--- title: "Lenny's Newsletter — 2025 年合集" date: "2025-01-01" source: "Lenny's Newsletter" url: "https://www.lennysnewsletter.com/" ---
# Lenny's Newsletter - 2025 (46 issues) This file contains 46 articles/episodes. --- ## [1/46] A guide to AI prototyping for product managers This post will transform how you build products, come up with new ideas, and operate as a PM. Whether you’re already deep into AI tools or just getting started, you’ll learn what tools you should be paying attention to, which tool to use when, and how to get unstuck when you run into an issue. You’ll find a collection of battle-tested prompts, real-world examples, and a step-by-step guide you can put into practice immediately. Imagine being able to turn Figma designs into a working app with a few clicks, or turn your PRD into a working prototype in minutes. This is all possible, and you’ll learn how. [Colin Matthews](https://www.linkedin.com/in/colinmatthews-pm/?originalSubdomain=ca) was a longtime PM and now teaches my favorite AI prototyping course: [AI Prototyping for Product Managers](https://maven.com/tech-for-product/ai-prototyping-for-product-managers?promoCode=LENNYSLIST). He also wrote my 9th most popular post of all time ([Become a more technical product manager](https://www.lennysnewsletter.com/p/become-a-more-technical-product-manager)). I absolutely love how simply and clearly Colin is able to explain complex topics, and I personally learned *so much* from this post on AI prototyping. I’m confident you will too. *If you’d like to learn more, [check out Colin’s free 30-minute lightning lesson](https://maven.com/p/30546f/ai-prototyping-for-product-managers-in-2025), where he shows you how to build a functional prototype in 10 minutes (including a few advanced techniques). And for much more, [check out Colin’s top-rated four-week cohort-based course](https://maven.com/tech-for-product/ai-prototyping-for-product-managers?promoCode=lennyslist), where you’ll master AI prototyping tools, learn to debug common issues, and build a fundamental understanding of how coding works. Use code “LENNYSLIST” to get $100 off.* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/5307ae42-888a-4ba5-8e9b-8cb6d3916c14_1456x1105.jpeg) If you haven’t been paying close attention over the past six months, you may have missed the rise of tools like [Cursor](https://www.cursor.com/), [Replit Agent](https://replit.com/), [v0](https://v0.dev/), [Bolt](https://bolt.new/), and other new cutting-edge AI tools that allow you to build working apps in minutes. For example, it took me 10 minutes to build [this 2-D tank game](https://cfjzdhoyqmiljtvb3fnjprf0stjcd8tg.vercel.app/) (with an AI opponent included), merely using this series of prompts: - *“Build a 2d tank game with an AI opponent.”* - *“Add collision for the shot when it hits a tank.”* - *“When health hits zero, play an animation and reset the game.”* - *“Improve the acceleration for player movement.”* - *“Make it so holding down the space bar has a timer to shoot a 2nd time.”* - *“Add power ups to the map.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/f2bb8af7-b477-4009-a792-6f997633bb67_3220x1922.gif) Pretty cool. But what’s cooler is that you can use these tools to build functional prototypes from a Figma design, convert a rough hand-drawn sketch to a working app, translate a PRD document into an interactive prototype, or even build a usable internal tool for your team, with no coding ability. In this post, I’ll cover the basics of AI prototyping, show how to get good results out of the most popular tools, and walk through an end-to-end example of building a prototype in less than 10 minutes. ## **Choosing your tooling** Current AI development tools come in three types: 1. **Chatbots (e.g. [Claude](https://claude.ai/), [ChatGPT](https://chatgpt.com/)):** The AI tools you probably know, which can also write and explain basic code 2. **Cloud development environments (e.g. [Replit](https://replit.com/), [Bolt](https://bolt.new/), [v0](https://v0.dev/), [Lovable](https://lovable.dev/)):** Full-stack platforms thatcan build and run your apps in the cloud 3. **Local developer assistants (e.g. [GitHub Copliot](https://github.com/features/copilot), [Cursor](https://www.cursor.com/), [Windsurf](https://codeium.com/windsurf), [Zed](https://zed.dev/ai)):** Development environments (i.e. IDEs) that help you write code with the help of AI ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/263e0c31-8ed5-40d0-b3d3-2fff6ef4f129_2912x1858.jpeg) Let’s review the most popular tools in each category to see what they can do and what we can build. ### Chatbots (ChatGPT, Claude) ***Best for:** Prototypes that are just one page and don’t have complex design requirements, like calculators, flip cards, or data visualizations* Chatbots are capable of writing code in response to a question or prompt. A prompt like ***“Build me a calculator with React”*** results in the following: ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/06c9d7ea-1cb6-41dd-8c88-5f8d9ff2ae4e_1080x641.gif) If you want to run this code, ChatGPT requires you to copy and paste the code into your IDE and run it on your own computer. Claude goes one step beyond ChatGPT’s abilities with their [Artifact system](https://support.anthropic.com/en/articles/9487310-what-are-artifacts-and-how-do-i-use-them). Artifacts allow you to run the code within Claude’s interface and deploy to a shareable link. An unfortunate limitation is that you can’t make any direct edits to the code, so you’re entirely reliant on using prompts to make code changes. ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/71c5a618-e663-49d4-986e-0ea9c18620f8_1080x641.gif) ([Perplexity](https://www.perplexity.ai/) is another tool you’ve probably heard of, which feels similar to chatbots but is more focused on search and not as useful for creating apps. It can write basic code because it’s built on top of other AI models like ChatGPT and Claude, but I wouldn’t recommend it for this use case.) Remember that chatbots can write code for any part of our stack (client, server, database) but can’t host your code (deploy) for us. They also can’t create complex prototypes with multiple pages, and it’s difficult to change the code directly. As a result, these tools are best used for very simple one-time prototypes—which sometimes is enough to get the job done. Think of chatbots when you’re looking to create a very simple landing page, individual inputs like a date picker, or small apps like a to-do list. ### Cloud development environments (Replit, Bolt, v0, Lovable) ***Best for:** Prototypes with more than one feature, specific design requirements, or many pages* Cloud development environments are one big step up from chatbots. These tools handle all the tasks required to turn your ideas into an actual working product. They can help you build end-to-end features, handle the backend infrastructure necessary to run your prototype, allow multi-file edits with agentic workflows, and take on more complex tasks across your codebase such as updating your database schema. One of the key differentiators among the various cloud development environments is hosting. As I explained in my [prior post](https://www.lennysnewsletter.com/p/become-a-more-technical-product-manager) on becoming a more technical PM, every software product is built with three parts: a client, a server, and a database. The client is what the user interacts with (often written in JavaScript), the server processes requests from the client to retrieve data or integrate with other products (often written in Node.js, Python, or Java), and the database is your permanent storage of data. Making prototypes that have real features requires you to host both your client and server code, and may require a database to power the app. One of the most popular tools today, [v0](https://v0.dev/), is capable of writing and hosting both client and server code. By default it uses specific frameworks called [Next.js](https://nextjs.org) and [Shadcn UI](https://ui.shadcn.com/) to do so (both were created by [Vercel](https://vercel.com/), the same company that owns v0). v0 can deploy your code and run backend servers—plus, one of its strongest features is that it has great styling as a default. Here’s a basic CRM I built in v0 with the prompt “Build me a basic CRM.” ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/baa2653d-0687-4214-8a50-df1281081b31_3226x1938.png) [Bolt](https://bolt.new/) is very similar to v0 in that it can also generate and deploy both client and server code. But a key difference is where the server runs. With v0, you deploy to real cloud hosting infrastructure, whereas Bolt runs the server code directly in the user’s browser. This means Bolt cannot natively support prototypes that need user identity like logins or accounts, multi-user interactions such as chat or collaborative workspaces, secure data operations like payment processing, or persistent data storage between sessions, because an isolated copy of the server is created on each user’s device. You can make up for this by integrating with external products like Supabase that offer servers and databases. Here’s a basic CRM I built with Bolt, using the prompt “Build me a basic CRM.” ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/fa08cc5e-283b-4da4-9d3a-441b93923b2b_3202x1938.png) Another popular tool is [Replit](https://replit.com). Replit allows you to build full-stack applications, including a client, server, and database. It can build web apps using both JavaScript and Python frameworks and particularly excels at building internal admin tools (e.g. file conversion, job applicant tracking) and data-driven applications (e.g. image resizing, multi-page dashboards) with simple UIs. I use Replit whenever I need a fully functional back end or I want to use Python code. I’ve used it to build an MP4-to-GIF converter and a Substack image resizer—both tools I use weekly. Here’s a basic CRM I built with Replit, using the prompt “Build me a basic CRM.” ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/40190b06-861d-4a75-b2eb-44fe6ee9800b_3220x1950.png) Finally, we have [Lovable](https://lovable.dev). It’s the newest of the bunch. Lovable is most similar to v0 and Bolt—it excels at generating websites, and uses JavaScript frameworks like React and Next.js. Its differentiation comes from its integrations with other popular tools. Lovable can connect to a GitHub repository, automatically add authentication and databases with Supabase, and help you connect to AI providers like Anthropic and OpenAI. All of these features make it one of the best AI coding tools for building products you actually want to use in production. One major drawback of Lovable is the lack of a code editor. To edit code, you have to ask the agent with prompting. This can make it difficult to debug issues directly in Lovable. I often find myself starting a new feature here but moving over to Cursor to resolve problems. Here’s a basic CRM I built with Lovable, using the prompt “Build me a basic CRM.” ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/5753e282-dacf-4349-9fd2-e6732810fe8f_3224x1938.png) To recap: - Choose v0 for beautiful designs by default - Choose Bolt for quick prototypes with flexible designs - Choose Replit for internal tools or products that store or transform data - Choose Lovable for building production apps that benefit from integrations with your current tools Regardless of your choice, cloud development environments all support building more complex applications than chatbots, with the ability to deploy to the cloud and easily share updated iterations over time. ### Local developer assistants (GitHub Copilot, Cursor, Windsurf, Zed) ***Best for:** People who know how to code and are working on serious applications they want to ship to production* The final type of AI development tool is local developer assistants. These products are targeted toward people who know how to write code. Tools like Cursor and GitHub Copilot can take prompts in a similar fashion to Claude but can then generate and apply changes within your own codebase and development environment (IDE). These tools do more than autocomplete—they can now write most of your code just using prompts. ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/eb9ff6f4-c9d2-4653-95c5-e63bd7e9e1f7_3246x2136.gif) For example, I built this presentation app (with live Q&A and polls!) in about 10 days using Lovable and Cursor. I started the app in Lovable to build basic features quickly, synced my code to GitHub to allow editing in other tools, and made final changes and fixed bugs with Cursor. This application uses authentication, databases, real-time updates, and more. Ten days may sound like a lot, but most of that time was spent resolving bugs and troubleshooting issues—something Cursor excels at compared with other tools, and something I get into a couple of sections down. ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/d66cb770-6f7d-4565-af71-02bd6d8da3f9_864x513.gif) GitHub Copilot is more popular in enterprise environments, as it comes from a trusted vendor, Microsoft. It supports multi-file changes from prompts, code explanations, and more. I’ve found it works best when given very specific direction and does not perform as well as Cursor at more general instructions. For example, when I asked for a new feature without providing context, Copilot re-created components of my app that already existed, whereas Cursor modified my existing files directly. Two tools I haven’t spent as much time with but people are excited about are Windsurf and Zed. Windsurf is another IDE that can suggest multi-line changes to files and suggest commands such as moving files on your behalf. It excels at working on larger, more complex codebases. Zed is a highly performant editor built with a variety of productivity features such as prompt libraries, slash commands, and keyboard shortcuts for common actions like applying AI-generated code. ## **Building your prototype** Now that we’re familiar with the basic tools, let’s build some prototypes. The two most common prototyping use cases for product managers are: 1. Converting an existing design to a functional prototype 2. Building an idea into a prototype from scratch ### Converting a design to a functional prototype Let’s turn the following design for Airbnb’s home page into a working prototype. Say you want to use this prototype to explore a new feature, such as a price filter. Here’s our initial design: ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/cf29900c-a0a2-4474-8620-e5a50a11a530_3080x1424.png) I chose Bolt for this task, as it’s better at building off a pre-existing design and we don’t need the backend database provided by Replit. Here’s a prompt I used (which you can copy and paste). Make sure to include a screenshot of the design! *“Build a prototype to match this design. Match it exactly.”* Here’s how this experience looks in Bolt: ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/000e35d0-08fd-471b-9f22-a12adae2b37f_1080x641.gif) Next, we’ll add our new price filter feature. Notice in the prompt below how I describe every feature in detail. One pro tip with prompting these tools is to be hyperspecific when describing changes for your subsequent prompts, as it helps the AI pinpoint what should change. *“Implement an inline price filter as a component of the search bar. It should appear next to ‘Add guests’ in its own section.* *Selecting the input should pop up a price filter with minimum and maximum values. The background of the pop-up should be white and should cover elements beneath it.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/6862d7f0-e5c6-4865-9716-69d171d5ceaa_3202x1920.gif) This is a great starting point. Let’s extend this feature with a slider for the minimum price. *“Can you add a price slider? It should have a blue line and a black node. Sliding the node should modify the minimum price.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/714f13e1-1534-4008-bf6f-1520264cdf4f_3218x1934.gif) We now have a functional initial prototype of your product idea within 10 minutes! Without any coding skill. Incredible. We could even continue to improve this prototype filter (e.g. showing listings updated in real time as you adjust the price). Check out the prototype [here](https://peaceful-shortbread-751d61.netlify.app/). ### Building a prototype from scratch If you’re like me, your design skills are probably not good enough to create the initial design we used in the prior example. Luckily, you can build a prototype using existing patterns and components from free and publicly available design systems like [Tailwind](https://tailwindcss.com/) or [Shadcn UI](https://ui.shadcn.com/). Let’s build a quick CRM with Bolt, then add a new feature to it. Let’s say we’re considering adding a feature that automates email outreach directly from our CRM and want to gather customer feedback about this feature before building it, by showing potential users a prototype of what it might look like. We can quickly put together a v1 with the following prompts: *“Create a comprehensive customer relationship management (CRM) system.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/c04b9082-1628-49c1-8814-5270966c8554_3198x1842.png) Again, let’s take a moment to pause here. We just created a working prototype of a CRM in less than five minutes—something that would have previously taken weeks of an engineer’s time. Unbelievable. Let’s keep going and see how we add a new feature. *“Please implement a mock AI email writer. This should be accessible from the left nav.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/c61c29eb-4947-43af-a4c6-70497f745483_1080x641.gif) Again, less than five minutes to be able to play with a new feature idea. Just think about how many ideas you can explore and how quickly you can bring them to market. AI prototyping tools have been available for less than six months and have already absolutely changed the speed at which teams can ship. Now we can take our prototype and get customers’ direct feedback without wasting time developing an initial version. This approach can speed up your discovery process by getting interactive examples in your customers’ hands as early as possible. Each AI tool will have very different outcomes based on their default settings and how specific you are. Here’s the exact same example with the same prompts using v0: ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/00ec3345-3c32-42dd-8891-d60a947973d0_2962x1626.png) ### Common use cases and prompt templates Here are a few good templates you can use to get started with each of the tools I’ve highlighted: #### **Task #1:** Build a prototype from an existing Figma design **Prompt:** `Build a prototype to match this design. Match it exactly. Use Tailwindcss.` `Match styles, fonts, spacing, and colors.` `[Include a single screenshot from Figma]` **Tool:** Bolt **Example:** [Deployment manager](https://spectacular-jelly-78231b.netlify.app/) #### **Task #2:** Build a prototype from scratch with good UI design defaults **Prompt** `Build a prototype for [x].` `This tool should: - [Behavior 1]` `- [Behavior 2]` `- [Behavior 3]` `Implement a simple initial iteration that meets these exact requirements.` **Tool:** v0 **Example:** [Sales calculator](https://v0.dev/chat/zMcDEi4HTuf?b=b_blmQipLxqTK) #### **Task #3:** Build a dashboard to visualize data **Prompt:** `Build a prototype for [x].` `Use Python and Streamlit.` **Tool:** Replit **Example:** [Product analytics dashboard](https://product-analytics-dashboard-colinmatthews2.replit.app/) #### **Task #4:** Convert a hand-drawn mockup to a prototype **Prompt:** `Convert the hand-drawn sketch to a functional prototype. Focus on frontend functionality.` `Make it in the style of [product you like].` ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/01579465-632f-4ffc-9656-25d4d2fd0a35_1600x1229.png) **Tool:** v0 **Example:** [Netflix-inspired blog](https://a4thjlf0qthccgos.vercel.app/) #### **Task #5:** Convert a PRD to prototype **Prompt:** `Implement a prototype to match the features in this PRD. Follow the exact specifications in the document. Focus on front end functionality -- do not include a server or database. Use Tailwindcss [Copy/paste PRD. Include any relevant images]` **Tool:** Bolt **Example:** [2-factor authentication UI](https://musical-pegasus-eb007e.netlify.app/) #### **Task #6:** Build your personalized productivity tool **Prompt:** `Build a tool that does [x]. This tool should: - [Behavior 1]` `- [Behavior 2]` `- [Behavior 3]` **Tool:** Replit **Example:** [Substack image resizer](https://substack-image-resizer-colinmatthews2.replit.app/) ## **Solving prototyping problems—without knowing how to code** Building your initial prototype is one thing. But one of the most common frustrations with AI prototyping is the errors the AI makes along the way toward a functional version. If you try to do anything even slightly complex, you’re likely to hit a wall. Knowing how to code becomes a superpower for debugging, but there are other approaches non-technical builders can use to still get great results. Let’s break down the four most effective ways to get to functional prototypes: 1. Reflection 2. Batching 3. Be specific 4. Lost context ### Reflection Reflection is the most important strategy for getting useful outputs from AI coding tools. These tools will respond to any question by writing code by default, which often means there’s no real plan. By forcing the AI to build a plan first vs. going straight to code, you’ll get much better results and have more visibility into what’s happening. Example: *“Build me a calorie tracking app with only a front end. Start by detailing out the minimum requirements. Do not write any code.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/5c6d213c-43c8-42b3-9292-a8ffe447778b_1080x641.gif) You can also use reflection to get out of a loop. Rather than asking the AI to fix your syntax errors, ask it to come up with a list of possible reasons an error exists. Be explicit about not wanting code, only an explanation of what’s happening. ### Batching Batching is a little unintuitive. Most people think that providing the most context up front will help the AI make the right decisions, but the opposite is true. You want to build the smallest iteration of something functional and then extend it. Try breaking down your prompts into smaller batches rather than asking for a complex prototype up front. I recommend starting with the data model first, as this is the backbone of how your prototype will store information. Example: *“Implement only the client-side view for calorie tracking. Use a basic data model that tracks entries with a description and number of associated calories. Display a table of all current entries and the sum of total calories in the top right corner.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/5a687bf0-350c-47a6-8431-bb140450f7e4_3218x1910.gif) ### Be specific The more specific you can be with AI, the more likely you are to get what you want. Just like if you were working with a junior engineer. Being specific comes in all forms—what technologies you’re using, what parts of the product should be built, and what files or even specific lines of code should change. Example: *“Add the ability to track calories on each day. - Extend the data model to include a date for the entry - Display a date picker in the entry form, defaulted to today’s date - Display today’s date inline with the total calorie amount - Add a left and right navigation arrow inline with the calorie amount to switch days backward and forward - The total calorie amount should show the sum of calories on the specified date.”* ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/89e8da38-877b-4808-9124-7528e659f871_3218x1910.gif) ### Lost context One of the most frustrating experiences of using AI prototyping tools is when they rewrite entire sections of your prototype and you lose hours of work. This often happens when your instructions are not specific enough and the AI cannot figure out the correct changes to make, so it rewrites everything. Fortunately, most tools have built a checkpoint system, where you can easily roll back to a prior version. ![Image from A guide to AI prototyping for product managers](https://substack-post-media.s3.amazonaws.com/public/images/c8855e71-6dc7-40f4-a77e-37ac3f9a2627_3218x1910.gif) You can also avoid losing context by focusing the AI on specific parts of the product that need to change. This can be done using a combination of the above strategies: 1. Reflection to determine what files need to change 2. Batching to limit the changes in each iteration 3. Being specific to minimize the chance of incorrect results ## **Putting it all together** AI prototyping is changing the way product teams work. Instead of spending weeks or months waiting for a feature to ship, you can build a prototype and get immediate feedback. Ship an internal tool in a day and see if it actually solves a problem. Sketch a mockup on a whiteboard and turn it into an app the same day. It sounds magical (and it definitely feels like magic), but there are limitations. Most product managers should use cloud development environments like: - v0 for beautiful designs by default - Bolt for quick prototypes with flexible designs - Replit for internal or data-driven tools - Lovable for building production apps and great integrations As you’re working on your prototype, you’re likely to run into issues. Be specific, reflect on the process, and try small incremental batches to get the best results. ### 📚 Further study 1. [Maximizing outputs with v0: From UI generation to code creation—Vercel](https://vercel.com/blog/maximizing-outputs-with-v0-from-ui-generation-to-code-creation) 2. [Prompt engineering overview—Anthropic](https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview) 3. [Build a fullstack app in 7 minutes with v0 (Figma to code)](https://www.youtube.com/watch?v=cyFVtaLy-bA) 4. [Bolt tutorial for beginners with the Bolt CEO Eric Simons](https://www.youtube.com/watch?v=1SfUMQ1yTY8&t=1607s) 5. [Windsurf vs. Cursor: which is the better AI code editor?](https://www.builder.io/blog/windsurf-vs-cursor) *Thanks, Colin!* *If you’d like to learn more, Colin is teaching a free 30-minute lightning lesson on January 14th where he’ll show you how to build a functional prototype in 10 minutes (including a few advanced techniques). [Sign up here](https://maven.com/p/30546f/ai-prototyping-for-product-managers-in-2025). For much more hands-on learning, [sign up for Colin’s live four-week cohort-based course on AI prototyping](https://maven.com/tech-for-product/ai-prototyping-for-product-managers?promoCode=lennyslist), where you’ll master AI prototyping tools, debug common issues, and build a fundamental understanding of how coding works. Use code “LENNYSLIST” to get $100 off. Enrollment closes January 23rd.* *Have a fulfilling and productive week 🙏* ## Hiring? 👀 I run a white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), working with a few select companies to fill their open roles. If you’re hiring, apply to work with us below. [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [2/46] Introducing Core 4: The best way to measure and improve your product velocity I talk a lot on my podcast and in this newsletter about the correlation between successful companies and an obsession with velocity. So I’ve been on the hunt for a framework that actually helps you measure and increase your velocity. I’ve finally found it. From the creators of DORA, SPACE, and DevEx, and in collaboration with [Laura Tacho](https://www.linkedin.com/in/lauratacho/) and the team at [DX](https://getdx.com/), I’m excited to introduce you to **Core 4**. Laura and her team spend every working hour researching, designing, and experimenting with ways to measure and improve team velocity (while avoiding burnout). Core 4 pulls everything they’ve learned from working with thousands of teams into a single unified developer productivity framework. Companies like Vercel, Intercom, and Dropbox have already put it in place and have seen great results. Below you’ll find everything you need to start using this framework within your team, including plug-and-play templates, industry benchmarks, case studies, tips for linking efficiency improvements to business impact, and specific advice for helping your team move faster. Gogogo! [Laura Tacho](https://www.linkedin.com/in/lauratacho/) *is the CTO at [DX](https://getdx.com/), has taught over 1,000 tech leaders through [her course on developer productivity metrics](https://lauratacho.com/developer-productivity-metrics-course), and on the side is an executive coach for engineering leaders. Her background is in developer tools and distributed systems. For more, check out her [LinkedIn](https://www.linkedin.com/in/lauratacho/) and her [blog](https://lauratacho.com/).* ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/f2e3ded9-41ac-4e1c-a51d-e82e00fd36ac_1456x1229.jpeg) If you’re reading this, you’re probably looking for ways to work more efficiently, “do more with less,” and move faster. I’ve yet to come across a software leader who isn’t. Unfortunately, if yours is like most companies, your product and engineering teams disagree about what “moving faster” means. Engineering leaders might zero in on development speed metrics like deployment frequency or application build times. Meanwhile, product leaders are focused on hitting deadlines and delivering value to customers. But “going faster” is only part of the equation. Product velocity is about speed *and* direction. When teams get too focused on “moving fast” but don’t have alignment on what to work on, they end up working on competing priorities and pulling in different directions. In those cases, trying to move faster can actually slow you down. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/a0926e59-22f8-4b8f-86f5-8bffe8fbab67_1798x720.png) Executing quickly means getting these arrows all pointing in the same direction. And executing quickly means getting to market faster and, as a result, learning what your customers need faster. “If we can learn faster than every other company, we’re going to win,” Mark Zuckerberg said in [a recent interview](https://www.acquired.fm/episodes/the-mark-zuckerberg-interview). That’s what product velocity enables. ### How to measure product velocity To improve, you first need to know how you’re doing today. But how do you decide what to measure? On the engineering side, developer productivity metrics have swelled in popularity, especially over the past several years, as companies try to figure out how to work more efficiently. In the past, most measurements focused on activity—like lines of code, number of commits, story points shipped—which don’t tell a complete story about performance. Some popular frameworks help connect the dots between engineering performance and business outcomes, all based on a lot of research. - **[DORA metrics](https://dora.dev/guides/dora-metrics-four-keys/)** are widely accepted “starter metrics” because of their popularity and focused scope. These metrics focus on software delivery capabilities (deployment frequency, lead time to change, change failure rate, and time to recover from a failed deployment), but they can often be misapplied. - **[The SPACE framework of developer productivity](https://queue.acm.org/detail.cfm?id=3454124)** offers a robust definition of developer productivity covering five dimensions: satisfaction and well-being, performance, activity, communication and collaboration, and efficiency and flow. SPACE is very thorough, but it doesn’t come with a turnkey list of things to measure (by design). - **[The DevEx framework](https://queue.acm.org/detail.cfm?id=3595878)** ties developer experience to three key dimensions: cognitive load, feedback loops, and flow. This framework is correlated with higher productivity but is somewhat disconnected from definitions of productivity used in other parts of the business. Still, even with these useful frameworks, many organizations feel lost. There are a number of concepts and metrics to consider, without a clear path forward. The most common question I get is: how can we decide between DORA, SPACE, or DevEx, or use them in combination? My answer used to be “It depends.” **Now I have a shorter answer: Core 4.** “Measuring developer productivity has been a notoriously tricky problem,” says Drew Houston, the co-founder and CEO of Dropbox. “Core 4 marries the state of the art in terms of research with a solution you can deploy in your company, and gives you a much more cohesive picture of what’s happening in your organization.” Core 4 is a set of metrics that unifies the principles behind DORA, SPACE, and DevEx. The framework was co-authored by [Abi Noda](https://www.linkedin.com/in/abinoda/), a co-author of the DevEx framework, and me, in collaboration with [Nicole Forsgren](https://nicolefv.com/) (founder of DORA), [Margaret-Anne Storey](https://www.margaretstorey.com/) (co-author of the SPACE framework), [Michaela Greiler](https://www.michaelagreiler.com/) (co-author of the DevEx framework), and other researchers. Our recommendations were informed by current research but also field experience at over 300 companies, including a few that were very early adopters of Core 4 (like Dropbox). “The SPACE framework was never meant to provide prescriptive metrics or tell you what exactly to measure. Core 4 gives you a list of things to start with that’s aligned to SPACE and DevEx, making it more practical for companies to use,” says Storey. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/bbec3635-771c-441f-9a27-6e8a8b8b4946_2912x1828.jpeg) The four dimensions of Core 4 are designed to hold each other in tension: we don’t want to increase speed at the expense of developer experience, or spend more time on new features while quality takes a nosedive. These metrics are designed to be used together as a system to provide a balanced look at overall team performance. This is not a recommendation of singular metrics to track, but rather a basket of metrics that must be used together to be meaningful. A few design choices are important to mention. First is the inclusion of PR throughput, measured by pull/merge requests per engineer, which can be controversial. We have seen that companies like Meta, Microsoft, and Uber have [successfully used this metric](https://newsletter.getdx.com/i/152463751/ive-talked-to-many-well-known-organizations-that-use-prs-per-engineer-as-a-central-part-of-how-theyre-thinking-about-and-measuring-productivity-whats-your-opinion-on-that-practice) as a key input for understanding and improving productivity. It can be misused easily, for example looking at PR output on an individual level, or as a single measure. It’s important that this metric is used only as a system health metric, and always alongside other metrics in the framework. “While this is an extremely hard metric to evaluate in isolation, it’s very easy to look at your organization’s diffs per engineer against peer companies’ diffs per engineer and make an informed judgment about whether things are going well for you,” says [Will Larson, the CTO of Carta](https://lethain.com/measuring-developer-experience-benchmarks-theory-of-improvement/). The Impact dimension, with its key metric of percentage of time spent on new capabilities, is unique in that more does not always mean better. Too much reactive maintenance will stifle a company’s ability to innovate. But too little investment in proactive maintenance results in a poor-quality product with problems kicked down the road. Another important design principle is that Core 4 balances qualitative and quantitative measurements. Quantitative measurements provide insight into what is happening, but qualitative insights tell us why. Ultimately, we need to understand the “why” of a team’s behavior in order to change it, which is the whole point of measuring in the first place. ### How to measure your Core 4 baseline today To get faster, it’s important to start with where you’re at today. With Core 4, you can collect a baseline measurement via self-reported metrics. Alternatively, if you already have access to clean data from your workflow tools, you can use that data along with survey data. Here’s a [template for a survey](https://docs.google.com/spreadsheets/d/1brKPLRJ9DDQAAFr1GM4hcFZg9zGUAGplQw2OkVx52Ls/edit?usp=sharing) you can send to your team to get a baseline on the key metrics in the Core 4 framework. You can use any surveying tool to do this—Google Forms, Microsoft Forms, Typeform, etc.—just make sure you can view the responses in a spreadsheet in order to calculate averages. Important: responses must be anonymous to preserve trust, and this survey is designed for people who write code as part of their job. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/345ead3a-b966-490d-a994-bfe5d7f6cb59_2912x7483.jpeg) Depending on your company’s size, you may want to collect certain demographic information, such as team identity and tenure, in order to analyze the results. There are some considerations to take into account, which I elaborate on in [this article](https://lauratacho.com/blog/a-deep-dive-into-developer-experience-surveys). ### Calculating your Core 4 metrics Once you’ve collected survey responses, it’s time to calculate your results. - For **Speed**, **Quality**, and **Impact**, find the average value for each question’s responses. - For **Effectiveness**, calculate the percent of favorable responses (also called a [Top 2 Box score](https://www.surveymonkey.com/mp/top-2-box-scores/)) across all Effectiveness responses. See the example in the template. To illustrate how to use these metrics, let’s look at a real organization’s results. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/37424000-51cc-416a-acb8-71f9cfaaa10f_2912x1310.jpeg) After looking into their data and using benchmarking data as a reference point (more on this later), they decided to focus on increasing Speed and Effectiveness to start with. Their hypothesis is that Impact will increase in the long term, after a short dip to focus on these internal improvements. To make this decision about what to work on to improve product velocity, they needed to drill down to the data on a team level, and also look at qualitative data from the engineers themselves. We encourage teams to triage their own results like this. From the qualitative results in the survey, many developers reported that release processes were hindering their velocity. Based on this, the company knew they needed to make investments in their continuous integration (CI) tooling and release workflows. Here’s where alignment with product comes in: how can teams weigh the relative benefits of making these investments now versus later, or compare their impact with the new features the business is asking for? ### Linking efficiency improvements to core business impact metrics Instead of presenting these CI and release improvement projects as “tech debt repayment” or “workflow improvements” without clear goals and outcomes, you can use Core 4 to directly link efficiency projects back to core business impact metrics. [Ongoing research](https://getdx.com/research/the-one-number-you-need-to-increase-roi-per-engineer/) continues to show a correlation between developer experience and efficiency, looking at data from 40,000 developers across 800 organizations. Improving the Effectiveness score (DXI) by one point translates to saving 13 minutes per week per developer, equivalent to 10 hours annually. With this org’s 150 engineers, improving the score by one point results in about 33 hours saved per week. This team also looked at supporting metrics to estimate the ROI of these projects: - PR cycle time (the time from opening a PR to merging it) was about four hours. The time to first review was about 60 minutes. Reducing these times by 25% would eliminate over 1,000 hours of waiting time per week (150 devs \* 5.4 PRs \* 1.25 hours). - Flaky tests are rated the #1 priority from developers, so a targeted focus here would improve developer experience. Targeting a three-point improvement to developer experience would mean recovering around 100 hours each week, equivalent to more than two full-time developers. There are two ways to frame these estimates. 1. Money: Weekly salary for two developers is $9,600 in this scenario, so each week, improving these aspects of developer experience is worth at least that much, or a half a million dollars annually. The 1,000 developer hours spent waiting to merge PRs come in around $120,000 each week. 2. Time: You now have capacity to bring revenue-generating features to market faster, investing those two full-time developers or 1,000 hours of saved waiting time into something else. Depending on the size and stage of your business, one of these frames may be more interesting to your teams and leadership. Are these calculations perfect? Absolutely not. Understanding the impact of these investments in terms of both time and money makes the stakes easier to understand when comparing these kinds of projects with new features, even if the comparison can never be apples to apples. The math makes it clearer to see that delaying this investment would have meant giving up on three full-time employees’ worth of time savings. That might make sense in certain situations. But this organization made the choice to invest immediately, even if it meant delaying some new feature work. The time savings alone meant a quick return on investment, and the project also supported the longer-term goal of improving the Impact score and spending more time on innovation and experimentation. ### Tracking velocity improvements with Core 4 Once you’ve got a baseline, you can start to regularly re-run this survey to track your progress. We recommend a quarterly cadence to begin with. For this organization, the investment is already starting to pay off as predicted. Speed, Effectiveness, and Quality improved by making investments in their CI and release workflows. Impact took a temporary dip, as they expected, as work shifted away from new features to work on these improvements. In their next quarterly measurement, they hope to see the Impact score surpass previous levels. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/d0063b6f-b87b-4834-a078-2fd2c4fae4b5_2912x1310.jpeg) They did hit their target of reducing merge wait times by 25%, with the actual reduction of PR cycle time coming in around 40%. That surpasses their target of saving 1,000 hours of wait time, taking into account their improved Speed metric of 6.6 PRs per engineer per week. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/fc2a41af-0095-42b5-a720-69a0c9aa9410_2912x1242.jpeg) Right now, there are hundreds of companies just like this one using Core 4 to decide what to invest in when it comes to productivity improvements. Increasing product velocity—or “moving faster”—is never done, but rather a horizon to chase. This habit of continuous improvement should repeat itself within your teams and across your organization and become part of the way that your company operates. With updated data, locate new areas of intervention, make new hypotheses, and repeat this process again. Intercom was an early adopter of Core 4. Iain Breen, who heads up developer experience at the company, says, “We’ve never had as comprehensive of a view before, which was a big win for us. We now have top-level metrics that help us measure and manage our work.” ### Using benchmarks Benchmarking data, both internal and external, will help contextualize your results. Remember, speed is only relative to your competition. Here are external benchmarks for the key metrics in the Core 4 framework. ![Image from Introducing Core 4: The best way to measure and improve your product velocity](https://substack-post-media.s3.amazonaws.com/public/images/bd6f681e-e959-4d58-968f-23fdd70fef71_2922x2070.jpeg) You can also [download full benchmarking data](https://getdx.com/research/benchmarks/), including segments on company size, sector, and even benchmarks for mobile engineers. We recommend looking at 75th percentile values for comparison initially. Being a top-quartile performer is a solid goal for any development team. ### The most common friction points that are slowing your team down Across hundreds of companies already using Core 4, here are the top points of friction in the developer experience, and some suggestions about what to do. #### **1. Poor build and test processes** Optimizing build and test processes helps reduce cognitive load on developers and accelerate feedback loops. Success signals: - Builds have predictable runtimes, allowing engineers to effectively manage their time around the pauses in their workflow. - Builds are stable and rarely require manual intervention or fail due to factors outside of the contributor’s code changes. What to do: - Implement caching for dependencies or other common build environment setup steps. This can immediately make builds faster by eliminating unnecessary setup time. - Parallelize steps when possible (i.e. frontend and backend tests running simultaneously). - Improve test efficiency by adding retries, removing dependency order, and breaking larger tests into smaller ones where possible. #### **2. Lack of time for deep work** Developers should have significant time for deep, focused work each day. Success signals: - Developers consistently have blocks of focus time that are over an hour long. - Developers have rapid feedback loops to avoid small but frequent disruptions. What to do: - Reassess recurring meetings and build a habit of regularly trimming meetings that accumulate over time. - Prioritize. Plan for engineers to focus on one project or priority at a time. [“Never half-ass two things. Whole-ass one thing.”](https://www.goodreads.com/quotes/10149245-never-half-ass-two-things-whole-ass-one-thing) - Limit the number of unplanned tasks or requests made to engineers. This might mean adjusting support rotations or creating workflows to avoid one-to-one pings. - Adopt no-meeting days or blocks to give developers control over their schedule. #### **3. Poor support for production debugging** Developers can easily investigate customer-facing issues and application performance. Success signals: - Developers can rapidly examine a variety of system events and metrics, in order to diagnose and address issues quickly. - Requests can be identified, isolated, and traced through systems effectively by engineers. What to do: - Adopt structured logging to keep log data organized, readable, and searchable. - Block off dedicated time for the team to add new log statements to existing code. This can provide an immediate boost to the team’s ability to trace and diagnose issues. - Develop a mechanism to track a user’s interactions in the UI. This assists in pinpointing potential bottlenecks or failures, making it easier to address problems. - Establish shared dashboards for system health metrics. This improves transparency and makes sure efforts to improve the dashboards benefit everyone. Many teams struggle with documentation, which impacts the developer experience globally. DORA’s research has [highlighted the importance of documentation](https://cloud.google.com/blog/products/devops-sre/deep-dive-into-2022-state-of-devops-report-on-documentation) for the past several years. Even small changes like standardizing documentation templates, deleting stale documentation, or using AI to transcribe and summarize recent product demo meetings can make a big difference in the discoverability of information. ### How to start using Core 4 to improve product velocity Going faster is about more than just speed. There should always be a healthy tension between feature projects and efficiency projects, but this tension can become unhealthy when competing priorities create misalignment. When this happens, it’s even possible for individual projects to move faster, but that effort does not lead to the outcomes expected by the business. Product and engineering want the same thing—to get more value to customers, faster—and should be moving in the same direction. But teams might not have the tools or common language to talk about tradeoffs and get on the same page. As Vercel CEO and founder Guillermo Rauch explains, “Engineering velocity is everything. Early-stage startups live or die based on product-market fit. With scale-ups, you live or die by velocity.” Using a framework like Core 4 to measure team velocity can help product and engineering teams answer the question “What should we work on to get faster?” together. If you want to use Core 4 to have data-informed conversations about what to prioritize, you should: - Share this summary [slide deck about Core 4](https://docs.google.com/presentation/d/15RQzf8UtXjhdq9AA6dxRUk9940vQzfta7dfHs9cy9mM/edit?usp=sharing), why it was designed, and how it can be used with the rest of your leadership team. - Get a baseline measurement. Use existing data if you can; otherwise use the templates in this article. - Compare your results with benchmarks in order to set high standards and see where you have room for improvement. - Dig into the data to see opportunities for improvement. Remember that your teams use these systems every day and know what needs to change. Involve them in the decision-making processes, and encourage teams to triage their own data. Finally, I will end with a question. In your next leadership meeting, ask this: Are your teams moving as fast as they could be, and moving in the same direction? *Thanks, Laura! For more, you can follow her on [LinkedIn](https://www.linkedin.com/in/lauratacho/) and her [blog](https://lauratacho.com).* *Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week [Delphi](https://www.delphi.ai/) is hiring for a [Lead Product Manager](https://delphi-ai.notion.site/Delphi-Product-Manager-13fa9fbe9ec880218413f1b4f73322c4) based out of SF. As their Lead PM, you'd get the opportunity to collaborate directly with their CEO to build and own an entire product category. **Why I think the company is interesting:** 1. Delphi powers [Lennybot](https://www.lennybot.com/)! They allow anyone to upload their content to create a digital representation of not only their knowledge base, but also their way of thinking. 2. With zero marketing spend they’ve managed to go from 0 to nearly $2 million ARR in their first year. 3. They are backed by Sequoia, Founders Fund, and a host of tier-one investors. 4. They a are still a small scrappy pre-series-A team. This is an opportunity to *actually* join at the ground floor and make a huge impact. If you’d like to get your profile sent directly to their team, just fill out this [quick form](https://airtable.com/apppne9fUg8CKw6s2/pag2vapuCgpIGh6WB/form). All submissions from my readers get priority (but no guarantees beyond that). — If you’re hiring, I run a white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), working with a few select companies to fill their open roles. Apply to work with us below. [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [3/46] What’s in your stack: The state of tech tools in 2025 *👋 Welcome to this month’s ✨ **free edition** ✨ of Lenny’s Newsletter. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. If you’re not a subscriber, here’s what you missed this month:* 1. *[A guide to AI prototyping for product managers](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product)* 2. *[Introducing Core 4: The best way to measure and improve your product velocity](https://www.lennysnewsletter.com/p/introducing-core-4-the-best-way-to)* 3. *[Top angel investors in the U.S.](https://www.lennysnewsletter.com/p/top-angel-investors-in-the-us)* *For more: **[Lennybot](https://www.lennybot.com/) | [Podcast](https://www.lennysnewsletter.com/podcast) | [Swag](https://lennyswag.com/) | [Hire your next product leader](https://www.lennysjobs.com/)** | **[My favorite courses](https://maven.com/lenny)*** I’ve always been fascinated by what tools people choose to use in their work. What started as a casual survey of PMs on Twitter has grown into something much bigger. Today, with insights from over 6,500 of you (thank you! 🙏), I’m excited to share the results of my first-ever large-scale “What’s in your stack?” survey. The results are both surprising and telling. We’re seeing AI already transforming people’s jobs, watching the methodical disruption of incumbents by beautifully crafted alternatives, and witnessing a few seismic shifts in how teams collaborate. To help me design and run this survey, I pulled in my former colleague [Noam Segal](https://www.linkedin.com/in/noamsegal/), who’s been a UXR leader at Meta, X, Intercom, Airbnb, Wealthfront, Upwork, and, most recently, Zapier. I’m delighted to collaborate with him on this and, hopefully, many more surveys. Let’s get into it. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/331167af-934d-40f4-8f05-c7b4d408bad1_1456x1215.jpeg) ## The evolution of tech tools: A five-year journey In 2020, Lenny conducted his [first (informal) survey](https://www.lennysnewsletter.com/p/product-stacks) of technology tool preferences, gathering responses from hundreds of people on Twitter and LinkedIn. Surprises included Slack coming in first, Notion beating Google Docs, and Linear becoming a fast-growing up-and-comer. When Lenny [revisited the PM tool stack in 2022](https://x.com/lennysan/status/1507040234000760836), he noted the continuing dominance of tools like Slack and Notion and the rise of Figma, Zoom, and Loom, signaling a shift toward more collaborative and async workflows. Today, with Lenny’s Newsletter reaching nearly 1,000,000 subscribers, we present our most comprehensive analysis anywhere of the tools shaping modern tech workplaces. ## What we asked and who responded The survey covered 13 categories, from AI assistants to project management to CRM. Beyond just asking, “What do you use?” we asked people what tools they most love, what tools frustrate them, and what they’d change if they could. Of the respondents, 50% work in product, 11% are engineers, 10% are founders, and the rest work in other cross-functional roles, including marketing, design, and growth. Based on prior research, the breakdown of company size among Lenny’s community is as follows: - ~45% work in 1-to-100-employee companies - ~25% work in 101-to-1,000-employee companies - ~20% work in 1,001-to-5,000-employee companies If we define sub-1,000-employee companies as “not enterprise,” about 70% of Lenny’s community work in startups or midsize companies. You could say this is the early-adopter crowd. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/16475c40-1929-4cd2-9871-315ece0b7c8c_1588x1450.jpeg) ## The big 10 headlines 🗞️ Before we get into the details, here are 10 key themes that emerged: 1. **ChatGPT has a commanding lead.** 90% of respondents use ChatGPT regularly. Only 35% use Claude, and 24% use Gemini. 2. **Cursor and other AI-native integrated development environment (IDE) tools are rapidly emerging.** 17% of respondents already use Cursor regularly (launched just two years ago!). 10% of all participants use v0 and Replit. 5% use Bolt. 3. **As the third-most-used tool overall, Slack continues to crush it.** 72% of participants use Slack regularly. It’s only behind ChatGTP and Gmail 🤯 4. **The Jira paradox and Linear’s insurgency.** 68% of participants use Jira, *but* it also tops the “we wish we could use a different tool” list. Enter Linear: the fastest-growing alternative to Jira, and already used by over 10% of participants. 5. **Figma Slides and Canva have become big players in presentations.** They’re already far ahead of Apple Keynote and closing in on PowerPoint. We’re also seeing new AI-native entrants in this space. 6. **Google Docs remains a go-to for collaboration, but Notion is gaining steam.** Notion is seen as “good for everything,” and it’s catching up to the big players, with 37% of respondents preferring it. Notion also came in second place for project management after Jira and fourth place for CRM. 7. **Figma continues to be the ubiquitous tool for design.** 97% of designers report using it as their primary design tool. Canva is still well behind but catching up, thanks to massive popularity with marketers and founders for general design needs. 8. **Miro continues to stay ahead of FigJam for virtual whiteboarding, just barely.** But FigJam is gaining ground (because everyone’s using Figma…). 9. **Notion and Slack are the CRM and customer support surprises.** Turns out that even with powerful incumbents, tool flexibility counts for a lot. 10. **Three meta-takeaways:** bundling, craft, and mix-and-match. More on this below. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/633ec6d6-395f-4261-8d62-53183f1350a5_1457x1516.jpeg) Now, let’s break down what’s happening in each space. ## ChatGPT has a commanding lead The most striking shift [since 2022](https://x.com/lennysan/status/1507040234000760836)? AI tools have become as essential as having a laptop. A whopping 90% of respondents use ChatGPT regularly. This is the most significant shift in the product team tool stack in recent memory. More participants use ChatGPT than Gmail (76%) or Slack (71%) 🤯 Interestingly, over 50% of participants combine AI assistants for specific use cases: - ChatGPT + Claude as a thought partner - ChatGPT + Perplexity for deep research - ChatGPT + Gemini for Google Workspace integration Role-specific AI tools are also gaining serious traction: - 40% of engineers use GitHub Copilot regularly - 21% of engineers have adopted Cursor - Though they weren’t included in the response options, tools like ChatPRD and Grammarly are becoming popular and were mentioned by 5% to 10% of people who chose to add their AI assistant tools under “other” ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/abf52b0d-cd64-4190-92ae-4cc27720eca1_1588x820.jpeg) ## Cursor and other AI-native IDEs are rapidly emerging The second most significant shift in the tool stack is the emergence of AI-native development environments, with tools like Cursor already being used by 17% of overall participants (and 21% of engineers in our sample), despite just being launched in 2023. Nearly 10% of respondents are already using tools like v0 and Replit, and 5% are using Bolt. Who’s using them? It’s a mix of ~60% product and ~40% other roles, including engineers, founders, consultants, marketers, designers, and UX researchers. Most of these tools were launched just over a year ago. This rapid adoption suggests developers are hungry for tools that make coding easier and that integrate into their coding workflow. It’ll be interesting to see how these AI-native coding tools fare in the next round of this survey. Almost two-thirds of participants use GitHub; however, the most popular “other” response was GitLab, GitHub’s primary competitor. Whoops. We’ll include GitLab next time to get a better sense of its use too. VS Code has established a strong position among engineers, with a 48% adoption rate. Its success reflects its technical capabilities and Microsoft’s successful platform evolution into a highly extensible, community-driven tool. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/bc53a82c-08e3-45cb-92f6-ef77f0e61f23_1588x1450.jpeg) ## As the third-most-used tool overall, Slack continues to crush it The numbers tell the story: - In Lenny’s [previous survey](https://x.com/lennysan/status/1507040234000760836), Slack emerged as #1 overall. This trend continued here, with an enormous 72% of respondents using Slack as their primary communication hub (just behind Gmail). - Microsoft Teams technically holds about 33% market share (according to [financial statements](https://www.notta.ai/en/blog/microsoft-teams-statistics)), but I believe this is due to their enterprise bundling strategy and their actual usage isn’t what people think. And as you’ll see below in the section on switching tools, users aren’t happy about using Teams. Also 👇🏻 ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/caea52eb-a05b-49e8-beba-8d0c5f82f32a_1320x534.jpeg) 🤿 Deeper dive: Why is Slack winning? Two hypotheses: 1. People do not think highly of the Teams user experience. We heard it’s cumbersome, slow, and even “impossible to use.” Slack is winning when it comes to user experience. 2. According to publicly available data, Microsoft Teams is mostly used by non-tech companies and has been adopted by most large U.S.-based enterprise companies. Conversely, Slack is predominantly used by startups up to midsize companies. Most respondents to this survey work in companies where Slack dominates. 🔍 Another interesting finding: WhatsApp is used by a whopping 20% of respondents in their work, and Telegram is making inroads in work communication, with 15% of respondents who chose “other” using it daily. WhatsApp and Telegram were never designed to be work communication tools, and yet a non-trivial number of people are using them as such 🤷‍♂️. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/2e9c170b-7e80-4231-a971-e584a344c4cd_1588x960.jpeg) ## The Jira paradox and Linear’s insurgency Here’s an intriguing contradiction: Jira dominates the project management market (53% of technical teams use it, biggest share by far) but simultaneously tops the “please let us switch” list. Its deep integration with development workflows, bundling with other products, and enterprise-first feature set keep teams locked in. Respondents’ feedback shows that Jira is overly complex, making it hard to learn and use. One participant said, “Jira is a mess, bad performance, and hard to maintain. It’s convoluted, full of feature creep and bad UX decisions.” Another said, “Jira is just too complicated/bloated for our needs. I find it cumbersome and could be much more simplified for what it does.” Enter Linear: the fastest-growing alternative to Jira, founded in 2019 and already used by over 10% of participants (vs. receiving only 10 mentions [in the last survey](https://x.com/lennysan/status/1507040234000760836)). Respondents praised its modern, intuitive interface and streamlined workflow management. One engineering manager said, “Just moved off the Atlassian suite to Linear, and we *love* it. *Much* more useful in terms of milestones and flexibility. Easier to filter views and build a personal workspace setup.” Linear’s popularity is on par with that of Asana, a company founded in 2008. Meanwhile, Notion is playing a clever game. It’s become the second-most-popular project management tool and the fourth-most-popular docs, and it’s gaining traction as a CRM tool. Participants consistently praised its flexibility (“It’s my go-to for … anything”) and how it enables teams to build a shared understanding. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/2220b1f6-dd5f-4438-b78f-51674b419de8_1588x1380.jpeg) ## Figma Slides and Canva have become big players in presentations The way people craft presentations is evolving in fascinating ways. Three distinct approaches have emerged, each telling a different story about how we communicate visually in 2025. First, there’s the old guard. Google Slides (#1) and PowerPoint (#2) continue to dominate traditional presentation creation, doing what they’ve always done well. However, design tools are staging a quiet revolution in the presentation space. Figma and Canva have become serious contenders, running neck-and-neck in adoption rates. This might surprise folks who know these platforms primarily as design tools. But the numbers don’t lie. What’s driving this shift? Participants consistently report that these tools offer something traditional presentation software can’t: creative freedom without complexity. AI is also entering the presentation game. Newcomers like Pitch, Gamma, and Beautiful.ai are fundamentally rethinking how presentations come together. Instead of starting with blank slides, these tools use AI to shape your content into polished presentations. While they’re still finding their footing, early adoption signals hint at a major shift in how we’ll build decks in the future. Another trend emerged from the data: Miro kept showing up in our “other” category, suggesting people aren’t just switching presentation tools—they’re questioning whether they need traditional slides at all. The line between presentations, whiteboards, and collaborative spaces is blurring, and this might be just the beginning. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/c2a305ed-f562-4bf5-81d1-9a0ff812e3d2_1588x1100.jpeg) ## Google Docs remains a go-to for collaboration, but Notion is gaining steam The docs space has consolidated around three platforms, each serving distinct needs: 1. Google Docs remains the go-to for real-time collaboration. 2. But Notion is an up-and-comer, capturing strongholds in team wikis, project management, and documentation. The most common sentiment about it is “It’s good for *everything*!” 3. Confluence ... well, it’s hanging on in enterprise teams (though not for lack of complaints). Google Sheets keeps gaining ground in the spreadsheet landscape, but Excel shows surprising resilience. Meanwhile, tools like Evernote, Quip, Coda, and Dropbox Paper are fading into the background as companies seek to consolidate their tool stack. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/efbecb0b-e6a9-4dd0-b4ee-a7fbb100f687_1588x1240.jpeg) ## Figma continues to be the ubiquitous tool for design and UX If you’re in design, you’re almost certainly using Figma. It’s that simple—90% of overall participants and 97%(!) of designers report using it as their primary tool. The surprise player is Canva. While it’s not competing directly with Figma for professional UX work, it’s democratizing design for everyone else. Product managers, marketers, and engineers are using it to create quick visuals without bothering their design teams. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/5f80f65a-66fd-449e-8bb4-93ae2ea6c628_1588x820.jpeg) ## Miro continues to stay ahead of FigJam in the virtual whiteboarding space, just barely In the battle to be the place where tech professionals think together, Miro remains the leader, but barely. As you just read, Figma is where design work happens, so FigJam has an advantage that could force Miro out of its leadership spot in future years. Perhaps having observed Figma’s approach, Atlassian and Microsoft developed virtual whiteboard products, which, funny enough, are both called “Whiteboard.” About 5% of the responses mentioned these products. Conversely, Mural and Whimsical don't seem to be growing. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/4726c5e5-ea40-43a3-96cf-a5a6d21e51b6_1588x820.jpeg) ## Notion and Slack are the CRM and customer support surprises Two companies dominate the CRM landscape, but there’s a surprise entrant to this market. Salesforce’s comprehensive enterprise features make it the default choice for large organizations, but smaller companies increasingly express frustration with its complexity and cost structure. This has created opportunities for alternatives that emphasize simplicity and specific use cases. HubSpot has stepped into this gap, particularly among small-to-midsize businesses. Participants said it’s more intuitive and streamlined than Salesforce, and accessible to teams without dedicated CRM specialists. But then comes a surprise. Coming in third, 11% of our participants selected Notion as their CRM of choice. According to the results, *flexibility* is the key to its success: “Notion offers a ton of flexibility that our team got, and perhaps would be the most difficult to replace.” ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/7448db9a-3201-4501-a6c5-7184a1771f4f_1588x1240.jpeg) The customer support tool landscape shows similar patterns of established leaders, as Zendesk maintains its position as the primary support platform, with 29%. But what’s fascinating is that Slack is also at 29%. Why? We hypothesize that participants in this survey skew toward earlier-stage companies. Slack is a must, and Zendesk or Intercom are expensive and complicated, especially if you don’t have a customer support (or success) team. Slack lets you create an external shared channel for each customer or design partner, where co-founders or early employees can engage with their customers directly. This insight reveals another variable behind Salesforce’s decision to acquire Slack for nearly $30 billion several years ago. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/b445cf0c-4ee0-4dac-ab77-6887a097d517_1588x1100.jpeg) ## Three meta takeaways: bundling, craft, and mix-and-match ### 1. Bundling is powerful, but it can get you only so far. Some of the most-used tools, like Jira, Microsoft Teams, and Google Slides, are all bundled within their respective corporate stacks, which locks people in for the long term and creates a massive switching cost. Thanks to the bundle, they end up “winning,” but we’re seeing some of these products top the “least valued” and “most interested in switching from” lists, so it may be only a matter of time until a better-crafted and well-executing startup (e.g. Linear, Figma Slides) [finds a wedge in](https://www.lennysnewsletter.com/p/wedge) and eats their lunch. ### 2. Well-crafted products are disrupting incumbents. Linear, Notion, Figma Slides, and Slack are all praised for their user experience, fit to people’s workflows, and focus on the perfect set of features. They’re rising fast (or already winning) and topping respondents’ wish lists to switch to them. ### 3. People are mixing and matching different tools in the same space. When we looked at the tool stack landscape more holistically, we saw that for core jobs, people use several (competing) tools within the same category, depending on their needs. For example, most respondents use more than one AI assistant, based on each assistant’s strengths. ## The tools people value most, and the ones they’d ditch if they could We asked participants to choose up to three tools they value most and least, expecting little overlap between the two lists. **We were wrong.** Tools like Slack, Jira, and even ChatGPT were mentioned often in both lists. We realized there was a need for a better metric to capture tool value while accounting for two critical things: 1. How often tools were selected *in general*. For example, Slack topped the original “most valued” list, but it’s also used by a significantly higher number of people than Linear. 2. How often tools made the *least* valued list in addition to the most valued one. Therefore, we adjusted the ranking based on the ratio of general popularity to most valued *and* the ratio of most valued to least valued, penalizing tools ranked highly on both lists. We ended up with the “adjusted most valued” (AMV) metric, and the top 5 tools tell a compelling story. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/41f1a808-8587-4451-a20e-f7757554647b_1457x1397.jpeg) #### Linear: Craft is winning Participants love Linear’s user experience, its match to their workflows, and its relative simplicity and focus compared with Jira. #### Cursor: The future of software development is AI-native Cursor is rising at an astonishing pace, and there are at least five other AI-native coding platforms in the mix. Over 20% of engineers are already relying on Cursor, and it was chosen above most other legacy coding tools like JetBrains, IntelliJ, and even Xcode. #### Slack: Where work happens Look, it ain’t all love when it comes to Slack. Slack topped our “most valued” list, but it also topped our “*least* valued” list. People heavily rely on it for communication, but some also see it as a productivity killer. Some respondents described it as a “detriment to focus,” “adding tons of cognitive load,” and “a noise generator.” Still, at a time when people are telling us that communication is becoming increasingly speedy, Slack is where work happens. #### Notion: Good enough at everything There’s a new guard of modern tools.Notion stands out as representing a new breed of tools prioritizing collaboration, intuitive design, and the flexibility to accomplish a range of jobs, from docs to project management to collaboration. #### Perplexity: Answers instead of links Perplexity making the top 5, in addition to ChatGPT’s and Claude’s popularity, tells us something important: AI tools aren’t just shiny toys anymore—they're transforming people’s workflows, replacing well-established tools (e.g. Google), and becoming essential to the workday. ## **The “please let us switch” list 😤** We also asked participants what tools they’d ideally love to switch out, and which tools they’d love to get their hands on instead. The biggest winners here? Linear, Slack, and Notion, a lineup people seem to consider the modern stack for project management, communication, and collaboration. Atlassian’s Jira and Confluence did not fare well, nor did Microsoft Teams. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/a9cee6fd-e377-44e6-b3b0-e165fe8689e1_1456x2021.jpeg) ## Other significant insights: user research, analytics, email ### 1. User research: glow-ups, who you research matters, and the rise of specialist tools #### Survey tools get a glow-up Google Forms dominates the landscape, especially among PMs, potentially because it’s simple, has sufficient functionality, and is bundled with Google Workspace. However, Typeform is the perfect example of what happens when someone reimagines a tired format. With a richer feature set, a less traditional approach, and a better-crafted design, it reached third place in user research tools. #### User Interviews is a recruitment game changer User Interviews has climbed to the #2 spot by solving one of the biggest headaches in research: finding the right people to talk to. Think about it—what good is the perfect research plan if you can’t get the right participants? Respondents love it because it turns the messy process of finding, scheduling, and paying participants into something manageable. #### The specialist squad But here’s where the landscape gets really interesting. A whole crew of specialized tools is changing how teams understand their users: - UserTesting still owns the usability space - Qualtrics handles enterprise-grade customer experience research - Dovetail has a stronghold on insight management - Maze makes research easy and speedy - Sprig is an always-on product experience platform - Optimal Workshop tackles information architecture - Dscout owns diary and longitudinal research ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/295ccc7a-3e78-463d-bc7e-9fa505c36163_1588x1520.jpeg) ### 2. Analytics: power players vs. specialists The data analytics landscape in 2025 tells a story of David vs. Goliath—or, more accurately, a few Goliaths vs. an army of Davids. Google Analytics remains the undisputed heavyweight champion, dominating general analytics usage. But here’s where it gets interesting: dig a little deeper, and you’ll find a thriving ecosystem of specialized tools carving out their territories. #### The power players When teams need to level up their business intelligence game, they’re increasingly turning to two leading players: 1. **Tableau:** The storyteller’s choice, a platform that simplifies turning complex data into compelling dashboards. “It’s effortless to slice and dice data. Tableau is our internal source of truth.” 2. **Looker:** The data scientist’s darling, Looker is helping democratize data across companies. “​​It’s the best way to make data accessible and usable across domains/functions without too much effort.” #### The behavior tracking faves Amplitude and Mixpanel remain the two biggest analytics players after Google Analytics, helping teams track everything from user behavior to feature adoption. One PM told us, “Amplitude empowers the product teams to learn instantly, finding answers on the spot.” #### The specialists are carving out their niches 🎯 We’re also witnessing the rise of highly specialized tools: - Hotjar is dominating the qualitative space with its heatmaps and session recordings - Metabase has become a startup favorite for quick, no-nonsense dashboards - Pendo owns a good chunk of the product feedback and onboarding niche - Segment is the go-to for data pipeline management - Fullstory and Heap are making names for themselves in behavioral data collection ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/e4d7152a-3b96-40ff-a615-476e2cde51ec_1588x1240.jpeg) 30% of the responses were in the “other” category. The two leading platforms, with around 200 responses each, were neck and neck: - **Posthog:** An open-source, all-in-one platform competing with the big players—Amplitude, Mixpanel, Fullstory, and Heap. Founded as a Y Combinator company almost five years ago, respondents seem to love them, and they’re [used by most YC companies](https://x.com/daltonc/status/1865108608805146804). They’re an intriguing up-and-comer that we’ll look at more closely next survey. - **Power BI:** Microsoft’s answer to Salesforce’s Tableau and Google’s Looker. Data analysis, visualization, dashboards... Often the default choice at companies using the Microsoft stack. ### 3. Email: The three camps The email universe basically falls into three buckets: 1. Gmail 2. Microsoft Outlook 3. Specialized tools for superusers Gmail is second in overall tool popularity rankings, behind ChatGPT but ahead of Slack (the previous survey’s winner) and every other email tool. Also, as an aside, Google’s Workspace was valued significantly higher than Microsoft’s Suite throughout the survey. Microsoft is maintaining its enterprise presence, but the satisfaction gap is widening. Their tools made several appearances in the “least valued” rankings, and, as one respondent put it with regard to Outlook specifically, “Everything about Outlook feels corporate and soulless compared to Google Suite, but enterprise inertia keeps us locked in.” Power users gravitate toward specialized tools like Superhuman and Front, which remain relatively niche, but their users report significant productivity gains that justify higher costs (“Superhuman is a game changer”). ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/a60c187b-fb9a-464f-8ab6-adf49df78bd5_1588x820.jpeg) ## The big picture ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/642c30dd-1ac7-4cc4-9b8e-b5e61c883fae_2913x4548.jpeg) When we zoom out of all this data, there are a few big ideas that become very clear. First, AI isn’t just the new normal; it’s ubiquitous. Teams aren’t just leveraging AI—they’re building entire workflows around it. ChatGPT isn’t just winning; it’s dominating. Respondents shared that ChatGPT packs a double punch: it expands their thinking when ideating or brainstorming and streamlines critical workflows like data analysis or writing. Another theme that’s coming through is that when it comes to picking tools, user experience trumps features. Teams are increasingly willing to sacrifice deep functionality for tools that are actually pleasant to use. The days of “but it has more features” are over. Tools like Linear, Notion, Slack, and Figma are appreciated for the craft that went into them and the flexibility they offer. Which is why we may see tool migrations accelerate. We’re noticing strong negative sentiment toward tools that aren’t up to par with the modern stack, and a strong willingness to switch from legacy tools to those modern alternatives. Bundling can get you only so far before well-built tools eat your lunch. ![Image from What’s in your stack: The state of tech tools in 2025](https://substack-post-media.s3.amazonaws.com/public/images/ef83f269-7795-4562-8c4b-e6bf6314f720_1174x288.png) And we’re noticing that savvy teams and individuals are mixing and matching tools within the same space, choosing the perfect one for each nuanced context or need. For example, people are catching on to which AI assistant can help most for specific use cases and holding strong opinions on which presentation tool is best fit for the type of presentation they need. One thing’s for sure: people in tech will always have one eye out for better tools. Shiny object syndrome is impossible to avoid. 😉 Special thanks to the more than 6,500 tech professionals who shared their insights for this survey. Your candid feedback will help everyone in Lenny’s community build better tool stacks 🙏 *Thank you, Noam! You can find Noam on [LinkedIn](https://www.linkedin.com/in/noamsegal/) and [X](https://x.com/noamseg).* *Also, thank you to [Ben](https://www.linkedin.com/in/benhawkins0/) for design and [Rebecca](https://www.rebeccaackermann.com/) for editing.* *Have a fulfilling and productive week 🙏* *Disclaimer: I may be an investor in some of the abovementioned companies.* ## 🏅 Featured role of the week [Delphi](https://www.delphi.ai/) is hiring for a [Lead Product Manager](https://delphi-ai.notion.site/Delphi-Product-Manager-13fa9fbe9ec880218413f1b4f73322c4) based out of San Francisco. As their Lead PM, you’d get the opportunity to collaborate directly with their CEO to build and own an entire product category. **Why I think the company is interesting:** 1. Delphi powers [Lennybot](https://www.lennybot.com/)! They allow anyone to upload their content, to create a digital representation of not only their knowledge base but also their way of thinking. 2. With zero marketing spend, they’ve managed to go from 0 to nearly $2 million ARR in their first year. 3. They are backed by Sequoia, Founders Fund, and a host of tier-one investors. 4. They are still a small, scrappy, pre-series-A team. This is an opportunity to *actually* join at the ground floor and make a huge impact. If you’d like to get your profile sent directly to their team, just fill out this [quick form](https://airtable.com/apppne9fUg8CKw6s2/pag2vapuCgpIGh6WB/form). All submissions from my readers get priority (but no guarantees beyond that). — If you’re hiring, I run a white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), working with a few select companies to fill their open roles. Apply to work with us below. [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [4/46] Introducing the Foundation Sprint: From the creators of the Design Sprint Everyone working in product today knows about the Design Sprint—a five-day method for teams to design, prototype, and test new products. The process goes back to 2010, when [Jake Knapp](https://www.linkedin.com/in/jake-knapp/) came up with the steps while helping build Gmail and Google Meet. His *New York Times* best-selling book, *Sprint,* has sold over 500,000 copies, been translated into more than 20 languages, and is a staple of product team bookshelves worldwide. I’m honored to share an exclusive excerpt from his new book, ***Click***, the anticipated sequel to *Sprint* that will hit bookshelves in April***.*** But this is not just an excerpt. Jake and his co-author [John Zeratsky](https://www.linkedin.com/in/johnzeratsky/) are sharing their new sprint method, the **Foundation Sprint**, for the first time ever, in this newsletter**.** Jake and John created the Foundation Sprint based on lessons from working with over 300 companies in the past 20 years. You’ll also learn about a framework I now encourage every founder to use—and that Jake used himself to create Google Meet: **the Founding Hypothesis**. If you [pre-order](https://www.theclickbook.com/) the book before January 31st, you’ll get instant access to a bunch of resources, including detailed guides and tools for running your own Foundation Sprint. Pre-order and claim these bonuses at [TheClickBook.com](https://www.theclickbook.com/). ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/fe39ea34-d1a7-4df0-b51a-80a125045630_1600x993.png) Now, here’s your sneak preview of *Click,* starting with a fun story about one crazy week in Stockholm when Jake and two other Googlers co-founded Google Meet. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/cd4d9478-bd18-467c-80fa-6f78ceef9b12_1456x1103.jpeg) ### **The Stockholm solution** In January 2009, I took a weeklong business trip to Stockholm. At the time, I was a design lead on Gmail—a job I loved—and I was in Stockholm to work on an exciting side project I’d started with two fellow Googlers. But as I walked to the Google office on that cold January morning, trudging through the miserable snow in the miserable dark, I was miserable too, because that exciting project was about to die. My colleagues were Serge Lachapelle and Mikael Drugge. We’d met a couple of years earlier, in 2007, when Google bought their startup. The three of us started talking about an idea for a new product: video conferencing software that could run in a web browser. In those days, multi-way video calls were a hassle, so hardly anybody used them. We thought easy video calls could change the way people worked. So we started a project to bring this idea to life. We compiled a giant slide deck. We wanted to come up with the perfect strategy to get the rest of the company excited. I became obsessed with the idea of a 3D virtual conference room, and we added more and more ideas on top of it. Interactive documents and agendas and whiteboards and on and on… But folks at Google never quite got it. Weeks went by. Months went by. A year and a half went by. We kept improving our pitch deck, showing it to team after team, trying to make our proposal perfect. Then the global financial crisis hit. In January 2009, Google announced it was closing its offices in Trondheim, Norway, and Lulea, Sweden. I freaked out. I figured Stockholm was next, and, if so, Serge and Mikael would be gone. And we still didn’t have a coherent strategy. It was now or never. I booked a trip to Stockholm. So there I was, in the dark and the slush. I slogged down crooked streets until I found the Google office. I climbed the stairs of a drab gray building and found Serge and Mikael waiting for me with big smiles and a cup of hot coffee. This felt like our last chance. But what should we do? It was do-or-die. We had one week to make people care about our idea. What would grab their attention? It wasn’t the complex bells and whistles. It wasn’t the 3D virtual conference room, or the interactive meeting tools. No, our core hypothesis was simple: We believed people would want our product because it would be the fastest and easiest video call software on the market. So we set an audacious goal. It was Monday morning. By Friday, we agreed, we’d have a prototype of our software. Not a proposal, not a slide deck. A prototype. We would *show* everyone how great video calls in the browser could be with a prototype that made dead-simple multi-way video calls and nothing more. We didn’t have time to get things perfect, so we made quick decisions. We hashed out a design that was good enough. Then Mikael, an engineering genius, built the prototype. I remember the moment he got it working. He emailed me a link. I clicked and … boom. There was Mikael. There was Serge. Mikael said hello, Serge said hello, I said hello. We could see each other! At the end of the week, we shared our prototype, and finally—finally!—people understood. And they wanted it, immediately. Googlers started using our prototype for actual meetings, it spread across the company, and eventually the software launched to the public. Today it’s called Google Meet and has hundreds of millions of users. ### **The quest for the perfect start** I remember flying home from Stockholm and thinking, “Wow. That was different.” After wasting *two years* chasing a vague idea, we hit warp speed with a single week focused on the basics. Realization struck me—*bang*—like a two-by-four between the eyes: *Teams need a better way to start projects.* I couldn’t get that idea out of my head. Several months later, I created the Design Sprint—my attempt to turn the magic of Stockholm into a repeatable recipe. The method took off, and guiding teams through Design Sprints became my full-time job, first at Google, then as a partner at Google Ventures, and today as co-founder of the venture fund Character Capital. But shortly after we started Character, I noticed that, while Design Sprints are awesome for solving problems, building prototypes, and testing ideas, founders at the very beginning of big projects need more. They need a plan for standing out from the competition and they need to choose a direction for their first steps. In other words, they need the same kind of clear hypothesis that allowed Mikael, Serge, and I to cut away the fancy stuff and get to our core idea. Every team needs their own version of “the fastest and easiest video call.” I realized this was the next stage of my quest. So I sat down with my co-founder and co-author John Zeratsky (JZ). We re-examined the most successful projects we’ve seen firsthand as designers and investors. Early in our careers, JZ and I had the good fortune to work on products like Gmail, Google Ads, and YouTube, each of which found product-market fit quickly and clicked with customers for the long haul. As partners at Google Ventures, we had a hand in smash hits like Flatiron Health, Gusto, One Medical, Blue Bottle Coffee, and Slack. We knew what they had in common, and it started with a clear hypothesis. ### **The Founding Hypothesis** Teams that build winning products share some fundamental traits. They know their customers—and what problem they can solve for them. They know which approach to take—and why it’s superior to the alternatives. And they know what they’re up against—and how to radically differentiate from the competition. JZ and I call this combination of customer, approach, and differentiation a **Founding Hypothesis**. The Founding Hypothesis is a handy Mad Libs–style sentence that we can fill in to understand and test a team’s strategy. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/56fa747d-40b3-40fd-a03c-afa670e9a03e_1478x866.png) Yes, the Founding Hypothesis is simple, but that’s exactly what makes it so powerful. Products click with customers when they make a compelling promise—and that promise must be simple, or customers won’t pay attention. Every winning team JZ and I worked with made one of these simple, compelling promises. For example, when I worked on Gmail in the 2000s, our promise to customers was “We’ll solve your overflowing inbox problems better than Outlook, Hotmail, and Yahoo because we offer more storage and great search.” The promise was simple. People found it compelling, so they tried Gmail, and it delivered, so they told their friends, and so on. Today, Gmail is one of the world’s most used products. It clicked. Looking back today, we can easily reverse engineer Gmail’s Founding Hypothesis: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/a6e0add6-a418-465c-bd76-e323f9c90b3b_1600x1023.png) In our book *Sprint*,we profiled several startups that built products that clicked. Again, looking back, it’s easy to identify the simple promise each one made to customers*.* Here’s how the Founding Hypothesis could have looked for Blue Bottle Coffee, a startup we worked with in 2012: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/6c81a8b9-7fd3-4697-b306-abd81b44a4aa_1600x991.png) For Flatiron Health, a startup we worked with in 2014: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/b9f455fb-598c-4b5e-8d17-71b4a7f056b5_1386x830.png) And for Slack, a startup we worked with in 2015: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/53ffa743-2707-48d0-9437-43b840249488_1600x1007.png) Each of these startups turned into a valuable business, and each was eventually acquired for hundreds of millions, or even billions, of dollars. In each case, they made a simple promise that clicked. Unfortunately, the simple stuff is *not* easy. When I first began working with startups, I was embarrassed to ask founders basic questions like “Who are your competitors?” or “How will you differentiate?” because I didn’t want to waste their time or appear naive. But once I worked up the courage, I learned a surprising thing: If I asked three co-founders to write down their startup’s target customer, I got three different answers. If I asked a team what differentiated their product from the competition, I got a 60-minute debate. Smart, motivated people who respect their colleagues can still struggle to get on the same page. The obvious stuff is not always so obvious. And even if a team *does* have a plan, there is no guarantee it will work. For every Gmail, thousands of new products fizzle. Most teams can’t find the right promise. They don’t build what people want. There are just *so many ways* to get it wrong. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/29e7dd40-77d6-4fe8-9de9-d700a2e77c14_1600x771.png) Every big project is full of predictions, but they normally remain hidden behind the scenes. With the Founding Hypothesis, you grab those predictions, drag them onto center stage, and hit them with a dazzling spotlight. And you might not like what you see. Remember the Google Meet story? Imagine the unspoken Founding Hypothesis behind my original idea for a 3D virtual conference room: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/59f154c9-47cd-475d-a50d-7781e4759d95_1408x822.png) What was I thinking? “3D aficionados”? That’s absurd! But when you shine a spotlight on a hidden hypothesis, you might not like what you see. Compare that with the hypothesis we formed in Stockholm: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/ce3518cd-2f6d-4afc-a274-b4503d2e7072_1486x976.png) The pressure in Stockholm pushed us to get real and clarify our strategy. But how can you create the best possible hypothesis if you’re not facing an existential risk, like, *right now*? This is where the Foundation Sprint comes in. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/7455f511-e527-497f-a424-a704317316d7_1600x919.png) # The Foundation Sprint The Foundation Sprint is a two-day workshop for a team at the beginning of a big project. On the morning of day one, you’ll define the **basics** of your project; then, in the afternoon, you’ll craft **differentiation** to help you stand out from the competition. On day two, you’ll evaluate multiple options and choose an **approach** to your project. By the end, you’ll have a Founding Hypothesis: a clear statement of what you believe that can be proved (or disproved) with Design Sprints. The Foundation Sprint creates an artificial—but very effective—series of deadlines that force you to make big decisions fast. Here’s how it works: ### Get ready ❏ **Form a tiny team** Gather the leaders of the team: no more than five people, including the Decider (the real decision maker on the team). ❏ **Block two full days on the calendar** Six hours per day provides plenty of breathing room to finish the activities *and* take ample breaks. Some teams will finish these activities faster, but it’s good to have the full time available to ensure you can get into deep focus mode. ❏ **Stock up on supplies** You’ll need a real or virtual whiteboard, sticky notes, blank white paper, and dot stickers for voting. ## Day 1 ### Morning: Basics The Foundation Sprint starts with the Basics. The Basics are the so-obvious-they-get-ignored fundamentals that should inform every big decision about your project. Who, precisely, is your customer? What problem are you solving for them? What’s your advantage? Who or what are you up against? On the morning of day one, you’ll make it all crystal clear. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/c53b69b2-d18e-479c-a272-64ecc183ff9e_1116x1446.png) #### 10:00 a.m. ❏ **Choose a target customer (about 15 minutes)** Use plain language to make sure you’re talking about real people (for example, instead of “distributed enterprise orgs,” just say “teams with people in different locations”). ❏ **Choose an important customer problem (about 15 minutes)** Trying a new solution will cost your customers time, energy, and/or money. What problem causes enough pain to justify that cost? What is the biggest way you can help them? ❏ **Identify your advantage (about 20 minutes)** What gives you and your team an advantage over anyone else who might try to solve this problem? What capability, insight, and motivation set you far apart? ❏ **List your competitors (about 20 minutes)** What options do customers have for solving the target problem? These might be other products, workarounds, or even “do nothing” (which can be a warning sign, since it indicates the problem may not be important enough to merit a solution). Be sure to identify the most challenging competitor—the 800-pound gorilla that will be most difficult for your product to beat. **Note-and-Vote for fast, smart decisions** Use our “[Note-and-Vote](https://www.character.vc/note-and-vote)” technique for all of the above steps, and throughout the sprint: For each question, give everyone five minutes to think in silence and write answers on separate sticky notes. Then review the answers in silence, using dot stickers to vote for the best. Finally, the Decider calls for a short debate and makes a decision (regardless of the votes). ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/d56a2687-bc27-4a41-bb38-d0e982008a0d_1280x1600.png) #### 11:30 a.m.-ish ❏ **Take a break (about 30 minutes)** Get a snack or lunch. Stand up and stretch or take a walk. ### Afternoon: Differentiation Next is the crux of your hypothesis: differentiation. How can you best use your advantage? How can you reframe the decision for customers and make the 800-pound gorilla look like junk? #### 12:00 p.m.-ish ❏ **Differentiation classics (about 20 minutes)** Use dot voting to mark where your solution couldstack up against the competition on these classic differentiators. Be very optimistic and aggressive—but also realistic. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/c61fe39f-d82d-47fd-9590-779f16d5e7a1_1600x1069.png) ❏ **Choose your own differentiators (about 20 minutes)** Next, write your own differentiators, using criteria at which your solution could excel—and which would make the competition look crummy. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/f2dd51f2-4a2f-480d-a57a-71ffec9a5ba1_1600x871.png) ❏ **Score your differentiators, part 2 (about 10 minutes)** Review the custom differentiators. Once again, mark where your solution could stack up against the competition. ❏ **Choose differentiators (about 5 minutes)** Vote in silence, then the Decider chooses two differentiators to try first. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/cc4bd064-18c0-48ff-8bf9-b2c5acee29f9_1600x688.png) ❏ **Take a break (about 60 minutes)** Get a snack or lunch. Stand up and stretch or take a walk. #### 2:00 p.m.-ish ❏ **Make a 2x2 differentiation chart (about 30 minutes)** Differentiation charts are typically unhelpful slide deck filler. But a 2x2 chart focused on customer perception can be a powerful expression of your product hypothesis. Keep experimenting until you find differentiators that put you all alone in that top right quadrant and push the competition into the other three quadrants (which form an “L” shape that I like to think of as Loserville). But be honest. Differentiation only works if you can deliver what you promise. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/267b8fb4-07a8-44ef-8533-d25f2dc0cafa_1600x914.png) ❏ **Write principles (about 45 minutes)** To end the day, use the Note-and-Vote method to write two or three practical principles that will help you make decisions and deliver on your differentiation. For example, if one of your differentiators is “fast,” you might choose the principle “Fast is better than slow” to guide your future decisions. #### 3:30 p.m.-ish ❏ **Fill out a Mini Manifesto** Combine your differentiation and principles on one page. Hang it on the wall—this is the Mini Manifesto for your first prototype. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/67402c3e-320c-49d4-8621-c832c1c29757_1244x1600.png) ## Day 2: Approach ### Morning: Options On day two of the Foundation Sprint, you’ll execute what you might think of as a “pre-pivot.” Before you commit to one direction, you’ll generate alternatives and consider which is best. By the end of the day, you’ll choose an approach and be ready to experiment. #### 10:00 a.m. ❏ **List all possible approaches (about 60 minutes)** Use a Note-and-Vote to generate multiple alternative approaches to the project. These might be approaches you’ve previously considered or brand-new ideas. Write a one-page summary of each approach (what it is, why it’s a good idea in one sentence, and a doodle that shows how it might work). ❏ **Assign a color to each approach (about 5 minutes)** The Decider chooses up to seven options. Assign each a letter or a color (or both). This will make it easier to spot patterns later. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/f6bcb2fb-e057-4e84-88ae-3c2e8e1a3902_1600x672.png) ❏ **Take a break (about 60 minutes)** Get a snack or lunch. ### Afternoon: Lenses In the afternoon, you’ll choose an approach using our “Magic Lenses” technique. We developed Magic Lenses to help teams make complex decisions. Instead of trying to hold different perspectives and arguments in your head, you can use Magic Lenses to *see* those perspectives, then analyze, compare, and make a quick but good decision. #### 12:00 p.m.-ish ❏ **Make classic charts (about 15 minutes)** Start with four classic perspectives: customer lens, pragmatic lens, growth lens, and money lens. Make a 2x2 chart for each, like this: ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/eea7018a-f5a4-47c1-be56-fcff15cd18a6_1600x1040.png) ❏ **Place your options on each chart (about 45 minutes)** JZ and I find it easiest to plot one axis at a time, asking an expert on the team or the Decider to help with relative placement. “Where does this go? Higher or lower than this one? Sounds like more of a six than a five?” Feel free to adjust the criteria on the charts as you go—the default labels above are just a starting point. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/edf7d23e-301e-4c13-af45-36679f92d5b0_1600x1035.png) ❏ **Take a break (about 30 minutes)** Get a snack or lunch. #### 1:30 p.m.-ish ❏ **Make custom 2x2 charts (about 45 minutes)** What other perspectives could help you decide which approach is best? Consider what criteria might make for a good decision. We’ve seen teams plot their options using all kinds of criteria, from *no advertising required* to *founder excitement*, *customer pain*, *unique to us*,or *delivers on our mission*.It can also be interesting to try plotting your options using your differentiation chart from day one. Most teams try at least one or two additional charts beyond the four classics. ❏ **Zoom out and review (about 10 minutes)** Stand back or zoom out in your whiteboard software. Look for patterns. Does one approach win every time? Does one 2x2 strike you as the most useful lens? Do one or more contradict the others? ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/7067f8c0-d55c-45d0-b641-9515f7cd9213_1600x307.png) ❏ **The Decider decides (about 5 minutes)** Choose one top bet (your first-choice approach for the project) and one backup plan (the next approach to consider if you need to pivot). #### 3:00 p.m.-ish ❏ **Fill in the Founding Hypothesis** At the end of day two, you have all the elements for your Founding Hypothesis. Fill out this sentence: *If we help [**customer**] solve [**problem**] with [**approach**], they will choose it over [**competitors**] because our solution is [**differentiators**].* ## After the Foundation Sprint ### Test (and adjust) your Founding Hypothesis until it clicks A good Founding Hypothesis should pass the common-sense test. But that’s not enough. Every time I look at a Founding Hypothesis, I think, “Prove it!” Each prediction in the Founding Hypothesis lends itself to an existential, testable question about the project. To make these questions plain, JZ and I give startup founders a scorecard that adds a question to each prediction: Do you have the right customer? Right problem? Right approach? Will people *really* choose your solution over the competition? Do they care about your chosen differentiators, and will they believe your solution is radically better when judged by those criteria? And of course, above all: Does it click? ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/6de8121b-8dbc-40c3-ad8e-d20db67bbd05_1600x796.png) At the beginning of a big project, every team’s top priority should be identifying their Founding Hypothesis and running experiments to check off those damn boxes. You shouldn’t have to waste two years—like I did—and wait until your project is about to fail to figure this out. You, and your team and your customers, deserve better. Here’s how JZ and I do it with the startups in our portfolio: After a **Foundation Sprint**,we run **Design Sprints** to answer the questions on the scorecard and prove—or disprove—the Founding Hypothesis. Every week, founders test prototypes (using [techniques](https://www.lennysnewsletter.com/p/finding-your-bullseye-customer-michael-margolis) we learned from Michael Margolis during our time at GV), adjusting and repeating until the product clicks with customers. ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/b95eea4e-4f64-47cd-a4c2-fc92c8e1a05e_1600x753.png) These sprints allow teams to establish confidence that people want their solution *before* they spend loads of time and money building it. Yes, it’s just a simulation—but for our money, it’s a heck of a lot better than relying on a hunch. And sure, clearing the calendar for several consecutive days can be challenging, but it’s a bargain compared with wasting months building the wrong product. — *Thank you, Jake and JZ!* Click *comes out in April, but if you [order now](https://theclickbook.com), you’ll get instant access to free guides, templates, and resources that will show you exactly how to run a Foundation Sprint, step-by-step. [Order](https://theclickbook.com)* [Click](https://theclickbook.com) *[and learn more about the free bonuses here.](https://theclickbook.com)* ![Image from Introducing the Foundation Sprint: From the creators of the Design Sprint](https://substack-post-media.s3.amazonaws.com/public/images/c8985368-d5d5-4cc3-ae67-200d24b05bf2_1600x993.png) *Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week [Delphi](https://www.delphi.ai/) is hiring for a [Lead Product Manager](https://delphi-ai.notion.site/Delphi-Product-Manager-13fa9fbe9ec880218413f1b4f73322c4) based out of San Francisco. As their Lead PM, you’d get the opportunity to collaborate directly with their CEO to build and own an entire product category. **Why I think the company is interesting:** 1. Delphi powers [Lennybot](https://www.lennybot.com/)! They allow anyone to upload their content, to create a digital representation of not only their knowledge base but also their way of thinking. 2. With zero marketing spend, they’ve managed to go from 0 to nearly $2 million ARR in their first year. 3. They are backed by Sequoia, Founders Fund, and a host of tier-one investors. 4. They are still a small, scrappy, pre-series-A team. This is an opportunity to *actually* join at the ground floor and make a huge impact. If you’d like to get your profile sent directly to their team, just fill out this [quick form](https://airtable.com/apppne9fUg8CKw6s2/pag2vapuCgpIGh6WB/form). All submissions from my readers get priority (but no guarantees beyond that). — If you’re hiring, I run a white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), working with a few select companies to fill their open roles. Apply to work with us below. [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [5/46] Announcing the greatest product bundle ever: Get a year free of Granola, Notion, Superhuman, Linear, and Perplexity with an annual subscription # [See the updates to the deal here](https://www.lennysnewsletter.com/p/an-unbelievable-offer-now-get-one). ## (You can safely ignore everything below) ![Image from Announcing the greatest product bundle ever: Get a year free of Granola](https://substack-post-media.s3.amazonaws.com/public/images/54950b6d-11ef-43c1-a6e2-ee33d6e3c9c4_1456x1007.jpeg) I’ve partnered with my favorite product companies to offer annual newsletter subscribers a deal you won’t find anywhere else. Paid subscribers to the Lenny’s Newsletter annual plan will now get: 1. 1 year free of **[Granola](https://granola.ai/lenny)** for your entire company ($10,000+ value) 2. 1 year free of **[Notion Plus](https://www.notion.com/pricing)** with unlimited AI for 10 seats ($2,000+ value) 3. 1 year free of **[Linear](https://linear.app/)** for two seats ($336 value) 4. 1 year free of **[Superhuman](https://superhuman.com/)** ($300 value) 5. 1 year free of **[Perplexity Pro](https://www.perplexity.ai/pro)** ($240 value) I specifically selected these tools because they were among the most beloved products in my recent [“What’s in your stack” survey](https://www.lennysnewsletter.com/p/whats-in-your-stack-the-state-of). They represent the height of what products can be—they’re beautiful and fast, and they make your life better. This is all in addition to the existing benefits of being a paid subscriber: full access to every newsletter post, the 5-year back catalog, and an invite to a thriving members-only Slack community. Sounds like a no-brainer to me. This bundle is worth over $13,000, all for just $200. Many subscribers expense the newsletter to their company’s L&D budget. ### Important deal details: 1. You must be a **new customer** of the products to take advantage of the deal. 2. You must get an **annual subscription** to Lenny’s Newsletter to be eligible for this bundle. Monthly subscribers won’t have access to the deal. 3. If you request a refund or chargeback for your subscription early, **your codes will be deactivated**. 4. What you get: 1. **[Granola](https://granola.ai/lenny)**: One year of the Business plan for you and your team (up to 100 seats) 2. **[Notion](https://www.notion.com/pricing)**: One year of the Plus plan (plus unlimited AI) for you and your team (up to 10 seats) 3. **[Linear](https://linear.app/)**: One year of the Business plan (two seats) 4. **[Superhuman](https://superhuman.com/)**: One year of the Starter plan 5. **[Perplexity](https://www.perplexity.ai/pro)**: One year of the Pro plan Other than Perplexity (which I partnered with on my original bundle), **none of these companies have ever offered this kind of deal before.** ### How to redeem the deal: 1. Once you [become a paid yearly subscriber](https://www.lennysnewsletter.com/subscribe?), [click here](https://lennysbundle.com/) to redeem your codes. 2. If you’re on a monthly plan, [upgrade to yearly](https://www.lennysnewsletter.com/subscribe?), then [go here](https://lennysbundle.com/) to redeem your codes. 3. If you’re already a paid annual subscriber, [click here](https://lennysbundle.com/) to redeem your codes. **If you have any trouble or questions, email [support@lennyrachitsky.com](mailto:support@lennyrachitsky.com).** Sincerely, Lenny 👋 --- ## [6/46] Pulling back the curtain on the magic of Y Combinator *Annual subscription includes a free year of [Perplexity, Notion, Superhuman, Linear, and Granola](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle), along with access to the entire 5-year back catalog and a thriving members-only Slack community. [Subscribe now](https://www.lennysnewsletter.com/subscribe).* To honor the final day you can apply to [Y Combinator](https://www.ycombinator.com/)’s first-ever Spring batch (i.e. X25), I teamed up with past collaborator [Palle Broe](https://www.linkedin.com/in/pallebroe/) on the most in-depth and intriguing analysis you’ll find anywhere of the world’s most successful startup incubator. Palle spent over 100 hours digging through all available public data to answer questions like: 1. How has YC performed over the past 20 years? 2. What are YC’s biggest winners? 3. What impact does joining YC have on your startup? 4. What does YC look for in startups? 5. What is YC betting on most going forward? If this post inspires you, it’s not too late to start your application and apply to YC. [Apply here](https://www.ycombinator.com/apply) by 8 p.m. PT tonight! *For more from Palle, make sure to subscribe to his newsletter, [Rule of Thumb](https://palle.substack.com/), and follow him on [LinkedIn](https://www.linkedin.com/in/pallebroe/). [Learn more about Y Combinator here](https://www.ycombinator.com/about).* ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/abe9cb24-962a-4044-8d15-d2aebb5dba0a_2912x2375.jpeg) Y Combinator (YC) is widely regarded as the most successful startup accelerator and the top choice for world-class entrepreneurs. For those new to the startup world, Y Combinator is like a boot camp for startups. Selected founders receive $500,000 in seed funding in exchange for [roughly 7% equity](https://www.ycombinator.com/deal). Four times a year, founders are immersed in a three-month cohort, or “batch,” with about 125 other startups, which are sorted into groups and mentored by successful founders. It’s an exclusive program—with an approximately 1% acceptance rate—offering founders a powerful network, future funding opportunities, and instant credibility. The evidence is clear. Roughly 4.5% of YC companies become unicorns (in contrast to the 2.5% outcome for other venture-backed seed-stage startups),and around 45% of companies go on to raise a Series A(higher than the 33% average).So far, YC has funded more than 90 billion-dollar companies. Top YC companies include Stripe, Airbnb, Coinbase, and Reddit. But what does the data say about the impact of YC and the success of their bets? We tried to demystify the giant by reviewing the 4,939 YC companies so that startup founders, investors, and others can see what’s really going on—and maybe learn from the successes or trends. There was a lot that was surprising in the data. But one theme throughout was that the magic of the incubator appears to be in picking the best founders, understanding trends, and the process within YC, not about repeating the successes of one particular startup profile. Read on to learn more about what we uncovered. ### **Key takeaways** 1. **YC has gone from being a Consumer investor to primarily a B2B investor.** Consumer companies have resulted in over $200 billion of market cap, while B2B companies are currently privately valued at some $170 billion and are on the rise. 2. **Based on batch profiles, founders are betting on AI (specifically, B2B AI) to be the next big thing.** The most promising subcategories include “Engineering, Product, and Design,” Infrastructure, and Sales. 3. **Solo founders are at a disadvantage.** Although solo founders are encouraged, the data does show a steep decline in the number of them accepted to YC. 4. **Success has so far been driven by U.S.-founded companies.** More than 70% of the startups have been founded in the U.S., and to date, 99% of returns have come from the U.S. 5. **The durability of YC companies is significantly higher than that of the average startup.** More than 50% of companies are still alive after 10 years (vs. 30% average). 6. **The chances of startup success are higher with YC.** 45% secure Series A (vs. 33% average), 4% to 5% become a unicorn (vs. 2.5% average), and 10% achieve an exit. 7. **The investors in YC companies are the “crème de la crème.”** Tier 1 VCs frequently invest in YC companies, and some have made several hundreds of investments. #### ***Methodology*** - *Sources: Y Combinator, Harmonic, and Crunchbase.* - *Included 4,939 YC startups from the first batch in 2005 (S05) to the fall batch of 2024 (F24). The one-off education-focused batch (IK12) not included.* - *Not considering YC continuity investments.* - *Data is as of December 10, 2024 (or otherwise, as stated on the graphs).* - *Some companies do not disclose valuation or acquisition price, so sometimes the data will not contain all companies. Treat the numbers below as directional, not as exact figures.* ## What does YC success look like? #### **Around 10% of YC companies achieve an exit** As of today, 17 companies have been backed by YC and gone public. Another 76% are still actively pursuing an exit. But one impressive fact is that only 13% of companies have gone out of business. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/c8112c55-55cc-45f7-80bc-40b2e1fe3a21_1583x1123.png) #### **More than 50% of companies are still active after 10 years** YC companies are more durable than other startups, which have an average lifetime of around five years. Batch W15, which is now 10 years old, still has more than 50% of its companies actively operating. All other batches since then have more than 50% of companies still operating. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/6907957f-3dac-4c57-a09f-01a41f3b31af_1600x1028.png) #### **Between 25% and 50% of companies get acquired after 10 years** A measure of success for any incubator or investor is the number of companies that achieve a liquidity event, either via an IPO or an acquisition. And no one does that better than YC. When you look at the mature batches, we are seeing between 25% and 50% of companies getting acquired. This speaks volumes about the quality of companies in YC. This is in contrast to the 15% of global companies that get acquired if they have raised at least one round of funding. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/fe4484e3-3486-4d82-87d5-5a0dd1e550d8_2500x1266.png) ## What are the biggest YC winners? #### **Consumer companies have driven the majority of YC returns** By far the largest wins for YC have come through some of their early consumer-focused companies. As a matter of fact, the four companies with the highest market capitalization are consumer companies: Airbnb, Coinbase, DoorDash, and Instacart. It’s no secret that technology startup outcomes follow a power law distribution, and that is also very much the case for YC, where the top four companies—Airbnb, DoorDash, Coinbase, and Instacart—account for more than 84% of the market value created. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/f9c4a0e4-6321-4a98-b92d-5718b3d73a7e_1728x2304.png) #### **More than $600 billion has been created in market value** YC companies attract a lot of funding. In total, more than $145 billion has flowed into the startups that were part of YC. This investment has led to the creation of more than $300 billion in market value, and with the existing private companies, this amount could double to more than $600 billion in the future (once private companies exit). Although acquisitions are the most common path for YC startups, they have “only” brought in $15 billion. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/b02a02db-b4dc-485a-9655-dfe6348b754b_1583x1123.png) #### **Consumer companies top the funding charts as well** Cruise and Stripe are the two most well-funded YC companies ever. Cruise requires large amounts of capital due to the expenditures required in the autonomous vehicle industry, and Stripe has been focused on blitzscaling their product globally—also requiring a lot of funding. Two-sided marketplaces such as Airbnb, Instacart, DoorDash, and Rappi are notoriously expensive to get started and require subsidizing one side of the marketplace. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/59efe5de-455d-4078-9f8c-84200167e528_1583x1123.png) #### **Stripe is the next big win for YC** Currently sitting at $70 billion in valuation and estimated to go public in the next few years, Stripe is gearing up to have one of the largest-ever outcomes for YC. Companies riding the AI wave such as Scale AI have seen a rapid increase in valuation and are likely to further benefit from the tailwinds, which could result in even higher valuations in the coming years. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/c80b5bf2-bcbc-4b7c-984c-918db918a2f6_1583x1123.png) Below you can see the capital efficiency of each of these private companies. That is the amount of funding relative to their valuation. Deel, Benchling, and Gusto are showing particularly strong capital efficiency ratios. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/b285590b-ac0d-43cf-8d89-2306d06521d3_1583x1123.png) #### **W09, S12, and S13 remain the best YC batches based on IPOs and acquisitions** During Paul Graham’s leadership, YC was fortunate enough to invest in some of the most industry-defining consumer companies. Airbnb, DoorDash, and Instacart were riding the shift to mobile, and Coinbase became the leader in the crypto revolution. Today these are the top companies YC has ever backed based on market cap value. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/88e7f364-5e47-4bde-94dd-3209d7065198_1583x1123.png) #### **The most promising batches are S09, W17, and W16** When we look ahead, there are some very interesting batches that may soon materialize on the public markets. S09, with Stripe, might overtake Airbnb as the most valuable YC company ever. W17, with Rippling, Faire, and Brex, is also looking exciting. Personally, I believe there is a lot of upside in S16, which is the vintage with Scale AI, one of the hottest AI companies out there now. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/363b635e-9fcd-4d03-a83e-a4ca9289e17b_1583x1123.png) ## What are the most successful conditions for YC startups? #### **YC has gone from being a Consumer investor to primarily a B2B investor** Although YC are famous for their Consumer investments (e.g. Airbnb, DoorDash, Reddit, Coinbase), the majority of their funding has actually gone into B2B companies. This is particularly true for their most recent batches, where more than 75% of the companies are defined as B2B companies. The shift away from Consumer is likely driven by the fact that most of the large consumer apps have already been invented. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/ad0d7a06-7904-4bc0-9c36-6b2f3653886c_2500x1607.png) #### **Future returns will be driven by B2B and fintech** The best past investments from a returns point of view have been great consumer startups like Instacart, Airbnb, DoorDash, Coinbase, and others. However, looking forward, the large exits for YC will come from B2B and fintech companies with combined valuations north of $270 billion. So far, YC has not had financial success with their investments in Healthcare, Bio and Industrials, or Real Estate and Construction. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/46a3282a-c9ba-49f6-9021-ecff7aae465a_1583x1123.png) #### **By far the majority of investments go to U.S.-based companies** It is probably no shock that the largest number of YC-backed companies are based in the U.S. More than 70% of companies are from the U.S., followed by countries such as India, the U.K., Canada, Mexico, and France. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/cf9c1f03-c66d-4446-80ee-6c62ef6e105e_1728x2304.png) #### **So far, 99% of the returns are coming from the U.S.** The only meaningful returns have so far come from U.S. investments. There are some promising bets from overseas, such as Razorpay at $7.5 billion (India), Rappi at $5 billion (Colombia), Zepto at $5 billion (India), Meesho at $4 billion (India), and Flutterwave at $3 billion (Nigeria). Slightly surprising is that European-based companies are still lagging behind when it comes to valuations and market cap value. The most valuable European YC company is GoCardless at $2.1 billion. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/1002b93c-9a19-409b-a327-7eb4f1088bf0_1583x1123.png) #### **Solo founders are at a disadvantage** There’s a lot of discussion of whether being a solo founder makes it statistically harder to succeed (or get into YC). When Sam Altman was the president of YC, he wrote, “We really prefer at least two founders, but it’s not a deal-breaker.”And the data speaks quite clearly. Over the past few batches, the proportion of solo founders in the batch has gotten smaller and smaller, to around 10% in the latest batches. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/07b1aa44-f086-49f0-b391-e1bb7328d5d8_2500x1607.png) ## What is the likelihood of success for YC startups? #### **40% to 45% of companies that go through YC raise a Series A round** Getting to Series A is no small feat for startups. According to Carta, on average only 30% to 35% of companies that raise a seed round can expect to also raise a Series A funding round. While batches differ, the vast majority (80% to 90%) of YC startups manage to close a seed round with investors outside the initial YC investment. Following that, around 40% to 45% of matured batches reach at least a Series A—significantly better than the industry average. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/896aeb28-6733-464c-b83a-b499517e64b5_1583x1123.png) #### **4% to 5% of companies become unicorns, about twice the industry average** It takes a bit of time for companies to mature. But if we look at vintages seven or more years after launch, roughly 4% to 5% of companies become unicorns. This is really impressive, particularly vs. the industry average of 2.5%. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/11cef644-897e-4f1e-b195-894168e5720b_2500x1607.png) #### **Funding rounds for YC companies are significantly higher than the global average** ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/a118366e-0580-49b3-8841-2faf4a3ef8ba_1583x1123.png) ## Who are the most successful YC investors? ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/ab38cd3f-f1d1-4aa4-890b-8fa3d642ebda_1728x2304.png) Reaching unicorn status is one of the core milestones for any private startup or investor. Below are the number of times investors have invested in a YC unicorn. *One very important note here: This shows the number of investments they made in unicorn companies; however, not necessarily before they were unicorns, i.e. for some of these, they invested when the company was already a unicorn company. This data is from Crunchbase and Harmonic, which is sourced from public announcements and thus isn’t complete but is directionally correct.* ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/05f13584-0465-4a5d-b71f-3b13efed28cc_1728x2304.png) ## What are YC founders betting on for the future? #### **In the new batches, it’s all about AI** A pretty natural question to ask is: What kinds of companies have YC founders been working on in the past few years? While YC’s ethos is all about picking the right founders/teams, it is useful information to know if you’re planning on applying to YC or even just interested in getting a feel for the latest trends in technology. The answer: AI. ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/24a3e086-f399-45b0-8009-ddba75d7b902_1583x1123.png) #### **Along with AI, founders are betting on B2B** First of all, YC has stopped specifically labeling companies as AI, Gen AI, AI Assistant, etc. I assume this is because all companies are essentially AI companies now—as in, they are leveraging AI technology in some capacity. But make no mistake, YC founders are betting big on AI in all forms. There is still a preference for starting B2B software companies, with over 63% being B2B-focused startups. Interestingly, the second biggest category is within Healthcare (10%), which has historically not been the case. As seen in the above, both from a funding and returns POV, that has not been a category that has done particularly well. This seems to be changing now with the myriad companies trying to revolutionize Healthcare with AI. It also seems like Fintech/Crypto is taking a little bit of a backseat right now, with only 6% of investments, despite some of the biggest winners coming from this category (Stripe, Coinbase). When analyzing the tags added by YC for the various companies, it’s pretty spread out, but three categories seem to have gotten the attention of founders: 1. Engineering, Product, and Design (25%) 2. Infrastructure (12%) 3. Sales (8%) ![Image from Pulling back the curtain on the magic of Y Combinator](https://substack-post-media.s3.amazonaws.com/public/images/108debad-1222-46c3-a7ec-fb530ce3a1c1_1583x1123.png) ## **To wrap it up…** The YC magic is about picking the world’s greatest founders and following the internal YC process, not repeating the same startup profile. Throughout the past 20 years of YC, different technological and societal trends have emerged that have shaped the big startup winners. What has stayed consistent is the ability to pick the best founders who are tackling the biggest opportunities. Today there are inflection points shifting founders away from Consumer and toward B2B, and all systems are go on everything AI. The most successful YC companies of the past (Airbnb, Coinbase, DoorDash, Instacart, etc.) don’t look like the most successful YC companies of the present (Stripe, Rippling, Brex, etc.), and they won’t look like the most successful YC companies of the future either (Scale AI and the new generation of AI companies). *Thank you, Palle! For more from Palle, make sure to subscribe to his newsletter, [Rule of Thumb](https://palle.substack.com/), and follow him on [LinkedIn](https://www.linkedin.com/in/pallebroe/). Also, thank you, [Rebecca,](https://www.rebeccaackermann.com/) for editing.* *Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week **[Bounce](https://bounce.com/)** is hiring for a Head of Growth based out of San Francisco. As their Head of Growth, you'll be expected to: 1. Lead strategy and execution across multiple channels, including product-led growth, SEO, and paid acquisition 2. Mix high-level thinking with hands-on execution as you build out Bounce’s growth team **Why I think the company is interesting:** - Bounce is one of the fastest-growing European startups ever. - They’re now in hypergrowth with a lean and scrappy team. This is an opportunity to join before they really mature. - They are backed by a16z and a host of tier-one investors. They just raised a Series B and are quickly expanding into a global presence. - This is not a 0-to-1 role; it’s a 1-to-10 role. Bounce’s growth is already epic, but they need a leader to take it to the next level. If you’d like to get your profile sent directly to their team, just fill out [this quick form](https://airtable.com/appTNAXC72V74HqPS/pagCczxszIZPdd00t/form). All submissions from my readers get priority (but no guarantees beyond that). *If you’d like to get your role featured here, [apply here](https://airtable.com/apphKJD0wLIHfPcZu/pag09IdQYrMpolMa5/form). If you’re interested in working with my white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), apply to work with us [here](https://www.lennysjobs.com/).* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [7/46] How much product managers make in the U.S., Europe, and Canada I love sharing data you won’t find elsewhere, so I’m excited to publish exclusive, in-depth salary benchmarks for product managers in the U.S., Europe, and Canada. This data is courtesy of [Pave](https://www.pave.com/), which gets its data by integrating with human resources information system (HRIS) software and equity management systems at over 7,500 top companies to build real-time compensation benchmarks. The data below is based on actual comp from over 23,000 product managers globally, plus a few select data points from [Levels.fyi](https://www.levels.fyi/). Below, you’ll find: 1. **Base salary, equity, and total comp for product managers in the U.S.** 2. **How U.S. comp compares to Europe, Canada, and Tier 2 and Tier 3 U.S. cities** 3. **A deeper look at Europe and U.K. compensation** ### **Key takeaways** 1. The median starting base salary for a PM in the U.S. is **$112,000** at a public company and **$96,000** at a private company. 2. A 90th-percentile senior individual-contributor (IC) PM can hit **$350,000** in base salary and close to **$1,000,000** in total comp. 3. New PM managers make less than senior ICs—**$265,000** vs. **$425,000** in total comp, respectively. 4. Moving from a Tier 3 U.S. city to a Tier 1 city can increase your salary by **20%**. 5. The median chief product officer makes **$1,425,000** in total comp, and the 90th percentile makes nearly **$5,000,000**. 6. The biggest salary jump happens going from L4 to L5 and M3 to M4. 7. PM base salaries are **15% to 20%** lower at private companies vs. public companies. They are over 25% lower for product execs (E7-E9). 8. U.K. PMs make **65%** of what the average U.S. PM makes, and the average European PM makes less than **50%** of the average U.S. PM. 9. If you want to maximize your base salary (i.e. stability), join a public company in a Tier 1 U.S. city. 10. If you want to maximize your upside, get into a senior IC or manager role at a private company in the U.S. ## First, an explanation of the levels you’ll see in the charts below Here’s how the 12 levels you’ll find below map to seniority: - P1 → Entry - P2-P3 → Mid - P4-P6→ Senior - M3-M4 → Manager - M5-M6 → Director - E8 → VP - E9 → CPO And here’s how these levels map to titles at a typical company, including where IC and manager tracks overlap: ![Screenshot 2024-08-26 at 10.18.08 AM.png](https://substack-post-media.s3.amazonaws.com/public/images/bce7cbc4-f248-48a1-8ca6-fec903c4646c_532x306.png) *For more, [check out this detailed explanation of levels](https://support.pave.com/hc/en-us/articles/4430985339799-Job-Levels) and how they may map to your organization.* ## **Product management compensation in the U.S.** Here are the 25th, 50th, 75th, and 90th percentiles for total comp for product managers in the U.S.: ![Image from How much product managers make in the U.S., Europe, and Canada](https://substack-post-media.s3.amazonaws.com/public/images/50f31a56-750f-4ce9-aa9b-f200bad7bab1_2913x2107.png) #### Top takeaways 1. The median starting total comp for a PM in the U.S. is **$139,000**. 2. 90th-percentile IC PMs make close to **$1,000,000** in total comp. 3. New managers make less than senior ICs. The median new manager makes **$265,000** in total comp per year—equivalent to an L4 IC PM. You have to hit M5 to make more than the most senior median IC PM. 4. The median chief product officer makes **$1,425,000** in total comp, and the 90th percentile makes nearly **$5,000,000**. 5. The median VP of product makes **$707,000** in total comp, and the 90th percentile makes over **$2,000,000**. 6. The biggest salary jump happens going from L4 to L5 and M3 to M4. And then from VP to CPO. 7. At [Meta](https://www.levels.fyi/companies/facebook/salaries/product-manager?country=254) (the highest comp of the Magnificent 7), the average M6 makes over **$2,000,000** in total comp, and an entry-level (non-associate product manager, or APM) PM makes **$232,000** in total comp. ### A deeper look at **salary and equity** Here’s the same data as above but splitting base salary and equity and comparing public and private company comp: ![Image from How much product managers make in the U.S., Europe, and Canada](https://substack-post-media.s3.amazonaws.com/public/images/89ccc297-8db0-4d02-ab57-c62e3f759121_1457x1849.png) #### **Top takeaways** 1. The median *starting base salary* for a PM in the U.S. is **$112,000** at a public company and **$96,000** at a private company. 2. A 90th-percentile IC PM at a public company can hit **$350,000** in base salary. 3. PM base salaries are 15% to 20% lower at private companies vs. public companies. They are over 25% lower for product execs (E7-E9). 4. Unsurprisingly, lower base salaries at private companies are made up for by more equity, especially when things go well. 90th-percentile IC PMs at private companies earn nearly **$600,000** in equity/year (vs. **$250,000** at public companies), and 90th-percentile PM managers (M6) earn around **$1,000,000**/year in equity (vs. less than **$500,000** at public companies). 5. The amount of equity 90th-percentile PMs can earn grows very quickly as they rise in the ranks. For example, at a private company, a 90th-percentile M3 earns just over **$200,000** in equity, an M4 earns **$500,000**, an M5 earns **$800,000**, and an M6 earns **$1,000,000**. 6. At [Tesla](https://www.levels.fyi/companies/tesla/salaries/product-manager), [Microsoft](https://www.levels.fyi/companies/microsoft/salaries/product-manager?country=254), [Nvidia](https://www.levels.fyi/companies/nvidia/salaries/product-manager?country=254), [Apple](https://www.levels.fyi/companies/apple/salaries/product-manager?country=254), [Meta](https://www.levels.fyi/companies/facebook/salaries/product-manager?country=254), [Google](https://www.levels.fyi/companies/google/salaries/product-manager?country=254), and [Netflix](https://www.levels.fyi/companies/netflix/salaries/product-manager?country=254) (the Magnificent 7), the starting base salary for a (non-APM) PM is $136,000, $156,000, $158,000, $158,000, $167,000, $199,000, and $343,000, respectively. Netflix and Tesla are low on equity, while Meta, Google, and Nvidia are quite high. ## **Comparing Tiers 1, 2, and 3 U.S cities to** U.K., Canada, and Europe broadly Now let’s look at how other parts of the world compare to the U.S.’s average PM compensation. ### **Base salary differentials relative to U.S. average** ![Image from How much product managers make in the U.S., Europe, and Canada](https://substack-post-media.s3.amazonaws.com/public/images/80058808-ef2b-4691-b5ac-2ceed42a5c66_2912x2008.png) #### **Top takeaways** 1. Tier 1 U.S. PMs make **10% more** than the average U.S. PM. 2. U.K. PMs make **65%** of what the average U.S. PM makes, Canadian PMs make **60%**, and the average European PM makes less than **50%**. 3. Moving from a Tier 3 U.S. city to a Tier 1 city can increase your salary by **20%**. 4. The average PM in a Tier 3 U.S. city makes more than the average PM in Europe or Canada. ### Deeper look at Europe and U.K. cash compensation vs. U.S. ![Image from How much product managers make in the U.S., Europe, and Canada](https://substack-post-media.s3.amazonaws.com/public/images/50c040a9-6465-486c-bdfb-df3bdfb560fc_2912x2064.png) #### **Top takeaways** 1. If you’re a PM in Europe, you want to be in the U.K. or Ireland. 2. Sorry, PMs in Bulgaria. ### **Equity differentials relative to the U.S. average** ![Image from How much product managers make in the U.S., Europe, and Canada](https://substack-post-media.s3.amazonaws.com/public/images/250b9b03-1665-4fc2-a5f7-ce7745e09255_2912x1986.png) #### **Top takeaways** 1. PMs in Tier 1 U.S. cities make **40%** more from their equity than in Tier 2 and 3 cities. 2. PMs in Tier 2 and 3 cities make about the same in terms of equity. 3. Living in Europe, you make **40%** of what U.S. PMs make from equity. See anything else interesting in this data? Share it in the comments! [Leave a comment](https://www.lennysnewsletter.com/p/how-much-product-managers-make-in/comments) Next in this series, I’m going to do an analysis of salary and equity benchmarks for *founder* compensation. Stay tuned! *Thank you, [Pave](https://www.pave.com/), for providing this data, [Ben](https://www.linkedin.com/in/benhawkins0/) for design, [Rebecca](https://www.rebeccaackermann.com/) for editing, and Ricardo and Dennis for providing feedback on early drafts. Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week **[Bounce](https://bounce.com/)** is hiring for a Head of Growth based out of San Francisco. As their Head of Growth, you'll be expected to: 1. Lead strategy and execution across multiple channels, including product-led growth, SEO, and paid acquisition 2. Mix high-level thinking with hands-on execution as you build out Bounce’s growth team **Why I think the company is interesting:** - Bounce is one of the fastest-growing European startups ever. - They’re now in hypergrowth with a lean and scrappy team. This is an opportunity to join before they really mature. - They are backed by a16z and a host of tier-one investors. They just raised a Series B and are quickly expanding into a global presence. - This is not a 0-to-1 role; it’s a 1-to-10 role. Bounce’s growth is already epic, but they need a leader to take it to the next level. If you’d like to get your profile sent directly to their team, just fill out [this quick form](https://airtable.com/appTNAXC72V74HqPS/pagCczxszIZPdd00t/form). All submissions from my readers get priority (but no guarantees beyond that). *If you’d like to get your role featured here, [apply here](https://airtable.com/apphKJD0wLIHfPcZu/pag09IdQYrMpolMa5/form). If you’re interested in working with my white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), apply to work with us [here](https://www.lennysjobs.com/).* [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [8/46] Strategy Blocks: An operator’s guide to product strategy Below, you’ll find what I believe is the most actionable, specific, and straightforward framework for crafting a strategy, for both your product and your company. [Chandra Janakiraman](https://www.linkedin.com/in/chandramohanj/) came on [my podcast](https://www.youtube.com/watch?v=WFLH8Af2f30) to share it, and afterward, we realized that it would be 10x more helpful as a newsletter post. So we teamed up to make that happen. As Chandra shares below, his framework sits on top of the best strategy wisdom out there (e.g. [Michael Porter](https://hbr.org/1996/11/what-is-strategy), *[Good Strategy Bad Strategy](https://a.co/d/ilE3kQq)*, *[Playing to Win](https://a.co/d/4NQzHmz)*), and turns that advice into a step-by-step guide that you and your leaders can put into practice immediately. The post includes plug-and-play strategy templates, recommended timelines, the stakeholders to involve at each step, and more 🔥 For more from Chandra, [follow him on LinkedIn](https://www.linkedin.com/in/chandramohanj/), and [VRChat is hiring](https://jobs.lever.co/vrchat)! ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/25acdc8f-9f25-4884-9946-213d4acea5cc_2912x2029.jpeg) At Headspace back in 2016, we had established our product roadmap and success metrics and our mission and vision, but teams were still confused about *why* we were working on the projects we chose. Without that clarity, teams weren’t able to get several projects off the ground and had unproductive debates on what to build. I worked closely with a seasoned board member to trace this back to a lack of product strategy—both articulated and aligned. With her help, I wrote the first strategy document for Headspace, which eventually led to the complete [reimagination of Headspace](https://youtu.be/llhF8ecRsSk?si=XVdwKDfm00bl7GAR), maximizing growth for our guided mindfulness product and adjacent spaces. I translated our journey from confusion to clarity into bite-size steps to create a product strategy, called Strategy Blocks. Over the past 10 years, I’ve used Strategy Blocks in six different high-stake contexts, at companies large and small, including [Headspace](https://www.headspace.com/), [Meta](https://about.meta.com/), and [VRChat](https://hello.vrchat.com/). It has helped in gaining alignment on complex topics with senior leaders at Meta (Sheryl Sandberg, Chris Cox, and Andrew Bosworth), paving the way for key launches like [Facebook digital well-being tools](https://x.com/facebook/status/1248698109271539712), [privacy protections for youth](https://about.fb.com/news/2022/11/protecting-teens-and-their-privacy-on-facebook-and-instagram/), and [Quest referrals](https://www.uploadvr.com/oculus-quest-2-referral-program/). It has also jump-started innovation at VRChat on the [VRChat+ subscription product](https://youtu.be/fyl8v_eia5w?si=IB2QqamPtmswTm8F), with a strong [community response](https://youtu.be/C9RMGv5gMdE?si=12CWSuo-ooLzrAg9), while helping to steer VRChat toward an exciting long-term future. Strategy Blocks is particularly helpful for product leaders looking to upgrade their strategy skills and for chief product officers looking for a comprehensive strategy architecture that balances action and aspiration. ### Basic definitions Strategy has benefited from several excellent foundational frameworks over the years, from [Michael Porter’s work](https://hbr.org/1996/11/what-is-strategy) to *[Good Strategy Bad Strategy](https://a.co/d/ilE3kQq)*, *[Playing to Win](https://a.co/d/4NQzHmz)*, and *[The Art of War](https://a.co/d/dxzsLah)*, to name a few. Strategy Blocks uses concepts from many of these frameworks and converts them into bite-size steps to map out the future of your company. Let’s establish some basic definitions before diving into the details of Strategy Blocks: #### 1. What exactly is product strategy? - **Product strategy sits in between the mission and vision and the plan, either at the company level or at the team level.** At the company level, the mission and vision is typically articulated by the founders/CEO and tends to be durable over time. The plan or roadmap is basically an ordered list of projects based on some notion of prioritization and sequence of delivery. There is a steep drop in elevation between the mission/vision and the plan, and strategy occupies this large void. - **Strategy exists to force a disciplined choice to deploy scarce resources for maximum impact.** Regardless of the size of a company, the resource pool and capacity to get work done is **always constrained** relative to the universe of work that could be done—making this choice a critical decision in every single context. - **A good strategy articulation typically includes three components**: - 3 to 5 areas for the company or the team to focus on, which we will henceforth refer to as **strategic pillars** - Several areas that should explicitly *not* be the focus - A clear set of explanations for why these choices were made - **Strategy Blocks consists of two sub-parts:** - **Part 1:** The crafting of a 2-year strategy, which is typically focused on solving problems with the current product, henceforth referred to as **small “s” strategy** - **Part 2:** The crafting of a 3/5/10-year strategy, focused on aspirational futures, henceforth referred to as **big “S” strategy** ## Part 1: Crafting the 2-year strategy (small “s” strategy) Crafting the small “s” strategy requires about 8 to 12 weeks to produce a **high-quality output** ([see output template](https://docs.google.com/document/d/1iNeYUaMnpicvkpVZO-gj9cCxLeHfWN0xtGm_QoxgemE/edit?usp=sharing)). The weeks are split into five distinct steps: 1. Preparation 2. Strategy Sprint 3. Design Sprint 4. Document Writing 5. Rollout This process is typically led by a senior product leader. ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/2314d431-64dd-41af-9c52-905c4d5722a0_1457x1461.jpeg) Let’s dive into detailed guidance for each step. ### Step 1: Preparation (3-5 weeks) ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/8fc5d7c3-4dda-4f7d-8669-bc9eb35c5a24_1344x896.png) The preparation step is a foundational effort where a lot of the groundwork and due diligence is done to inform the strategy selection process. The work here is not full-time and can run alongside other projects for the people involved. Let’s go through the steps and action items. 1. **Strategy working group:** The product leader should assemble a core strategy working group, at a minimum, consisting of an engineering lead, a design lead, and a data lead, besides the product lead themselves. Other functions could be added depending on availability/resourcing (e.g. product marketing, user research, content design, etc.). The product leader should set the stage through a kickoff meeting and explain the upcoming 8-to-12-week process, helping people understand their roles and commitments through the process. 2. **Discrete action items:** An important element of the kickoff meeting is assigning discrete action items to the different members of the strategy working group and making expectations clear on the output needed from each person. 1. **Behavioral insights:** The data lead is responsible for producing a comprehensive meta-analysis of past behavioral analyses available at the company. Some net-new work could be commissioned to close critical gaps in understanding user behaviors, especially if these gaps are identified early. 2. **UXR insights:** The design lead or the user researcher (if available) is responsible for producing a comprehensive meta-analysis of past research and user insights work (both qualitative and quantitative). There is usually not enough time in the preparation step to commission net-new research work, but scrappy studies could be mobilized to close any critical gaps in understanding users. 3. **Leadership interviews:** This is a critical input to the process and is often forgotten or skipped for various reasons. Key leaders like founders/CEO have really important nuggets of wisdom tucked away in their back pockets, which structured leadership interviews can unearth. And I’ve found that these leaders typically enjoy engaging in these kinds of conversations as a creative outlet, and a channel for their desires and frustrations. Since there are typically multiple leaders that need to be interviewed, this load should be distributed across the strategy working group members as distinct one-on-one sessions. Some of the questions that I recommend asking are: 1. **What does success look like? What does failure look like?** 2. **What are some measures of success?** 3. **What are principles to keep in mind while building the product?** 4. **Why have things not worked in the past?** 5. **What are some of the leader’s favorite/pet ideas?** 4. **Competitive and comparables analysis:** Competitors are players operating in the same space in an effort to win a similar set of customers. Comparables are players that are tackling similar problems but perhaps in a different market/space. Both should be included in this analysis. The product lead or the product marketing lead (if available) should run this analysis with a clear output consisting of the following components: 1. A head-to-head summary comparison of the features and strengths of each player, with color coding or a heat map to easily understand areas of strengths and weakness 2. Commentary of where each player seems to be focusing and their themes of investment 3. Key feature deconstructs and screenshots 5. **Adjacent roadmaps:** This is applicable only in large companies where one team’s strategy could be influenced by an adjacent team’s investments and roadmap. When this is the case, it’s important for the team to download adjacent teams’ strategies and roadmaps as a key consideration. Sometimes these can amplify certain areas of investment or discourage other areas of investment. 6. **User observations:** Each of the strategy working group members should watch between one and three user interviews or sit in on live user observation sessions and write down their takeaways to share with the rest of the group. This is not intended to be used as insights but is meant to build empathy toward users and ground the strategy process from being abstract to being human-centric. All the above material is consolidated into a comprehensive share-out deck. People should spend zero effort in polishing this deck. The focus is really on substance over style, and the consolidated deck will power the next step of the process: the strategy sprint. ### Step 2: Strategy sprint (1 week) ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/9187a478-d334-44be-b60d-2f0286fff24a_1456x970.png) The strategy sprint is the heart of the strategy process. This is where strategy selection happens. The product lead facilitates and the full strategy working group participates. Let’s walk through a day-by-day account of how to run the sprint. 1. **Comprehensive share-outs:** Day 1 is all about share-outs from the consolidated deck. Members of the strategy working group who have prepped individual sections run through their respective slides, while the rest take notes in the form of problems that are affecting users or preventing the product from growing. These notes are critical to power day 2’s process. Overall, day 1 is about achieving a shared sense of consciousness about the “state of the union.” 2. **Core process:** Day 2 of the strategy sprint is the single most important day of the whole 8-to-12-week strategy process. This is where the actual strategic pillars are selected. Let’s walk through the steps to get there: ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/b6b6c198-1fa2-4d36-993c-1d1140fcb178_1456x1378.jpeg) 1. **The first step is for the team to generate a list of problems based on the notes from day 1.** Each member of the strategy working group adds their problem statements to a shared repository (e.g. Google Sheets). Typically anywhere between 50 and 150 problem statements should be collected this way. The importance and quality of each problem is not important in this first step. Focus on generating a high volume of problem statements. 1. **Next, cluster related problems into broader themes, and give each theme a relatively short and pithy name.** Typically, 10 to 15 problem clusters emerge from 50 to 150 problem statements. 2. Flip each problem cluster into an opportunity framing (e.g. difficulty finding features → discovery, unwanted content → relevance, etc.). 3. **Next, score each opportunity area along four key dimensions:** 1. **Expected impact:** Think about the number of people impacted by the problem, the frequency of impact, and the relative pain that people experience in that area. 2. **Certainty of impact:** Establish if the need is well-proven or more speculative at this stage. 3. **Clarity of levers:** Understand if the team has a rough sense of the levers that could be deployed to improve the situation. 4. **Uniqueness of levers:** Assess if the levers can have a unique and differentiated experience as a result of being built by the team/company. This usually relates to unique competencies or pre-existing advantages the team/company might have (e.g. brand advantage, network advantage, skill advantage, technical advantage, process advantage—some secret sauce that has helped the company differentiate in the past that can also be applicable to these levers). ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/7e40c9be-77d0-4a54-842c-f70ea8cca3d8_1236x824.png) 4. In situations where hard data isn’t available, qualitative scores based on team debate are a reasonable fallback. Evidence will eventually come with execution. 5. Once these scores are generated across the 10 to 15 opportunity areas, a simple sum of scores and a sort based on totals helps automatically bubble up the top three opportunity areas. These become the **strategic pillars**. 6. For each strategic pillar, generate about three “how might we’s” as a way to spark ideation. This is a preparatory step for the next step, the design sprint. 3. **Winning aspiration:** Once the strategic pillars and the “how might we’s” are established, kick off day 3 with a creative exercise. Imagine a day two years into the future when the team has made significant progress on all three opportunity areas, and a journalist covers the company/team’s work. Ask the team to imagine the **headline**, and express it in the form of how the team’s work has made the consumers’ lives better and grown the company. The working group generates multiple versions of these headlines and then blends and mashes them up into one statement that incorporates common elements from across the different versions. This ultimately becomes the **2-year winning aspiration** for the effort. ### Step 3: Design sprint (1 week) The design sprint builds on the strategy sprint. It takes the three strategic pillars and the “how might we’s” as its primary input and aims to produce a set of illustrative concepts to bring the strategy document to life. As the saying goes, a picture is worth a thousand words, so these illustrative concepts can paint a better picture for leadership of how the strategic pillars might translate into product concepts and directions. These illustrative concepts are not intended to be “shovel-ready” for engineering to build on. Instead, they’re meant to help the reader understand the strategy document better. The strategy working group design lead heads up the design sprint, while the others take on a participatory role. There are different ways to run the sprint (e.g. some variation of the [Google Ventures Design Sprint](https://www.gv.com/sprint/) could be used). Any style can work as long as there is clear alignment on the inputs (strategic pillars and “how might we’s”) and the outputs (illustrative concepts for each strategic pillar). ### Step 4: Document writing (1-2 weeks) At this point, the product leader has significant leverage built up from steps 1 to 3. Let us recall everything that is now available: user insights (behavioral and user research), competitive analysis, three strategic pillars, associated “how might we’s,” 2-year winning aspiration, and illustrative concepts. These are excellent building blocks for writing the strategy document. The product lead should approach this activity in a relatively solo fashion and combine, connect, and edit all the pieces into a coherent document. The integrity of the process ensures that there are no major surprises for the stakeholders when the document is fully written. Here is a [template](https://docs.google.com/document/d/1iNeYUaMnpicvkpVZO-gj9cCxLeHfWN0xtGm_QoxgemE/edit?usp=sharing) for that document. ### Step 5: Roll-out (2-3 weeks) The final step is to roll out and operationalize the strategy. A few key steps are recommended for this step: 1. **Review the strategy 1:1** **with key gatekeepers:** These are the people you absolutely need support from to make the rollout successful. This could include the founders, CEO, etc. Then iron out any adjustments as a result of these reviews. 2. Once you have gatekeeper alignment, call for a **group review among all key stakeholders**, and drive toward alignment. This could include other functional leaders and partner team leaders. 3. Next, run through **team roadshows**, in groups of 8 to 10 team members each, to encourage more participation and dialogue. During these team roadshow sessions, mainly check for clarity of the document, and adjust phrasing to improve clarity. Do not tinker with the strategic pillars unless some dramatically new signal is unearthed at this point. Defend the strategic pillars through the clear framework for selection that you’ve established earlier. 4. And finally, **empower your teams** to build roadmaps with the strategy document as a key input and reference, and iterate from there. The advantages of this approach are that: 1. Alignment within the team and with leadership is built into the process. Alignment is critical to mobilize any strategy. 2. This approach produces better strategic framing and pillars due to a broader set of inputs in the strategy formulation process. 3. The output is clearly defensible and transparent based on the selection criteria and scoring that is exposed to everybody. This kind of problem-focused 2-year strategy is absolutely necessary to drive action and near-term execution *but* is only half of the story. To complete the picture, let’s rotate to a complementary and more aspirational strategy process called Big “S” strategy. ## Part 2: Crafting the 3/5/10-year strategy (Big “S” strategy) *“Life has to be about more than just solving problems.” —Elon Musk* There needs to be an aspirational and exciting component to a company’s strategy, which is not just focused on solving problems with the current product but maps more closely to the eventual mission and vision of the company. The approach to craft this kind of strategy is called the big “S” strategy. It also takes about five steps, but the process is intentionally a bit more open-ended and greenfield and can take up to six months. Typically a senior design leader at the company heads up this process. ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/2790e523-3aac-4e4e-b635-52883eb10cb3_1457x1455.jpeg) Let’s dive into detailed guidance for each step. ### Step 1: Preparation 1. Form a small working group with a senior design leader, one or two designers, and a UX researcher. 2. Start with the company mission and vision as a grounding anchor. 3. Run some lightweight analysis on long-term cultural, social, competitive, and technological trends that are particularly relevant to the product space. It is OK to use external research or AI tools to quickly gain some of these insights. 4. Conduct interviews with senior leaders and stakeholders to understand their desire for the long-term direction of the company/product. Some questions to consider during these interviews are: 1. What does a day in the life of a user look like in 3/5/10 years? 2. What does the product look like in 3/5/10 years? 3. Why is the world better in 3/5/10 years as a result of our efforts? 4. What is the most exciting version of that view? ### Step 2: Distinct futures 1. Synthesize all of the above information by clustering related ideas into cohesive narratives and fiction pieces. Generate at least three fairly distinct descriptions of a future where the product is significantly different and has a major impact on the world. As a hypothetical example, imagine if we were doing big “S” strategy work for a company plotting the future of transportation. One could imagine three distinct futures. 1. Fully autonomous travel: An experience where people can move from point A to point B through fully automated systems 2. Extreme-speed travel: An experience where cross-world travel can be achieved at a fraction of time and cost relative to what is possible today 3. Virtual travel: An experience where physical travel would be substituted by virtual travel in a large number of use cases due to the maturity of AR/VR technology 2. Once these distinct futures are generated, outline a set of learning goals or key questions associated with each of the distinct futures. These should help to uncover the assumptions and parameters that would make each future widely adopted and commercially viable. ### Step 3: Prototypes Once these distinct futures and the associated key learning goals are verbally articulated, the team should generate prototypes to help with discovery and deeper understanding. Think of these prototypes as concept cars. Concept cars are never commercialized, but they spark inspiration, and there are certain components or technologies that make it into mainstream production down the road. ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/36339312-2672-4cb7-94fa-2c234813274e_1456x839.jpeg) ### Step 4: Convergence UX research should then test these prototypes with a set of carefully selected users to answer the key learning goals previously established. Through this process, a lot of sub-ideas are eliminated, some are modified, and others are combined with each other, and eventually clarity emerges on the **winning components** that are likely to resonate with users even in the present-day context, which accelerates the company toward the desired future. ### Step 5: Roadmap/testing The winning components are then pushed into the active roadmap so that they can be tested live with a broader set of users. And this process could be triggered again depending on a need to explore a potentially new area. ## Combining the two The product roadmap and plan is built through a combination of small “s” (present-forward) and big “S” (future-backward) work, which can run in parallel. This is akin to building a bridge from both sides of a river. ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/262d7eb5-58b3-4612-9959-3ddb3d71ebc6_1457x795.jpeg) Now that we’ve established both parts of Strategy Blocks—small “s” and big “S”—let’s wrap with some practical tips for success. # Overcoming challenges and myths ### **Do we need so much time?** One of the most common challenges with any strategy effort is justifying the time to do it properly. Here are some tools and tips to manage this challenge. 1. The product leader should proactively plan ahead for strategy work, by including both small “s” and big “S” work as part of their onboarding plans in a new role. They should communicate these plans ahead of time to key stakeholders (CEO/founders/the board) and gain alignment. 2. For those who aren’t convinced, it usually helps to break down the steps involved so that stakeholders understand the buildup toward the final output. It further helps to explain that while the small “s” process extends over two to three months, it requires only two to three weeks of fully dedicated time, whereas the rest of the process can be run as part of people’s jobs alongside other projects. 3. Contextualize the relatively small two-to-three-month process against the significant focus, clarity, and momentum that this would unlock over a two-year period. 4. Don’t gate execution work on the availability of strategy. Get started on execution in parallel with “no-regrets” work that can unlock quick wins. 5. And if time is really of the essence and a much quicker turnaround is needed, I’d recommend fast-tracking through the steps rather than skipping any steps: two days for preparation, three days for strategy sprint, two days for design sprint, two days for document writing, and one day for rollout. This gives you a two-week turnaround for the strategy document. Keep this as a last-resort approach. ### **Strategy vs. execution** You might run into vociferous debates on the importance of strategy vs. execution, and tensions can run high on both sides of the argument. I think the “versus” framing is fundamentally flawed, as it pits these two against each other in some kind of eternal conflict for time, resources, and level of organizational importance. Here are some harmonious ways to articulate strategy and execution: 1. Strategy and execution are interconnected and have a circular symbiotic relationship. Execution becomes more refined with strategy, and strategy becomes more refined with execution. They improve each other and make each other stronger. 2. Execution need not be gated by strategy work. I’ve always found that there are some areas where the opportunity is very clear and execution work can start right away, even as strategy work is pursued in parallel. 3. There are some situations where strategy work has low incremental benefit and high opportunity cost. Recognize these situations and consider moving directly into execution (as an exception rather than as a rule). One such situation is when a company decides to “fast-follow” a competitor, where there is a combination of extreme time urgency and a relatively clear direction proven out by the competitor. Even in these situations, it’s worth having a mental model of how to ultimately differentiate in the market. ### **Process vs. content** You might also run into tensions around the importance of process (people who embrace process as the recipe for success) versus content (people with strong opinions on the appropriate direction). I’ve found that people who are on either ends of that spectrum are the least impactful—they’re either too married to a process or too married to specific directions. Instead, the most effective people **can move fluidly between process and content**, wielding each as powerful tools to achieve a certain objective. These people embrace a process as a philosophical approach to bring the team together, rather than as a rigid rule book. They also embrace ideas to spark inspiration and move people in a certain direction. And they are hostage to neither process nor content. Explain this philosophy and use these people as your allies. ### **Managing energy** The strategy working group can feel fatigued through the process, given some of the more intensive team exercises. It’s important to manage the team’s energy by pacing the sprints well, injecting fun team activities to break things up, and building in explicit down cycles during the 8 to 12 weeks. ### **Dead ends, entropy, and hopelessness** While this is a deeply satisfying exercise at the end, expect a healthy dose of difficulties, frustrations, and dead ends along the way. Furthermore, expect a certain level of entropy, where different members of the strategy working group push in different directions based on their individual passions and/or biases. There will be moments when the whole thing feels futile. This is all completely normal. The person driving the effort has to have grit, optimism, and strong synthesis skills to overcome these odds. Keep pushing through and you will eventually see the light of day. ## Results and business impact Ultimately, strategy in itself has no intrinsic business value and is only as good as the results it can produce. Test and iterate the strategy through systematic execution. Assess performance honestly, and pivot away from the parts of the strategy that don’t work and double down on parts that show traction. ## Recap ![Image from Strategy Blocks: An operator’s guide to product strategy](https://substack-post-media.s3.amazonaws.com/public/images/4cfde755-47be-4e3b-8cfa-3916887fdf3f_1486x1621.jpeg) Also refer to the [podcast episode](https://youtu.be/WFLH8Af2f30?si=aJ60o7DGhQOa7J1Y) for additional commentary and support. All the best with your strategy work, and please share what worked and didn’t work through the comments below! *Thank you, [Chandra](https://www.linkedin.com/in/chandramohanj/)!* *Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week [Halo Science](https://www.halo.science/) is hiring for a Head of Product. In this role, you’ll: - Be hands-on executing, while also helping craft strategy for the future of Halo’s product - Work cross-functionally across GTM, design, and engineering—with exposure to the entirety of the business’ moving parts - Report directly to their CEO **Why I think the company and role is interesting:** - I’m an investor! I’m a big fan of what they’re doing. - They are well funded but also very early. This is an opportunity to not only lead product for a company that is disrupting the science R&D space, but also help build the exact team you want. If you’d like to get your profile sent directly to their team, just fill out [this quick form](https://airtable.com/appTNAXC72V74HqPS/pagCczxszIZPdd00t/form). All submissions from my readers get priority (but no guarantees beyond that). *If you’d like to get your role featured here, [apply here](https://airtable.com/apphKJD0wLIHfPcZu/pag09IdQYrMpolMa5/form). If you’re interested in working with my white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), apply to work with us [here](https://www.lennysjobs.com/).* [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [9/46] 1,000,000 *👋 Welcome to a**✨ free-edition ✨**of my weekly newsletter. Each week I tackle reader questions about building product, driving growth, and accelerating your career. Annual subscribers now get a free year of **[Perplexity Pro, Notion Plus, Superhuman, Linear, and Granola](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle).** For more: **[Lennybot](https://www.lennybot.com/) | [Podcast](https://www.lennysnewsletter.com/podcast) | [Courses](https://maven.com/lenny) | [Hiring](https://www.lennysjobs.com/) | [Swag](https://lennyswag.com/).*** This newsletter just hit a big milestone: one *million* subscribers. This is both exciting and absurd. No part of me imagined it would reach these heights. ![Image from 1,000,000](https://substack-post-media.s3.amazonaws.com/public/images/b02b7f87-00d9-48cf-bdbc-e10f3e8f58af_480x264.gif) What started as an “[experiment](https://www.lennysnewsletter.com/p/advice-on-growth-product-and-leadership)” just over five years ago is now the world’s largest product newsletter, podcast, and community. I’m feeling very proud. Milestones like these are a great opportunity to pause and reflect, so I put out a call on [X](https://x.com/lennysan/status/1892279027626365015) and [LinkedIn](https://www.linkedin.com/posts/lennyrachitsky_im-fast-approaching-1-million-newsletter-activity-7298050236277563392-7zei?rcm=ACoAAABGvmoB4S920iEQfSFO_P91nw2wPqfPoic) and asked y’all what burning questions you had for me. Below is my first-ever mailbag edition. If you have any additional questions, post them in the comments, and I’ll do my best to answer them. [Leave a comment](https://www.lennysnewsletter.com/p/1000000/comments) If you’re wondering what might change now that we’ve hit this milestone, the answer is: nothing. You can expect a high-quality, high-value newsletter in your inbox every week, like always. That being said, I’m cooking up some exciting new stuff (a bigger and better [bundle](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle), a new podcast, and a few other surprises). But mostly, it’s back to work. Let me also say this: From the bottom of my heart, thank you. Thank you for subscribing to this newsletter, for sharing it with your friends and colleagues, and for making this work possible. Your support means the world to me. An added thank-you to Jean-Francois Monfette, Rashi Kakkar, David Anderson, Jonah Wolfraim, Patrick Morgan, Mike P. Lewis, Siddhant, Sherry Jiang, David Lobo, Daria Aleksandrova, Anxo, Ender, and Axaia for the questions below, and to everyone else who submitted questions 🙏 On to the mailbag. . . ![Image from 1,000,000](https://substack-post-media.s3.amazonaws.com/public/images/df80d241-d9c0-40ce-8766-d85d3a0a948a_2913x1530.jpeg) ### **What! So many product managers?** I know, right? Seriously, though, there are apparently [over a million product managers](https://www.linkedin.com/search/results/people/?keywords=product+manager&origin=FACETED_SEARCH&sid=xzE&titleFreeText=product+manager) globally. But more importantly, my audience is much wider now. Roughly half are PMs, but over a quarter are founders, and the rest (and the segment growing fastest) are a mix of adjacent functions: engineers, designers, growth/marketing, data, UXR, etc. The newsletter (and podcast) are growing fastest among non-PMs because, with the rise of AI engineers, [as I’ve written](https://www.lennysnewsletter.com/p/why-pms-are-best-positioned-to-thrive), the skills that matter more and more are PM skills: - figuring out what to build - prioritizing the highest-ROI opportunities among endless ideas - getting buy-in - clearly and succinctly articulating requirements - having the taste and experience to know if what you’ve built is good - driving adoption, growth, and retention That’s exactly the kind of stuff I write about. While I initially positioned the newsletter around product management so it was easy for me and others to explain what it was about, I’ve always intentionally kept the focus much wider (“A weekly advice column about building product, driving growth, and accelerating your career”) because I knew I’d be incredibly bored just writing about product management week after week. ### Do you have employees helping you with the newsletter? Why or why not, and how many? I have no full-time employees (that’s been a goal of mine), but I do have an all-star team of freelancers and contractors. I look for stuff I can delegate, and for people who can help me level up the quality of my work. Here’s what the team looks like today: 1. **Newsletter:** I do all of my own writing and editing (10 to 30 hours per post), but I’ve also got an amazing editor (who gives me content and structure feedback), copy editor (who edits for grammar, typos, and clarity), and designer. For the first couple of years it was just me, but now I can’t live without this crew. 2. **Podcast:** I work with two boutique agencies that help me with basically everything outside of finding guests and recording the episode. They handle audio and video editing; titles, thumbnails, and descriptions; coordinating with guests and sponsors; social clips and episode trailers; publishing each episode, and more. I did most of this myself to start, and I still review everything (refining titles and thumbnails, tightening episodes, simplifying top takeaways). I’m also experimenting with a researcher to help me get background on the fancier guests. Btw, check out [this post on my podcast stack](https://www.lennysnewsletter.com/p/my-podcast-tech-stack-workflows-and). 3. **Community:** I’ve got a (new!) community lead and a small team of contractors and volunteers who help me run the 20,000+-member Slack community. The community that’s formed around the newsletter and podcast is the thing I’m proudest of, of all the things I’ve done. It’s very special. 4. **Other:** I also have an EA and an engineer who help across all of the areas above. I suspect the team won’t grow too far beyond this. ### What was the moment you realized, “OMG, this is actually working, and I can do it full-time”? There was a moment nine months in, after publishing something every single week, when I realized that it was still fun for me, people were finding it valuable, and I still had dozens of ideas for things I wanted to write about—after 36 posts. I realized then that maybe this could be a real thing. And that’s also the moment I decided to add the paid plan. (Because adding a paid plan means you need to commit to writing for at least another year, since people are buying annual plans.) After adding the paid plan, people started to pay for it. I found my Ikigai. ![Image from 1,000,000](https://substack-post-media.s3.amazonaws.com/public/images/ef483c6f-5451-4f72-b825-08e460e33de8_1414x1524.png) ### In the beginning, did you feel like you were preaching in the desert? The reason I started writing was to crystallize my own thinking and to not forget what I’d learned. After leaving Airbnb, I was planning to start another company, and I asked myself, “What did I actually learn over the course of my time at Airbnb?” I wasn’t totally sure. I started putting together a bullet-point list of some lessons. I realized those might be useful to other people, so I turned that into [this Medium post](https://marker.medium.com/what-seven-years-at-airbnb-taught-me-about-building-a-company-e1d035d49c56) (a cliché, I know). The post did really well, and that motivated me to keep writing. I wrote a few more things on [Medium](https://medium.com/@lennysan) and got good enough feedback from people I respected to motivate me to keep going. So I kept going. After a handful of posts, I moved over to Substack (because I got advice that collecting people’s email addresses was useful no matter whether I started a company or kept up this writing thing), and that led to adding the paid plan, and now we’re here. As I write out the answer to this question, I see that the key was getting enough positive signals in the early days (while the idea was an “[ugly baby](https://www.forbes.com/sites/andyboynton/2014/03/17/pixar-chief-protect-your-ugly-babies-your-unsightly-ideas/)”) to motivate me to keep going. So there’s a tip: Find the folks who can give you the motivation to keep going. ### What did the growth curve look like in the early days? How did you sustain momentum when you were under 5K? Here you go: ![Image from 1,000,000](https://substack-post-media.s3.amazonaws.com/public/images/b87bb799-d3e6-4fb5-904a-defb7ebaece2_2913x1530.jpeg) Three things that most helped me grow in the beginning: 1. Two guest posts in other newsletters to kick things off (in July): [The First Round Review](https://review.firstround.com/the-power-of-performance-reviews-use-this-system-to-become-a-better-manager/) and [Andrew Chen’s blog](https://andrewchen.com/grow-marketplace-supply/) 2. A popular series of posts that led to a large spike (in late November): [How to kickstart and scale a marketplace business](https://www.lennysnewsletter.com/p/how-to-kickstart-and-scale-a-marketplace) 3. Otherwise, simply publishing something valuable every single week, week after week, for 9+ months ### Have you thought about your off-ramp plans? Someday you may not feel like writing or podcasting. Will you build a team? Shut down? I am always curious about solopreneurs’ long-term plans. This is a very astute question. I have no idea how this story ends. I can obviously just stop if I want to, but then everything I’ve built falls apart. I don’t want that. I’m also not aware of anyone who has done what I’m doing and found a clean off-ramp. This is a real problem and an open question. But I’m very lucky to be able to do this for a living, so I’m not complaining. I’m also experimenting with ideas that may reveal an answer about the long-term plan: 1. Leaning into guest posts (which also end up being [my most popular posts](https://www.lennysnewsletter.com/archive?sort=top)) 2. Incubating new podcasts (with new hosts) 3. Finding ways to delegate more of the work (mixed results so far) 4. Giving myself more ways to take time off (e.g. [PTO policy](https://www.lennysnewsletter.com/p/pto-policy)) If you have ideas here, I’m all ears👂 ### What were the key inflection points in your journey to 1M subscribers? What growth strategies have worked best for you in growing your newsletter, YouTube, and podcast audiences? What growth tactics didn’t work? I’ve experimented with all the growth levers—paid ads, SEO, referrals, BD—and none of them have done a damn thing. The only way this newsletter has grown, with two exceptions, is good old-fashioned word of mouth. That’s actually one of the best parts of an email newsletter—it’s very easy to forward and share. Outside of that, two things did make a dent: 1. [Substack’s recommendation feature](https://on.substack.com/p/recommendations). If you look at the chart at the top of this post, the hockey-stick moment was the day they launched that feature. Today, over 5,000 other newsletters recommend my newsletter. Without it, if you extrapolate from the trajectory beforehand, I’d maybe be at 500,000 subscribers. It’s important to note, though, that the Recommendations feature only works if other writers find your stuff worth recommending. Which is essentially word of mouth. So it still comes back to WOM. 2. [The Greatest Bundle Ever](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle). This didn’t accelerate my free subscriber growth, but it did do a lot of good for my paid subscriber growth. However, this sort of approach works only if you already have a large following. So it wouldn’t work for folks just getting started. And to get here . . . WOM. How do you drive word of mouth? In the words of Arnold Schwarzenegger: [Be useful](https://beusefulbook.com/). As I shared in my post when reaching [500,000 subscribers](https://www.lennysnewsletter.com/p/500000): > There’s no trick to making something like this grow. Growth comes from publishing something valuable, that people want to share with their friends and colleagues, over and over and over. [Here’s some tactical advice](https://www.lennysnewsletter.com/i/136342278/quality-consistency-all-that-matters) on how to actually do this. ### What would you have done differently? Put another way, are there opportunities you missed? Things you wish you’d started or stopped sooner? I honestly can’t think of anything too meaningful. I did do a Maven course at one point that I stopped doing. It went great and people liked it, but it didn’t bring me enough joy (something I try to look for in the things I take on). So I stopped. That ended up creating space to try a podcast, which proved to be much better aligned with my goals. Outside of that, I joke that maybe I should have thought harder about the name of this newsletter, instead of going with the default name suggestion the Substack onboarding flow gave me. ### How do you think about balancing writing what you know will do well vs. writing what you feel is important to you to express? Initially, my heuristic was 80% writing about what I’m energized/curious/excited to write about and 20% writing about what people want me to write about. These days, it’s actually 100% what I’m curious about. I now never publish anything only because I know it’ll do well. This is actually a very important lesson I’ve learned: In this line of work, individual (viral) posts come and go, but it’s all about how long you can stick with it. One viral post: easy. One post every week for five years: much less easy. You need to prioritize stamina over anything else. In other words, [play “infinite games](https://www.amazon.com/Finite-Infinite-Games-James-Carse/dp/1476731713).” ### How do you allocate your time between newsletter and podcast? Which one gives you more pleasure and which one more revenue? If contradictory, how do you reconcile that? I bet this answer will surprise a lot of people. The podcast takes less work *and* generates more revenue than the newsletter. However, the newsletter is the main reason the podcast even had a chance at breaking through. Very few podcasts have the luxury of a massive newsletter promoting their podcast. To be more concrete, each podcast episode takes me 3 to 5 hours, including prep time, recording time, and reviewing draft edits, titles, thumbnails, etc. Each newsletter post takes me 10 to 30 hours. People always ask me how I’m able to put out so much content, but if you add this together (one newsletter post plus one to two podcast episodes a week), that’s basically a 40-hour job. Very doable. ### Who was your favorite guest on the podcast, and why was it [Nan](https://www.youtube.com/watch?v=nTr21kgCFF4)? Haha, that actually *is* one of my favorites and has been the most mentioned over the past month when I run into folks. A few others to check out that you may have missed: 1. [Tobi Lütke](https://youtu.be/tq6vdDJQXvs?si=-FqZvVQ4cvBbC7sY) 2. [Elizabeth Stone](https://youtu.be/2XgU6T4DalY?si=77-CVhVGgkstVLRw) 3. [Drew Houston](https://youtu.be/egdYKLBswgk?si=aO-RS9ibQpstLotk) 4. [Nikita Bier](https://youtu.be/bhnfZhJWCWY?si=RJxCWlrwtqTWTXY3) 5. [Elena Verna](https://youtu.be/IHwS2By9UKM?si=SFU_5bhBJyysIu_f) 6. [Dalton Caldwell](https://youtu.be/m7LvNTbaqSI?si=4pGkqaFNk8enQiS4) 7. [Mihika Kapoor](https://youtu.be/uDq6_CPaRjM?si=DhbJyzU-TdHUHrej) 8. [Jen Abel](https://youtu.be/969dwgu98qc?si=O5Vlsyz2aN1tkneY) 9. [Todd Jackson](https://youtu.be/yc1Uwhfxacs?si=LIXF6SABF3yh2LYU) 10. [Katie Dill](https://youtu.be/gfEEcssu304?si=7273FsNWR1ZahKkO) 11. [Jeff Weinstein](https://youtu.be/qbZQjprTnrU?si=YG2XMnqb_Y4Nz12L) ### How do you keep your content fresh and engaging when scaling to such heights? Guest posts. A few years ago, I realized that there’s no world where I have 52 lessons a year for the rest of time. The solution is to use this platform to help the best product and growth minds in the world share their best lessons. The trick is that most people aren’t necessarily great at writing. To make this work, I spend 10+ hours with each guest author editing, refining, and tightening their post. We go through 5 or 6 iterations before we get to something that hits the bar I set for this newsletter. And it’s working. Most of my most popular posts are now guest posts, and most of the guest posts are far better than my posts. It takes a lot of work, but it pays off. ### What do you think modern product management is going to look like in the age of AI? As I was answering this question, Marty Cagan published a post about the same topic, and I 100% agree with his take. [Go read it](https://www.svpg.com/a-vision-for-product-teams/). The gist, to answer your question, is that I expect product teams will shrink by 25% to 50% (mostly, fewer engineers); PMs will have even *more* influence and leverage, and oversee more scope, but they’ll spend more time in discovery and GTM, and less time designing and building. It’s actually a really good time to be a PM. Companies will continue to need product people who can help them figure out what to build, work with AI and humans to build it, make sure it’s great and correct, and then drive adoption. But the bar will continue to rise. If you want to stay on top of what’s going on, check out these posts: 1. [A guide to AI prototyping for product managers](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product) 2. [Product manager is an unfair role. So work unfairly.](https://www.lennysnewsletter.com/p/product-manager-is-an-unfair-role) 3. [How to become a supermanager with AI](https://www.lennysnewsletter.com/p/how-to-become-a-supermanager-with) 4. [Five proven prompt engineering techniques (and a few more-advanced tactics)](https://www.lennysnewsletter.com/p/five-proven-prompt-engineering-techniques) 5. [How to use ChatGPT in your PM work](https://www.lennysnewsletter.com/p/how-to-use-chatgpt-in-your-pm-work) (and [Perplexity](https://www.lennysnewsletter.com/p/how-to-use-perplexity-in-your-pm)) ### Any dream podcast guests you haven’t yet had on the show? Yes. Here’s my current list: 1. [Gwynne Shotwell](https://en.wikipedia.org/wiki/Gwynne_Shotwell) 2. [Levelsio](https://x.com/levelsio) 3. [Mira Murati](https://x.com/miramurati) 4. [Jensen Huang](https://www.linkedin.com/in/jenhsunhuang/) 5. [Jony Ive](https://en.wikipedia.org/wiki/Jony_Ive) 6. [Andrej Karpathy](https://karpathy.ai/) 7. [Sheryl Sandberg](https://en.wikipedia.org/wiki/Sheryl_Sandberg) 8. [Tony Fadell](https://en.wikipedia.org/wiki/Tony_Fadell) 9. [Paul Graham](https://www.paulgraham.com/) 10. Elon 😬 Anyone else you’d love to see come on the podcast? Leave a comment and let me know 🙏 ### What three pieces of advice would you give yourself to reach this point even faster? Why try to reach this point faster? What’s the rush? Play infinite games. Enjoy the ride. You’ll do great. *Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week [Halo Science](https://www.halo.science/) is hiring for a Head of Product. In this role, you’ll: - Be hands-on executing, while also helping craft strategy for the future of Halo’s product - Work cross-functionally across GTM, design, and engineering—with exposure to the entirety of the business’s moving parts - Report directly to their CEO **Why I think the company and role are interesting:** - I’m an investor! I’m a big fan of what they’re doing. - They are well funded but also very early. This is an opportunity to not only lead product for a company that is disrupting the science R&D space but also help build the exact team you want. If you’d like to get your profile sent directly to their team, just fill out [this quick form](https://airtable.com/appTNAXC72V74HqPS/pagCczxszIZPdd00t/form). All submissions from my readers get priority (but no guarantees beyond that). *If you’d like to get your role featured here, [apply here](https://airtable.com/apphKJD0wLIHfPcZu/pag09IdQYrMpolMa5/form). If you’re interested in working with my white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), apply to work with us [here](https://www.lennysjobs.com/).* [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [10/46] Why you’re so angry at work (and what to do about it) Recently, executive coach [Natalie Rothfels](https://www.linkedin.com/in/nrothfels) has noticed a concerning pattern across all of her co-founder and executive clients: everyone’s angry. Very angry. Below, she shares a guide to identifying and then transforming anger into wisdom that she’s been putting into practice with her clients. It honestly made me well up the first time I read it. This post may completely change your relationship with work. Before becoming a coach, Natalie spent a decade as a product manager at [Quizlet](https://quizlet.com/gb) and [Khan Academy](https://www.khanacademy.org/). She is a certified Internal Family Systems practitioner, a facilitator for Stanford Graduate School of Business’s Interpersonal Dynamics course, aka “Touchy Feely,” and the author of two popular guest posts: [Why no productivity hack will solve your overwhelm](https://www.lennysnewsletter.com/p/why-no-productivity-hack-will-solve) and [On asking for help (even when you really don’t want to](https://www.lennysnewsletter.com/p/on-asking-for-help-even-when-you). For more from Natalie, follow her on [LinkedIn](https://www.linkedin.com/in/nrothfels) and [X](https://x.com/natatouille). ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/e9655a16-b9ee-4394-8ea4-7fb7fd056831_2912x2322.jpeg) In the past six months, every one of my coaching clients has been vibrating with anger. Not mild frustration—deep, persistent anger that was depleting their motivation and sense of ownership and making them emotionally exhausted at work. Yet few clients could name or work with that anger effectively. We have tons of sophisticated frameworks for building product strategy, giving feedback, making decisions, and leveraging AI in the workplace. But when it comes to navigating anger, we’re mostly winging it. The result? An army of professionals walking around with forced smiles, internally fuming, unconsciously stabbing each other with their emotional swords in the hallways or on Zoom. We’re all feeling angry, no one’s talking about it, and it’s burning us out. At work, anger most commonly arises when your expertise is challenged, your autonomy is threatened, your values are compromised, or your identity is dismissed. I have been taken over by anger for all of these reasons. But I’ve learned that anger isn’t something to suppress or feel ashamed of. It’ssomething that we need to be more skillful at owning and engaging with because, while it contains immense wisdom, it can cause harm to ourselves and others if it erupts unconsciously or goes unchecked. Here’s the process I now use with my clients to help them identify and then transform their own anger into wisdom that they can then appropriately express in a work context. Informed by Buddhist philosophy, [Internal Family Systems](https://www.lennysnewsletter.com/p/why-no-productivity-hack-will-solve), and other somatic methodologies, it’s the resource I wish I had earlier in my career for navigating these high-intensity emotions at work. ## Anger: an inherently useful emotion Before we get into the process of transforming anger, let’s briefly touch on what anger is and what it does for us. Anger is part of being a human. We all have it. And it’s an inherently intelligent and useful emotion. Think of anger as a smoke alarm in your home. When the alarm goes off, the *alarm* isn’t the problem—it’s doing its job by telling us there might be a fire. If we smash the alarm or ignore it, we risk damage. If we panic and do nothing, the fire spreads. How we respond to the alarm once we hear it determines the outcome of a potentially dangerous situation. Anger indicates that something we need or care about is under threat—and most often the “fire” is **an unmet emotional need**. ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/a4b11080-a5a0-4541-a0ee-443c494199ee_2048x1634.png) When the anger “alarm” goes off, it triggers different unconscious reactions: - **Exploding** - Getting bigger: aggression, control, over-functioning - Pushing outward: blame, criticism, punishment of others - **Withdrawing** - Getting smaller: shutdown, under-functioning - Turning inward: self-criticism, perfectionism ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/62b829e0-e6e3-4f43-b529-6e3f7f080b81_1996x830.png) If we let unconscious reactions take over again and again at work, there can be real costs. - Exploding leads to conflict escalation, damaged relationships, and lack of trust. - Withdrawing leads to simmering resentment, passive-aggressive behavior, and burnout. These reactions happen *fast*. Here’s an example from my own professional life to illustrate how quickly the takeover happens. Several years ago, I was facilitating an important meeting with eight cross-functional leaders to make a decision about the launch of a new product line. Differing opinions shot across the room, and I started to notice a pattern as the hour went by. Each time an idea was offered by one of the quieter voices in the room, it was quickly bypassed, shut down, or even criticized by one of the more vocal members of the team. I noticed myself starting to feel angry after the third, fourth, then fifth instance. I didn’t want to take sides, but the dynamic felt so obvious and unfair to me that my body tensed up. My heart was racing, my hands were clammy, and I was on the verge of saying something I would regret. And then I did. Raising my voice significantly higher than usual, I firmly stated that I was feeling angry and that I really wanted everyone in the room to stop shutting down the quieter voices. I don’t know how I came across, and I wasn’t even aware of what I was saying. Inside, I was raging. An unconscious reaction had taken over. The counter-reaction was instant and intense—where was my comment even coming from?! One person raised their voice back at me, another claimed I was judging them unfairly, and a third said I was derailing the conversation for my own personal needs. So, naturally, I started to feel even *more* outraged. Before I knew it, the meeting was more about *my* anger than the team’s goal of making a launch decision. By the time the hour was up, I was nearly in tears, and we were nowhere close to making a call. Oops. *Feeling* angry wasn’t the issue. It was my *response* to the anger itself that had consequences. I had unconsciously gone into an aggressive reactivity—exploding and sending others on the defense. As a result, the group’s attention was taken away from the real decision and toward managing a heated interpersonal conflict. It was a painful but pivotal moment to learn about my own boundaries, triggers, default responses, and needs. And it gave me the key to understanding my options for responding to my own anger. While my anger manifested as exploding in that meeting, we all respond to workplace threats very differently based on our personal history, the situation, and systemic issues. Later, I’ll walk through an example from my client Etienne (who had a different response than mine) so you can get a sense of the range of possibilities. ## Anger is the alarm, unmet needs are the fire If anger is your body’s way of saying “Something important is under threat—pay attention,” it pays to investigate what*’*s on fire inside you. We may think of our time at work as a place of pure logic and practicality, but as humans, we bring emotional needs to every situation and every setting. And when those needs aren’t met—even in the workplace—they can trigger an emotional response in the form of anger. Here are some common and fundamental needs that can set off our alarms when they go unmet: - **Connection:** Feeling included and heard by colleagues - **Autonomy:** Having the freedom to choose how you spend your time and structure your work - **Agency:** Having control and input over your work decisions and direction - **Clarity:** Getting transparency around decisions and processes - **Security:** Having stability and predictability in your work environment - **Belonging:** Feeling genuinely part of the team or community, accepted, and valued for who you are - **Dignity:** Feeling your inherent worth is recognized and respected - **Respect:** Having your contributions and identity acknowledged and valued - **Competence:** Seeing your efforts make a real impact and feeling capable in your role - **Growth:** Having opportunities to deepen your skills and advance professionally - **Fairness:** Believing decisions, promotions, and resource allocations are equitable - **Collaboration:** Exchanging support and guidance with those around you ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/a822f82d-84a9-47e5-b863-f0475080d7c9_2048x1634.png) When a fundamental emotional need goes unmet, anger often pops up—and sometimes so powerfully that it overwhelms any other feeling or logical analysis. That’s what happened for me in my earlier example, when I exploded with anger instead of articulating the group’s (and my own) unmet needs for fairness, security, and respect. I see this dynamic play out across different product teams and levels all the time. Recently, a product leader had her carefully crafted roadmap completely overturned by an executive after they used a competitor’s product and decided to pivot the company’s strategy as a result. She was furious, but instead of naming her need for respect and acknowledgment, she spent two weeks preparing a detailed competitive analysis to prove herself “right,” stalling her team’s work in the meantime. The unmet need festered, and so did the anger. ## Four steps to translate anger to wisdom Below are the four steps I use with my coaching clients (and myself) to transform anger from a destructive force into an opportunity for deeper wisdom and more effective action: 1. Recognize your alarm 2. Do a U-turn (look inward first) 3. Identify the unmet need 4. Choose a conscious response ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/ad88fa84-e5de-4863-8dc0-0a7e99839e7d_2048x1634.png) To illustrate how this process unfolds in real time, let’s look at Etienne, one of my clients who found himself battling anger when a new coworker seemed to threaten his sense of being valued on the team. ### 1. Recognize your alarm Many of us are so used to labeling anger as “bad” or unprofessional that we avoid even acknowledging we’re experiencing it. The first step is noticing we’re in anger territory. I have found that paying attention to physical sensations is the best starting point. If you’re new to recognizing anger in your body, the Buddhist paradigm of **hot and cold anger** can be very helpful. **Hot anger** feels like urgency, agitation, a pressing need to act or speak now. It is characterized by: - Physical sensations: constricting or tight chest, warming sensations in your chest or neck, rapid breathing, clenched fists, excessive sweating, or your heart pounding in your ears - Expression: exploding**,** going outward - Protecting: pain, feeling hurt, betrayed, or wronged **Cold anger** surfaces as withdrawal, shutdown, or quiet contempt. It is characterized by: - Physical sensations: goosebumps, a staggered or held breath, slumped shoulders, or a frozen feeling in your extremities - Expression: withdrawing or going inward - Protecting: fear of conflict, loss, being too much, or being unsafe If you run hot, you might need to slow down to access the wisdom. If you run cold, you might need to make yourself bigger in order to hear yourself more fully. What does that mean? Imagine your anger alarm as the main character in a play. Bring it to life by first creating a shape with your body that represents this character’s personality. Then have the character verbally or physically express all the things it may never get to express in “real life.” I’ve had clients spontaneously start beating their chests, throwing their arms around emulating a tantrum, or fully fold in on themselves like a crumpled piece of paper. ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/aa9c18fd-994a-441e-9286-f6a499402c27_1812x1083.png) A skilled product leader known for his behind-the-scenes effectiveness, Etienne came to me struggling to navigate the working style of his new peer Corey. While Etienne led through quiet competence, Corey was more vocal and politically savvy. Corey knew how to advocate for himself, while Etienne knew how to advocate for his team. Etienne found Corey’s style to be self-serving and threatening, and he was concerned about losing opportunities for advancement for fear of not knowing how to self-advocate like Corey. When Etienne’s boss, Manuela, suggested that Corey join in to lead one of his projects, Etienne’s anger alarm went off immediately. In our session, I noticed his tight smile (a common deflection when we’re uncomfortable with anger) as he calmly outlined all the operational reasons why Corey shouldn’t join. When I asked Etienne to check in with his body, he noticed constriction, physical heat building, and a feeling of anxiety. And this was just in reporting the events back to me! Then he would smile and say, “The situation honestly is not that bad.” As we tracked his internal state, his pattern became clear: oscillation between passive-aggressive outburst and complete withdrawal, a mix of hot and cold anger responses. **👉 TRY THIS IN THE MOMENT:** When you feel anger rising in real time: - Take three deep breaths (long exhales) - Name your physical sensations (“My heart is pounding, my shoulders are tense.”) - Identify one immediate trigger (“They dismissed my idea. I notice I’m getting defensive about my team’s work.”) - Write down what you’re feeling rather than saying it **👉 TRY THIS FOR DEEPER REFLECTION:** When you’re in a more grounded state: - Think about the past couple of times you felt angry at work (roadmap presentation to execs? scope negotiation with engineering?). Do you tend to respond to your alarm with hot anger or cold anger? - Keep a simple anger log for a week (see example below) - What patterns trigger your anger (e.g. scope changes, dismissed research, bypassed processes, boss interrupting you)? - How does your body signal anger in these situations? Pay attention to your physiology (e.g. clenched fists, tight chest, internal screaming voice). - What’s your default response pattern in different contexts (e.g. outward expression, inward expression, gossiping, crying, biting your tongue)? Patterns will emerge. ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/2f4fee3c-71c8-4fec-bfbf-b3c2d104fbb8_2048x1634.png) This self-awareness work is difficult and an ongoing practice. I’ve spent decades dissociating from my internal experience and mostly operating from the neck up at work because most workplaces reward that! Maybe you have too. Given that, don’t underestimate the power of this step: recognizing and naming your own anger alarm may be a huge first step toward building a different response. It will get easier over time. ### 2. Do a U-turn (look inward first) When anger spikes, some of us often look outward: *They’re the problem!* And if we have a tendency to withdraw or turn inward with anger, we may be quick to dismiss our own suffering and instead harshly direct it toward ourselves: *I’m an idiot!* This is exactly what happened with Etienne. His first instinct was to focus entirely outward, preparing arguments about operational overhead and how much the team would be disrupted if Corey joined. He was building a case against Corey rather than understanding his own reaction first. The second step is to pivot inward with compassion. **What am I actually feeling and why?** We do this not to let others off the hook but because we’re the only ones who can truly attend to our own reaction. When I invited Etienne to turn inward, he could start to see that his reaction wasn’t just about Corey or the project structure. There was something deeper being triggered about his own way of working and his sense of being valued by Manuela, his manager. During our session, I noticed how Etienne’s energy shifted as he moved from listing all the reasons Corey shouldn’t join the project to actually feeling what was happening inside his body. His shoulders dropped, his voice got quieter, and he admitted, “I guess I’m concerned that if Corey joins, my contributions won’t be as obvious anymore.” This moment of vulnerability opened the door to understanding what his anger was really trying to protect. Doing a U-turn is vulnerable and counterintuitive, especially if you believe you have been wronged. But we cannot see what’s happening beneath the anger—what our alarm is truly signaling us about—if we stay outside of ourselves. **👉 TRY THIS IN THE MOMENT:** When you feel triggered in a meeting or conversation: - Use [Tara Brach’s RAIN](https://www.tarabrach.com/rain/) system, a super-powerful tool for doing the U-turn. - Recognize what’s happening (“I am getting defensive.”) - Allow it to be as it is (“It’s OK to feel this way.”) - Investigate with interest (“What’s underneath this anger? What hurts right now?”) - Nurture with self-compassion (“This is hard, and I’m doing my best.”) - Focus on your breath and physical sensations. You don’t need to fix or change them. If possible, ask for a quick break to collect your thoughts. **👉 TRY THIS FOR DEEPER REFLECTION:** When you have time to process: - Pick three emotions using the [Feelings Wheel](https://feelingswheel.com). - Journal for three minutes each, writing freely without editing - “I’m [emotion 1] because…” - “I’m [emotion 2] because…” - “I’m [emotion 3] because…” - Get curious. What need might these emotions be pointing to? The goal here is to make the intentional move to turn toward yourself and get curious about your experience. This step, much like the first, takes time and practice. The journal prompts are there to help you learn the mechanism for going inside, but once you get more practiced you’ll be able to identify what’s going on internally with more ease. ### 3. Identify the unmet need Once you pivot inward, you can see which of your core needs has been violated or threatened. This is the crucial step where we help anger transform from a blaring “DANGER!” alarm to a clear mirror reflecting your innermost priorities. That mirror is what helps turn anger into functional insights for the future. Staying with your internal experience, ask yourself: **What do I need that is getting expressed through my anger?** Revisit the needs list below or check out the resource section at the end for a deeper [list of needs](https://www.sociocracyforall.org/nvc-feelings-and-needs-list/). ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/44af1328-cdcc-4b41-a0f0-822f66d11f02_2048x1634.png) For Etienne, having a sense of autonomy on the project pointed to a deeper need. When I asked him, “What does having autonomy give you?,” we found layers: a need for dignity and respect, and a deep need for being seen and valued for his contributions. Working alone had been his way of ensuring that his contributions were clear, visible, and easily attributable to his own unique efforts. The thought of Corey joining threatened this strategy he’d developed for feeling valued at work. ![Image from Why you’re so angry at work (and what to do about it)](https://substack-post-media.s3.amazonaws.com/public/images/0f2aa9fd-54e0-43c6-9213-6158ca3985a7_2048x1634.png) **👉 TRY THIS IN THE MOMENT:** When you feel anger rising, ask yourself: - What am I wishing were different right now? - What feels most important to me in this moment? - What do I need to feel better about this situation? **👉 TRY THIS FOR DEEPER REFLECTION:** When you have space to process: - Review the list of common needs. Which feel threatened? - Imagine the perfect scenario. Ask: “If this were perfect, what would I be getting that I’m not getting now?” - Dig deeper: “What would having that give me?” Continue with that question until you find something that feels tender or vulnerable. ### 4. Choose a conscious response With new clarity about your own unmet needs and values, you can decide how to respond rather than letting your anger drive you. That may mean speaking up, setting a boundary, or simply taking a break to process on your own. Etienne’s clarity about his core need—to feel valued—opened up new possibilities for action. Instead of arguing against Corey joining the project (a reaction driven by anger), he decided to have a different conversation with Manuela about how he experienced recognition and support at work. This wasn’t about preventing Corey from joining anymore; it was about ensuring that his own needs for visibility and acknowledgment could still be met. Not every work environment allows for the vulnerability of speaking from needs. Still, I love [Tara Brach’s advice](https://insighttimer.com/tarabrach/guided-meditations/anger-responding-not-reacting) on this: “Whenever it has a chance of increasing understanding and connection, speak out.” I’d add to that “without the energy of anger.” **👉 TRY THIS IN THE MOMENT:** - Pause and breathe. Ask: “Will this response build or break trust?” - Try: “I’m noticing myself feeling [emotion], and I think it’s because I need [x] (or my team has done [y]). Could we discuss this so we can find a solution that addresses both of our needs?” - If you’re leading, pause and ask: “What’s important to each of you here?” This can reset the room and bring everyone’s needs to the forefront. **👉 TRY THIS FOR DEEPER REFLECTION:** - Journal in two three-minute sessions: - Prompt 1: “From my anger, I want to say...” - Prompt 2: “What I really need is...” - Draft two versions of your response: one from anger, one from your identified need - *Example from anger: “When my roadmap gets changed without consultation, I want to say…”* - *Example from identified need: “What I really need in stakeholder relationships is a sense of collaboration and mutual respect.”* - Consider sharing: “I’ve been reflecting on [situation], and I realized I need [x] to do my best work. Could we discuss how to make that possible?” These exercises help you notice and express your underlying need rather than letting anger hijack the conversation. Etienne ultimately chose to have a conversation with Manuela that focused on his need for recognition rather than his resistance to Corey. This led to a deeper discussion about what would help him feel valued on the team, including ways his contributions could remain visible even in collaborative projects. Just like Etienne discovered, the heart of turning anger into wisdom is about pausing to recognize what deeper needs you may have, and then addressing those through a broader or more personal conversation. ## Conclusion Workplace anger doesn’t always announce itself as a fiery outburst. Whenever you feel a trickle or rush of frustration, there’s an opportunity to get ahead of the anger before it gets out of control. Remember: The goal is not to avoid ever feeling anger (internally or from others). The hope is to transform anger’s energy from something that drives unconscious reactions (like quiet bristling) into a source of clarity about what really matters. When you feel the alarm ringing, pause. Shift from **alarm** to **mirror** by looking inward with compassion. Let that mirror reveal which core need or value is being threatened, then choose a response that builds rather than breaks down trust. Start small. This week, look for one moment of frustration in a meeting and run it through these four steps. See what you notice. Then reflect on what might change if your whole team had access to this approach. ### 📚 Further study 1. *[Anger: Wisdom for Cooling the Flames](https://terebess.hu/zen/mesterek/Thich%20Nhat%20Hanh%20-%20Anger%20-%20Wisdom%20for%20Cooling%20the%20Flames.pdf)* by Thich Nhat Hanh 2. [Taking Care of Anger](https://www.youtube.com/watch?v=9OvLOna5_1A) by Thich Nhat Hanh 3. *[The Dance of Anger](https://www.amazon.com/Dance-Anger-Changing-Patterns-Relationships/dp/0062319043)* by Harriet Lerner 4. [RAIN practices](https://www.tarabrach.com/rain/) by Tara Brach 5. *[When the Body Says No](https://www.amazon.com/When-Body-Says-Understanding-Stress-Disease/dp/0470923350/)* by Gabor Maté 6. [NVC needs list](https://www.sociocracyforall.org/nvc-feelings-and-needs-list/) (to print for reference) 7. [When Getting Angry Makes You Happy](https://greatergood.berkeley.edu/article/item/when_getting_angry_makes_you_happy) by Lauren Klein *Thank you, Natalie! For more from Natalie, follow her on [LinkedIn](https://www.linkedin.com/in/nrothfels) and [X](https://x.com/natatouille).* *Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week **[Bounce](https://bounce.com/)** is hiring for a Head of Growth based out of San Francisco. As their Head of Growth, you'll be expected to: 1. Lead strategy and execution across multiple channels, including product-led growth, SEO, and paid acquisition 2. Mix high-level thinking with hands-on execution as you build out Bounce’s growth team **Why I think the company is interesting:** - Bounce is one of the fastest-growing European startups ever. - They’re now in hypergrowth with a lean and scrappy team. This is an opportunity to join before they really mature. - They are backed by a16z and a host of tier-one investors. They just raised a Series B and are quickly expanding into a global presence. - This is not a 0-to-1 role; it’s a 1-to-10 role. Bounce’s growth is already epic, but they need a leader to take it to the next level. If you’d like to get your profile sent directly to their team, just fill out [this quick form](https://airtable.com/appTNAXC72V74HqPS/pagCczxszIZPdd00t/form). All submissions from my readers get priority (but no guarantees beyond that). *If you’d like to get your role featured here, [apply here](https://airtable.com/apphKJD0wLIHfPcZu/pag09IdQYrMpolMa5/form). If you’re interested in working with my white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), apply to work with us [here](https://www.lennysjobs.com/).* [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [11/46] Which companies produce the best product managers *For more: **[Lennybot](https://www.lennybot.com/) | [Podcast](https://www.lennysnewsletter.com/podcast) | [Courses](https://maven.com/lenny) | [Hiring](https://www.lennysjobs.com/) | [Swag](https://lennyswag.com/)*** My first-ever deep dive a few months ago into which [companies accelerate PM careers most](https://www.lennysnewsletter.com/p/which-companies-accelerate-your-pm) led to a lot of great feedback and ideas, so I’m back with a follow-up, going one level deeper into the question: Which companies produce the best product managers? We arrived at the answer by triangulating a bunch of juicy data. #### **In this post, we**’**ll look at:** 1. Which companies create the **most successful PM alumni founders** 2. Which companies **accelerate PM careers most**—with four new data points since the last deep dive 3. Why Stripe PMs didn’t score higher—this one is fascinating ### Across the board, here are the top companies that seem to produce the best PMs: ![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/339c7fbf-6760-4478-8ed6-294565235016_2913x2374.jpeg) The big change we made in this updated edition was zeroing in on just the past 10 years vs. the company’s entire history. We also went deeper into the founder data, looking at not just the raw number of founders but also how they’ve done. And we added a few companies to the list (e.g. Airbnb, DoorDash, Salesforce, Dropbox) that people pointed out we missed. A big thank-you to [Live Data Technologies](https://www.livedatatechnologies.com/product-jobs) ([Jason Saltzman](https://www.linkedin.com/in/jason-salt/) and [Ethan Elias](https://www.linkedin.com/in/ethan-elias-70021472/)) for sharing this data with us and for helping with the analysis. *Caveats:* 1. *We are looking only at companies that have enough data (i.e. enough PM alumni) to draw meaningful conclusions from. Your startup may be producing the best PMs ever—we just won’t know this yet.* 2. *A confounding variable is the quality of PMs these companies hire. If a company hires only amazing senior PMs, they will naturally go on to do amazing things—and that doesn’t mean the company made these PMs great. However, if the goal of this post is to help you figure out where to work, getting a gig at one of the companies on this list will put you on a good track either way.* ## 1. Which companies create the most founders—and the most *successful* founders? First, we looked into which companies’ PM alumni go on to start the most companies, irrespective of how these companies do. **Palantir** dominates. Over a third of their PMs have started a company. 🤯 **Intercom** comes in second, at over 18% (almost one in every five alumni PMs!), followed by **Dropbox**, **Plaid**, **N26**, **Revolut**, **Duolingo**, **Uber**, **LinkedIn**, and **Coinbase**. I’ll also give props to **Ramp**, **Airbnb**, and **Notion**, which came in just below the top 10 but all have 10% of alumni PMs who start a company. That’s impressive as hell. Go work at one of these companies if your goal is to start your own company: **Palantir**, **Intercom**, **Dropbox**, **Plaid**, **N26**, **Revolut**, **Duolingo**, **Uber**, **LinkedIn**, **Coinbase**, **Ramp**, **Airbnb**, or **Notion.** But keep reading. . . ![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/67eb925f-7188-4b0a-89ba-6c11e931dfa3_1457x2441.jpeg) Next, we looked more closely at which companies create the most successful founders. It’s one thing to start a company—it’s another to start something that works. Based on the data we have available to us, the best proxy we found for measuring “successful” was to look at which companies have gone on to raise a Series A. Yes, some founders choose to stay bootstrapped and never raise money, and raising a Series A doesn’t mean you’re successful, but it’s a simple way to zero in on founders aiming to build large venture-scale companies. The headline is **Chime**. Wow. Over 20% (one in five) of Chime’s PM alumni go on to not just start a company but also raise a Series A. **Scale**, **Palantir**, and **Faire** aren’t too far behind, at around 15%. And then we have **Dropbox**, **Robinhood**, **Stripe**, **Block/Square**, **Coinbase**, and **Salesforce** rounding out the top 10. Go work at one of these places if you want to build a venture-scale company: **Chime**, **Scale**, **Palantir**, **Faire**, **Dropbox**, **Robinhood**, **Stripe**, **Block/Square**, **Coinbase**, **Salesforce**. ![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/7e857c7d-104f-4be7-b48a-7fe7b02ee916_1457x2159.jpeg) ## Which companies accelerate PM careers most? Building on our analysis in [the previous edition](https://www.lennysnewsletter.com/p/which-companies-accelerate-your-pm) and adding four new data points, below we’ve triangulated which companies’ alumni PMs create the biggest inflection in the career of their PMs by looking at: 1. Rate of PMs getting promoted within the company 2. Rate of alumni PMs getting promoted after leaving 3. The average time to promotion after leaving 4. The average time to reach a leadership position (e.g. VP of Product, Head of, CPO) #### Takeaways: 1. **Intercom** dominates. It’s the only company to rank in the top 10 on all four dimensions (1st in promotions internally, 5th in promotions externally, 7th in fastest to promotion, and 9th in fastest rise to leadership). 👏 2. **Revolut** and **Nubank** are tied for second place, both ranking in the top 10 in three different categories. Fintech? More like Win-tech. 🤣 3. **N26** comes in fourth, ranking in the top 10 in two categories: rate of promotion externally and fastest rise to leadership. Nice. 4. A special mention to **Palantir**, **Deel**, **HubSpot**, **Discord**, **Block/Square**, **Faire**, **Chime**,and **Cruise** for ranking highly in at least one of the categories. Go work at one of these companies if you want the best education on the skill of product management: **Intercom**, **Revolut**, **Nubank**, **N26**, **Palantir**, **Deel**, **HubSpot**, **Discord**, **Block/Square**, **Faire**, **Chime**, **Cruise**. Here’s the data that informed the above: ![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/5dcfca3f-e2e5-4b78-ae28-a365f5dc2d6c_1457x2486.jpeg)![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/dc8e81e0-0c43-4280-912d-6211f524dddd_1457x2192.jpeg)![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/263a6663-5eff-4890-9add-a57c28c2988f_1457x2527.jpeg)![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/46673f66-0220-4f17-91d4-db10f8f2529d_1457x2277.jpeg) ## Which companies create the most product leaders? Next, we looked at which companies’ alumni PMs go on to become product leaders most often (e.g. CPO, Head of Product, first PM hire at a startup). #### Takeaways: 1. **eBay**, **Intercom**, **N26**, **Palantir**,and **Notion** stand out as companies that produce the highest rate of product leaders across the board. Whatever they are doing we should try to learn from. 2. **Revolut** and **Intuit** are just a bit behind, with solid rankings but not quite as strong as the companies above. 3. **PayPal** and **Dropbox** get a special mention as the remaining companies ranked in at least two categories, though they ranked fairly low. Go work at one of these companies if you want to one day lead product at a company: **eBay**, **Intercom**, **N26**, **Palantir**, **Notion**, **Revolut**, **Intuit**, **PayPal**,or **Dropbox**. ![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/4467cefd-c5d3-4a14-9edf-e4e301e8f7c2_1457x2023.jpeg)![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/e1c29d38-2fad-4be2-bf8b-46b08ad94a5b_1457x2230.jpeg)![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/0e374915-3514-430a-bd53-73bfc2cbb365_1457x2274.jpeg) ## Why didn’t Stripe PMs do better? In this analysis and the previous edition, I was stumped as to why Stripe didn’t perform better. I’ve been nothing but impressed with every PM I’ve ever met from Stripe, and I’ve had more Stripe and ex-Stripe people on my podcast than any other company. WTF. Well, the answer is fascinating. It turns out Stripe’s PMs get hired to be star PMs at rocketship companies. Instead of starting their own companies or climbing the ladder, they go on to become key PMs at top companies like OpenAI, Anthropic, Mercury, Adyen, Scale, etc. Here’s a breakdown of where Stripe PM alumni go work: ![Image from Which companies produce the best product managers](https://substack-post-media.s3.amazonaws.com/public/images/47a8fb65-880c-4d41-ba50-c5cb196415f3_1456x1917.jpeg) It’s interesting how many Stripe PM alumni go to earlier-stage companies and not to big tech companies (e.g. Google, Microsoft, Amazon, Meta). That may also explain why they see less career advancement on paper. This opens up some interesting opportunities for future explorations! I’m excited to hear what you think of this data, and what additional analysis you’d find valuable to guide your career if I were to continue this series. Leave a comment with suggestions and questions👇 [Leave a comment](https://www.lennysnewsletter.com/p/which-companies-produce-the-best/comments) *Thank you, [Jason Saltzman](https://www.linkedin.com/in/jason-salt/) and [Ethan Elias](https://www.linkedin.com/in/ethan-elias-70021472/) from [Live Data Technologies](https://www.livedatatechnologies.com/product-jobs), for helping with this analysis. Have a fulfilling and productive week 🙏* ## 🏅 Featured role of the week **[Bounce](https://bounce.com/)** is hiring for a Head of Growth based out of San Francisco. As their Head of Growth, you’ll be expected to: 1. Lead strategy and execution across multiple channels, including product-led growth, SEO, and paid acquisition 2. Mix high-level thinking with hands-on execution as you build out Bounce’s growth team **Why I think the company is interesting:** - Bounce is one of the fastest-growing European startups ever. - They’re now in hypergrowth with a lean and scrappy team. This is an opportunity to join before they really mature. - They are backed by a16z and a host of tier-one investors. They just raised a Series B and are quickly expanding into a global presence. - This is not a 0-to-1 role; it’s a 1-to-10 role. Bounce’s growth is already epic, but they need a leader to take it to the next level. If you’d like to get your profile sent directly to their team, just fill out [this quick form](https://airtable.com/appTNAXC72V74HqPS/pagCczxszIZPdd00t/form). All submissions from my readers get priority (but no guarantees beyond that). *If you’d like to get your role featured here, [apply here](https://airtable.com/apphKJD0wLIHfPcZu/pag09IdQYrMpolMa5/form). If you’re interested in working with my white-glove recruiting service specializing in senior product roles (e.g. Directors, VPs, and Heads of Product), apply to work with us [here](https://www.lennysjobs.com/).* [Start hiring](https://www.lennysjobs.com/) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [12/46] How to ship like a startup [Mihika Kapoor](https://x.com/mihikapoor) is one of my all-time favorite podcast guests. Her passion, hustle, and ability to get sh\*t done have inspired so many listeners and readers to think bigger and bring more joy to their work. In fact, [her podcast episode](https://www.youtube.com/watch?v=uDq6_CPaRjM&t=583s) is the most popular episode I’ve ever done with a non-founder or non-exec. Mihika was one of the early leads on FigJam, Figma’s first-ever new product launch, and most recently, she pitched and launched Figma’s latest product, Figma Slides, which [in my recent survey of readers](https://www.lennysnewsletter.com/i/155187022/figma-slides-and-canva-have-become-big-players-in-presentations) is already ahead of Apple Keynote and tied with Canva as your most used presentation software. For something that launched less than a year ago (and just went GA last week), that is astounding. **Below, Mihika shares the seven biggest lessons she’s learned about successfully pitching big ideas, getting buy-in, rallying a team, and getting new products out the door—essentially, how to keep shipping like a startup as a company grows.** *Check out Mihika on [X](https://x.com/mihikapoor) and [LinkedIn](https://www.linkedin.com/in/mihikakapoor/), and, if you’re interested in more, she recently launched a live cohort-based course on Maven—[Product Storytelling: Pitch & Build 0-1 Products at Your Company](https://bit.ly/3RgoNjk)—where she goes even deeper on this topic. Her next class starts March 31st, so get on it.* ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/9458b655-5bcc-4334-93cb-4436c2cfe271_2912x2061.jpeg) We often hear about companies trying to maintain a “startup” mindset as they grow. They want to build products rather than follow processes, assemble irrationally passionate teams, and empower every employee with a great idea to ship a blockbuster. But BigCo status quo often gets in the way. Based on my experiences building Figma Slides, FigJam, and a few others in development, I’ve learned a lot about how to actually keep teams scrappy while launching products on the ground. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/32144d3b-fa9d-4f1a-a8e7-2947abb6ebcf_3840x2161.png) [Figma Slides](https://www.figma.com/slides/) was a bottom-up initiative that came to life through a mixture of prototyping, camaraderie, and optimism bordering on delusion. Evangelizing the idea at hackathons, vision pitches, and product reviews bought our team a small amount of headcount to “explore” the space. But that was all that we needed to sprint to the finish line. When Figma Slides debuted at Config 2024, Figma’s annual user conference, it proved to be a huge hit. Since the launch, we’ve seen people create millions of Figma Slides decks, and, as of last week, [it’s out of beta and available for all Figma users](https://x.com/mihikapoor/status/1902431297705804013) 🙌. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/e03649d8-5be3-49ad-9347-14e6ab4cd345_1456x821.jpeg)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/3602eeec-7242-44c2-9264-5acc4f115935_1600x900.png) Below are seven strategies for replacing the **BigCo status quo** with a **startup mindset**. Regardless of whether you are pushing your own idea within a larger company, fleshing out a new product space, or expanding existing initiatives, these tips will accelerate you, your team, and your company. ### 1. Replace your PRDs with prototypes 🤨 **BigCo status quo:** A PM’s job is to write detailed PRDs to kick off projects and keep teams aligned. 🏃 **Startup mindset:** Speed and experimentation matter more than documentation. Anyone with an idea should be empowered to explore it through prototypes. Learning how to write a PRD (product requirements document) constitutes one of the hallmarks of being baptized as a BigCo™ PM. But in fast-moving environments, PRDs often fail because: 1️⃣ They block valuable design and eng work. 2️⃣ They are almost never up-to-date, as specs change frequently. 3️⃣ Once a product exists, a PRD is just a watered-down version of the actual product being built. At Figma, everyone is encouraged to design and prototype—PMs included. Additionally, with the rise of [AI-assisted features](https://www.figma.com/blog/introducing-figma-ai/) to quickly get to a first draft, the barrier to getting started is plummeting even further, and PMs are much better off spending their time prototyping than writing docs. Conviction comes from prototyping, not paperwork. Tips for working without PRDs: #### **Create a prototyping culture.** If I had written a PRD for every idea we explored for Figma Slides, I’d *still* be writing those docs. Instead we built, tested, and iterated—fast. Encourage engineers to code their “what-ifs.” Some of them might turn into flagship features. Here are some of the very first prototypes of some of Figma Slides’ core features: slide grid, focused view, and design mode: #### **Align in lo-fi.** Use FigJam and quick wireframes to jam on ideas and explore solutions. Speed-run option spaces instead of spending weeks writing specs. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/215255e4-05a1-4f5e-9dbd-21dae98785ee_1600x579.png) #### **Let design files be the source of truth.** Once you align on a direction, don’t create a standalone PRD—annotate specs *inside* your designer’s mocks. A design file with notes is infinitely more useful (and accurate) than a text doc. #### **Invest in evergreen, visual artifacts.** You still need to communicate *why* you’re building something, but don’t rewrite it for every feature. Instead: - 📌 Build a single **vision artifact** up front—so everyone understands the broader impact you’re aiming for. Think about this as your pitch deck. - 📌 Update a **strategy artifact** each quarter—so teams see how near-term work ladders up. - 📌 Invest in **visuals**—directional designs tell the story better than text ever could. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/24a283d0-193c-486a-88d0-64c252fe107e_1600x900.png) **Two prototypes from the original deck of some core features: AI tone dial and the alignment scale:** ### 2. Build hype by working in the open 🤨 **BigCo status quo:** Sharing work in progress with colleagues leads to scrutiny of rough edges, depleted trust in the product, and an increased likelihood the project gets killed. 🏃 **Startup mindset:** Showing behind-the-scenes progress gets key stakeholders invested in the product’s success. Think of your product as a small flame. Your job? Grow it into an unstoppable wildfire inside your company. Do that by sharing progress early and often. Here’s how: #### **Light the first spark with hackathons.** Hackathons aren’t just for fun—they’re a launchpad. At Figma, previous attempts to get buy-in for Figma Slides through traditional pitches went nowhere. But when we built a working prototype at Maker Week, Figma’s company-wide hackathon, everything changed. The demo at our internal showcase took off because: 1. **It was big.** Everyone expected small feature demos, not an entirely new product built in three days. One engineer even asked if we had 40 people working on it (in reality it was only eight). ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/df48ae6b-c10e-4994-a0d1-89e9272c06f3_988x214.png) 2. **It was funny.** We held a live vote for the product name, with ridiculous contenders like “Feck” (Figma x Deck), “Sligma” (Slides x Figma), and my personal favorite, “Flides.” The absurdity of these suggestions made it an instant hit. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/d40c10de-f850-4f72-8939-4788e9c3614f_495x536.png) 3. **It was relatable.** To set the scene, I pretended I was a hackathon attendee late for submitting my demo video for Maker Week and used Flides to help me out. The demo mimicked a real-world scenario—being late on a presentation deadline. It resonated because it was exactly how we worked. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/270c533c-a08d-4e0c-9a70-51de6c8087ae_3024x1964.png)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/cafb31b9-ca86-4a46-aa97-3a71f77f3724_3024x1964.png) #### **Share raw progress async.** Don’t wait for polished releases—share raw progress. One engineer on our team, Jon, posted over 70 prototype videos in Slack before launch. Soon everyone followed his lead. A steady drip of updates builds trust and keeps excitement high. #### **Turn your spark into a wildfire at the all-hands.** Figma Slides got the green light to launch in beta at Config after it was pressure tested in a real-world scenario—I advocated to use our minimum viable product (MVP) at the first company all-hands meeting of 2024. Though our IT team was wary of betting on unproven software, the team paused new feature development for two weeks to refine the MVP for that moment, addressing bugs as soon as they came up. The demo was the best possible showcase of progress and agility, and the team learned so much about user needs. The flawless presentation in front of the whole company sealed the deal. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/8399080b-78aa-4710-a235-d1ab982834af_1600x899.png)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/1ffe7e02-cbca-4c5d-ac5c-a146cfc40938_916x514.png)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/a34dfe6d-0565-4ad8-afef-e038b6a1056a_1600x1471.png) #### **Bonus: Share beyond your company.** We launched Figma Slides in open beta at Config so we could gather as much feedback as possible before GA. The team shared some of our early WIP vision on platforms like X (with the GTM team’s blessing, of course), which proved to be an effective way to drive hype. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/5d789b70-5a8f-40db-a0af-654b50d890ec_2576x1448.png) ### 3. Build a cult(ure) 🤨 **BigCo status quo:** PMs are responsible for the product and the product alone. 🏃 **Startup mindset:** A product’s success depends on a team culture so strong it feels borderline religious. Culture isn’t just a nice-to-have—it’s a cheat code for momentum, creativity, and execution. Here’s how to build one that people want to be part of: #### **Make it weird.** Inside jokes create attachment. The weirder, the better. For Figma Slides, the absurd name “Flides” became an internal meme: - Slack emoji? ✅ *[Make one for your team today if you don’t have one!]* - Workstream names? ✅ *[The animations sync? “Flanimations.”]* - IRL team photo shoot . . . on actual slides? ✅ The more irrational the name, the stronger the inside joke—and the deeper the buy-in. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/d4534ab0-a745-4c4e-a637-85ebb6a38938_1456x644.jpeg)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/996b6424-3040-4ec8-88ba-45025f357e7e_4375x2024.png)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/4b553569-a4ed-43f8-930f-6cefd08995f7_5568x3712.jpeg) #### **Steal each other’s quirks.** Shared quirks = shared identity. Our team leaned into this, hard. Everyone changed their Slack profile picture’s background to match. A Gen Z engineer always opened chats with “yoooooo”—so the whole team adopted it. Even small things, like entering a group chat, became dopamine hits. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/a2308c7c-96d4-4c7c-ab07-f7b0e066d776_2000x1578.png) #### **Celebrate like it’s the Oscars.** On launch day, we threw an award ceremony. Red carpet, custom trophies, full production. What started as a joke turned into one of the most emotional moments of the entire project. The team was so hyped, we closed out the office at midnight. Want maximum inside jokes with minimal effort? Crowdsource superlative awards from the team. P.S. Trophies are $20 on Amazon. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/85112a0c-cfed-412a-a46e-e147ca4efa1f_1600x1200.jpeg)![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/68fe86cd-37e4-4912-9599-540342b2ba6b_1200x1600.jpeg) ### 4. Find the believers instead of swaying the skeptics 🤨 **BigCo status quo:** People work on the products they are assigned to. 🏃 **Startup mindset:** Players who opt in to their teams build the most successful products. The Figma Slides team was the most inspiring group I’ve worked with—not because they were assigned but because they *wanted* to be there. We pulled team members from across the org by spreading the word and empowering the most excited and invested folks we heard from to make the switch. Here’s how to build your dream team: #### **Find your early adopters.** A piece of feedback I was given early on, that to this day resonates with me, is: *If you need to sell someone really really (like, really really) hard to join you, they are probably not the right fit*. The best teammates are the ones begging their managers to work on your project. Make sure your project is known across your org (and ideally the whole company) so they can find you. #### **Look for “run with it” energy.** The best team isn’t always the most senior—it’s the one that can move forward without perfect specs. At Figma, we call this the “run with it” mentality. Find engineers and designers who thrive in ambiguity. #### **Recruit scrappily.** Don’t let org swimlanes prevent you from reaching out. I love working with engineers and designers both on and off my team. The best part is, if they like working with you, they’ll make the time to help out. If they love working with you, they’ll do whatever they can to join you. #### **Empower every team member to pitch.** Your team should be your biggest evangelists. The best pitches will actually come from the folks on your team. Ensure that they know when recruiting is a priority and that you are available as a resource. ### 5. Make every team member a product owner 🤨 **BigCo status quo:** Research hands off to Design, Design hands off to Engineering, and everyone stays in their lane. 🏃 **Startup mindset:** The best teams blur the boundaries. Everyone is a part-time PM. Before Figma, I was told to stay in my lane. “That’s a designer’s job” or “This should be done by a data scientist” were constant refrains. Deviating from the original plan was considered “thrash.” At Figma, it’s the opposite. Extreme collaboration isn’t just encouraged—it’s the norm. We’re able to improve decision-making and expand the solution space when we lean into the following practices: #### **Remove the black box around the product and design process.** Engineers shouldn’t be kept in the dark until handoff. Involving them early gives them better context, improves decision-making, and allows them to shape product behavior. Engineers are often brought into design syncs where product decisions are being made. That way: - Design proposals are already informed by engineering feasibility. - Engineers can proactively contribute insights, not just execute specs. #### **Go from 0% to 100% together.** At Figma, teams don’t work in silos—they *pair* on solutions. PMs and designers jam together in a Figma file. Handoffs between design and engineering aren’t just “Here, build this”—they’re real-time pairing sessions, tweaking interactions and refining details together. The result? A stronger product, fewer roadblocks, and a team that feels true ownership over what they’re building. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/0bc6669b-9f8c-4cec-b5aa-9ff2f6ee063d_1600x481.png) ### 6. Remove the wall between your team and your users 🤨 **BigCo status quo:** Researchers are the only ones talking to users. 🏃 **Startup mindset:** Everyone on the team engages with users to demonstrate, deploy, and debug products. One of the most fascinating phenomena I’ve witnessed is that if an employee hears a user directly express and elaborate on a pain point, they will be a bajillion times more likely to act on it than if they hear a secondhand summary. As a PM, your job is to engineer as many of these anecdote-driven crusades as possible. How to get your team talking to users: #### **Require research attendance.** Set an expectation that each person on the team attends *at least* 50% of research sessions. #### **Invite questions.** At early-stage startups, *everyone* talks to users—not just dedicated researchers. Encourage your team to chime in on live user calls and follow promising lines of questioning. *(Pro tip: Avoid leading questions!)* #### **ABR (Always Be Researching).** User interviews aren’t the only way to gather insights. At Figma, we often reach out directly to users on X and other community forums to get their thoughts on what we’re building. ![Image from How to ship like a startup](https://substack-post-media.s3.amazonaws.com/public/images/cce30015-bc4f-4bdd-b7e6-4f9589febba0_1076x390.png) Once the product is public, ask a friend. Ask your mom. Ask a random person in a coffee shop. The best feedback often comes when you least expect it. #### **Prototype for proof.** Throwaway code is a superpower. Quick, scrappy prototypes help validate ideas, clarify tradeoffs, and get stronger user reactions. A two-week experiment that saves four months of misguided development? Always worth it. ### 7. Meat and potatoes come before dessert and soufflé 🤨 **BigCo status quo:** I can only work on the projects I’m assigned to. 🏃 **Startup mindset:** Once I build trust with the people funding and resourcing me, I can sprint on *any* idea. If there’s one takeaway from the Figma Slides journey, it’s this: *Anyone* with a strong idea can turn it into something real. But first you need to become a trusted person. A product leader once told me, “The way to get leadership to fund your idea (your soufflé) is by first executing on their ideas (the meat and potatoes).” #### **How to earn that trust:** ✅ **You need reps.** → Ship things. ✅ **You need wins.** → Ship *good* things. ✅ **You need a line of communication.** → Leaders need to *know* you. ✅ **You need excellent communication.** → Leaders need to *respect* you. Once you check these boxes, the world is your oyster. For me, launching some of FigJam’s earliest features—Sections, Music, Voting—gave me the reps and wins I needed. Those *meat and potatoes* paved the way for my soufflé. If you’re sitting on a burning idea, I hope you too get to relish your soufflé. 😉🔥 *Thank you, Mihika! For more from Mihika, check her out on [X](https://x.com/mihikapoor), [LinkedIn](https://www.linkedin.com/in/mihikakapoor/), and [her Maven course](https://bit.ly/3RgoNjk). Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [13/46] The definitive guide to mastering product sense interviews *👋 Welcome to a **✨ free edition ✨** of my weekly newsletter. Each week I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Podcast](https://www.lennysnewsletter.com/podcast) | [Courses](https://maven.com/lenny) | [Hiring](https://www.lennysjobs.com/) | [Swag](https://lennyswag.com/)*** *Annual subscribers now get a free year of **[Perplexity Pro, Notion, Superhuman, Linear, and Granola](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle)**. [Subscribe now](https://www.lennysnewsletter.com/publish/post/https://www.lennysnewsletter.com/subscribe?).* A lot of you are having a hard time finding a job. I hear stories from readers who send out hundreds of resumes and get zero responses, go through month-long interview loops that often end in getting ghosted, and continue on without ever hearing honest feedback about what they could do differently. I want to help. In the coming months, I’m going to be sharing a number of newsletter posts with the goal of helping you find a job you love. Today’s post is the first in that series, and it’s a doozy. The product sense interview is one of the most important and most mysterious steps of the PM interview loop, and [Ben Erez](https://www.linkedin.com/in/benerez/) just wrote your new bible for mastering that interview. Ben shares exactly what the product sense interviewer is looking for, a simple framework for structuring your answer in real time, how to practice for these interviews, and even specific phrases you should be using. **If you read this post and take its advice, you will** ***significantly*** **increase your chances of landing a job.** *[Here’s a link to part 2 of this series—[The definitive guide to mastering analytical thinking interviews](https://www.lennysnewsletter.com/p/the-definitive-guide-to-mastering-f81)—which was published after this came out]* Ben has been a founder, a PM at Meta, and the first PM at three different startups, and, as a coach, he’s now helping PMs land their dream roles. Ben’s got [a free 30-minute lightning lesson coming up on April 16th](https://maven.com/p/d8e439/practice-for-product-sense-pm-interviews-with-ai) where he’ll show you how to use Claude Projects to practice for your product sense interviews, and he also teaches [a two-week immersive PM interview bootcamp](https://bit.ly/4iQm5xq) where you’ll master the product sense and analytical interviews with live case discussions and personalized expert coaching. Use code “LENNYSLIST” to get $100 off. Enrollment closes April 22nd. Now, on to today’s post. . . ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/549dbd65-871f-4438-91be-e73e8d21ec09_2912x2124.jpeg) In 2020, I interviewed for a product manager role at Facebook after spending nearly a decade in early-stage startups. While I had plenty of interview experience at Series A and B startups, structured Big Tech PM interviews felt intimidating to me. To land the offer at Facebook, I would need to ace three interviews: Product Sense, Analytical Thinking (“Execution” back then), and Leadership & Drive. I found myself wondering: How should I prepare? What does a strong response even look like? Over the course of four intense weeks, I spent countless hours watching every mock interview I could find on YouTube and picking the brains of PMs in my network who passed these interviews. I noticed patterns in the way successful candidates tackled the questions in the YouTube mock videos and drafted a basic template for each interview based on the patterns. I practiced new questions with those paper templates, increasing my skill and confidence. I ultimately aced the interviews and landed the offer. During my time at Facebook, I interviewed over 50 candidates, which gave me a deeper understanding of the interviewer’s perspective. Over the past few years, I’ve coached hundreds of candidates for interviews and noted common behaviors of successful candidates. I’ve been distilling these insights into a continually updated set of frameworks and templates that have helped dozens of candidates land total comp packages of upward of $600,000. People say the best way to learn something is to teach it, and, through coaching, I’ve leveled up my own understanding of these interviews. Originally popularized by Facebook and Google, product sense (PS) and analytical thinking (AT) interviews have now been widely adopted across the tech industry, with companies like Stripe, OpenAI, Block, and many others including PS and AT screens in their PM interview loops. Even earlier-stage companies have adopted these in their process. To land my next PM role after Facebook (at [Attentive](https://www.attentive.com/)), I had to pass PS and AT interviews. With the goal of making my materials more widely accessible to the largest number of PM candidates in today’s competitive hiring market, I partnered with Lenny to publish this comprehensive guide for PM candidates to master product sense interviews, which are becoming ubiquitous in PM hiring. Whether you’re actively preparing for upcoming interviews or simply want to understand how top tech companies evaluate PMs, this guide will equip you with all the tools and frameworks you need to start your preparation. ### Understanding the product sense interview Product sense interviews assess your ability to identify user needs, articulate problems, and craft compelling solutions while demonstrating empathy, creativity, and structured thinking. These interviews are typically 45 minutes long, giving candidates roughly 35 minutes for the core exercise after accounting for introductions and closing questions. To pass, candidates need to achieve at least a passing score on each dimension within the time box: - **Clear communication:** Delivering an easy-to-follow response with solid structure - **Product motivation:** Examining the product’s purpose and mission - **Segmentation:** Defining the audience and prioritizing a target segment - **Problem identification:** Listing and prioritizing key pain points faced by the chosen segment - **Solution development:** Brainstorming and prioritizing a solution for the problem, including a v1 Excellence in one area can’t compensate for weakness in another. But you don’t need perfect scores across the board; a solid score for each dimension is sufficient to pass. A structured approach gives you the best chance of delivering all the signals interviewers need before running out of time. I’ve included in this post concrete ways a candidate might discuss each of the above evaluation dimensions for three example product sense questions: - “Tell me about a product you love and how you would improve it.” - “You’re a PM at Meta. Design a product for gardening.” - “Build a podcast product for Netflix.” *Note: For the “Tell me about a product you love and how you would improve it” question, I chose **Claude Projects** because it’s an incredible product that I use regularly.* I generally coach candidates to structure their game plan for the interview in a way that enables them to provide strong signals on the interview dimensions in a linear flow: ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/c8182661-5596-44d2-b56c-14c4f4423d21_1456x2317.jpeg) #### **A quick note about interviewer empathy** Understanding the interviewer’s perspective can help you structure your responses effectively. Product sense interviewers operate under constraints you might not consider: - Most interviewers are busy PMs conducting interviews to help the organization, balancing this responsibility with demanding day jobs. - Interviews are typically thrown on the interviewer’s calendar by a coordinator. - The rubric for PS and AT interviews is generally consistent across seniority levels. What does this mean for you? Since the interviewer’s job is to efficiently collect specific signals to complete their evaluation, your job as a candidate is to generate clear, easy-to-identify signals. This post offers a deep dive into those signals for **product sense** interviews. Next, I’ll share my framework for each of the five dimensions of the product sense interview. ## Baseline skill: Clear communication The opening minutes of a product sense interview can make or break your performance. I’ve watched countless candidates struggle because they jumped straight into brainstorming without aligning with the interviewer on the structure. The best candidates approach these interviews as a specific game with clear rules rather than a casual conversation. Generally there are three components to address for clear communication: 1. **Waypointing:** Before each section, candidates should take a thinking pause to work for 1-2 minutes and come up with what they want to say. Once they’re ready to share, candidates should walk the interviewer through their response for that section and check in with the interviewer before proceeding to the following section. 2. **Assumption setting:** Make 2-4 assumptions at the start of the interview that narrow the scope to a manageable exercise without prematurely limiting your solution space. 3. **Game plan articulation:** Clearly outline your approach for the entire interview, showing the interviewer how you’ll structure your time. To set yourself up for success, these are a few potential assumptions to state: - **Role and context:** State your assumed role and the company/product context for the exercise. - **Geographic focus:** Specify whether you’ll focus on a specific market or region for your response. - **Platform and constraints:** Identify any technical or strategic constraints that will help focus your solution, without over-limiting possibilities. These assumptions should help you focus the discussion without prematurely limiting your solution space. The goal is to establish enough structure to proceed efficiently while remaining open to the interviewer’s guidance. After stating assumptions, strong candidates review their game plan for the interview and check in with the interviewer to make sure they’re on the same page about how they’ll be spending the time together. Regardless of the question asked, I recommend that candidates state something along these lines: > *“Before I dive in, I’d like to walk you through my plan for our time together:* > > - *I’ll start by describing the product/experience and why it matters.* > - *Then I’ll break down the target audience and define a segment to focus on.* > - *From there, I’ll identify key problems for that segment and prioritize one.* > - *I’ll then brainstorm solutions and pick one.* > - *If we have time, I’d love to describe a v1 implementation of that solution.* > > *Does this plan sound good to you?”* This is music to the ears of the interviewer. It assures them that the candidate is prepared and has a plan to generate the signals they’re looking for. It’s like discovering they’re about to start a game with someone who knows how to play it. It also offers the interviewer a chance to redirect the candidate, preventing wasted time heading down an undesired path. Interviewers value candidates who lead with structure while remaining adaptable to the interviewer’s input. ### Elevating clear communication What separates good communication from exceptional communication in these interviews: - **Strategic structure ownership:** Drive the conversation with clear assumptions and a well-articulated plan, demonstrating leadership while remaining adaptable to interviewer feedback. - **Explicit waypointing:** Use verbal signposts throughout your response that help the interviewer track your thinking, clearly distinguishing between your process and conclusions. - **Balanced scope management:** Choose assumptions that provide helpful guardrails without prematurely closing off creative solution paths, showing you can manage complexity effectively. ### Examples of articulating assumptions #### **Question 1: Tell me about a product you love and how you would improve it** > - *I’ve been using Claude Projects and love it, so I’ll focus on this product and assume I’m the PM for Claude Projects at Anthropic, building within Anthropic’s existing product ecosystem.* > - *I’ll focus on the Claude Projects feature within the Claude Pro subscription, targeting the U.S. market initially but leaving the door open to serving a global audience with our solution.* > - *I’ll focus on improvements for individual Claude users rather than Anthropic’s enterprise API customers, since Projects doesn’t apply to API customers.* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** > - *I’ll assume I’m the PM for a new zero-to-one product for gardening at Meta.* > - *I’ll focus on the U.S. market for this product because it’s what I’m most familiar with, and we can likely scale our solution globally in the future.* > - *For distribution advantages and accelerated impact, I’ll plan to design a solution within Meta’s family of apps (Facebook, Instagram, WhatsApp) rather than as a standalone product.* > - *I’ll focus on personal/home gardening rather than commercial or botanical gardens.* #### **Question 3: Build a podcast product for Netflix.** > - *I’ll assume I’m the PM for a new zero-to-one podcast product at Netflix.* > - *I’ll focus on the U.S. market initially, though Netflix is global.* > - *For distribution and accelerated impact, I’ll aim to build within Netflix’s existing platform/ecosystem.* > - *Podcasts include both audio and video, but I’ll focus on audio for this exercise, since most podcasting usage is via audio.* > - *While we could build something for people who don’t yet use Netflix, I’d like to limit our exercise to serving existing Netflix subscribers, since we can make a large impact by serving hundreds of millions of people.* ### Top pitfalls for clear communication Understanding what makes for clear communication is only part of the equation. Just as important is knowing what to avoid. Here are the most common pitfalls I’ve observed candidates step into when structuring their communication: - **Thinking out loud without structure:** Starting to brainstorm without a clear framework forces the interviewer to piece together your thought process, making it difficult for them to follow your reasoning and identify the signals they need to evaluate. - **Asking the interviewer for direction:** Frequently asking, “What would you like me to do next?” or “Would you prefer I focus on X or Y?” signals a lack of ownership and leadership. Strong candidates proactively drive the conversation while remaining adaptable to feedback. - **Premature narrowing:** Making assumptions that prematurely narrow your solution space (like assuming specific features or user demographics) limits your options later. Focus on assumptions that provide structure without closing doors too early. ### Practice tips for clear communication To boost your confidence and effectiveness when starting a product sense interview, consider these tactical preparation tips: - **Create a standard template:** Develop a simple template with 3-4 placeholder assumptions (role, geography, platform constraints) that you can quickly adapt for any product sense question. This gives you immediate structure and helps you avoid the blank-page anxiety that can derail your opening. - **Build muscle memory through repetition:** Practice stating your assumptions and game plan for 20-30 different product sense questions. By rehearsing these opening elements repeatedly, you’ll develop the confidence to start strong even when nervous or faced with an unexpected question. - **Master your waypointing transitions:** Practice explicit verbal signposts like “Now that I’ve established assumptions, I’ll outline my plan for our time together” or “Before I think through the product motivation, does this game plan sound good?” These transitions help the interviewer follow your structured thinking and create natural check-in points. After establishing clear communication fundamentals, the next crucial step is articulating why the product matters. This transition from “how we’ll spend our time” to “why this product exists” creates the foundation for all your subsequent decisions. Let’s explore how to develop a compelling product motivation that will impress interviewers and set you up for success. ## Step 1: Product motivation The product motivation section sets the tone for the entire interview by demonstrating your holistic understanding of the purpose behind products, rather than just their features. While it might be tempting to rush through this part to get to “meatier” sections, taking 3-5 minutes to thoughtfully establish why the product matters creates a strong foundation before diving into segmentation, problem identification, and solutions. Generally there are three components to address in this section: 1. **Product description and value:** Start by clearly describing what the product or experience entails and connect it to deeper human needs that it addresses. Why does this product matter to users? Why is the world better with this product in it? 2. **Strategic and competitive context:** Articulate how this product advances the company’s strategy and fits within its broader ecosystem. Consider market trends, key competitors, and what makes this offering uniquely valuable in the current landscape. 3. **Mission statement:** Finish this section with a concise, purpose-driven mission statement that will guide your decision-making throughout the interview. This statement should be specific enough to provide direction but broad enough to allow creative exploration. When articulating the product motivation, strong candidates go beyond superficial descriptions to articulate deeper purpose. For instance, rather than saying, “This product helps users share photos,” you might say, “This product enables people to maintain meaningful connections with loved ones through visual storytelling, even when physically apart.” A compelling mission statement serves as a North Star for your interview, guiding your segmentation, problem identification, and solution development. It should be specific enough to provide direction but broad enough to allow for creative exploration. For example, instead of “To help people take better photos,” a stronger mission might be “To empower anyone to capture and preserve meaningful moments through intuitive photography tools.” ### Elevating product motivation What separates good product motivation from exceptional product motivation: - **Human-centered purpose articulation:** Connect the product to deeper emotional and practical human needs beyond surface functionality, demonstrating empathy and insight into why people would truly care about this product. - **Strategic ecosystem positioning:** Articulate how the product creates unique value within the company’s broader ecosystem and competitive landscape, showing business acumen and systems thinking. - **Mission-driven decision framework:** Craft a mission statement specific enough to guide meaningful decisions but broad enough to allow creative solutions, creating a powerful reference point for the rest of your interview. To illustrate how effective product motivation might look in practice, here are examples for our three interview questions, showing how to articulate the product’s purpose within its broader context: ### Examples of articulating product motivation #### **Question 1: Tell me about a product you love and how you would improve it.** > *Claude Projects is a feature that allows users to create specialized Claude instances trained on custom documents and knowledge, enabling each chat to be primed with specific context.* > > *While general-purpose AI assistants have broad knowledge, they often lack depth in niche areas. Claude Projects addresses this gap by allowing users to augment Claude’s capabilities with their own materials. Domain-specific AI assistants are increasingly valuable for knowledge workers who need tools that understand their unique contexts.* > > *Key competitors include:* > > - *OpenAI* > - *Perplexity* > - *Google* > > *Claude Projects directly contributes to Anthropic’s mission of building AI systems that are helpful, harmless, and honest. Improving this product would increase Claude’s adoption engagement, accelerating Anthropic’s ability to achieve its mission.* > > *Let’s use this as a placeholder mission statement for Claude Projects:* > > *“To empower users to create AI assistants with specialized knowledge and capabilities tailored to their unique needs, making AI more personally relevant and useful in their daily lives and work.”* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** > *Gardening is a popular hobby that involves cultivating plants, whether for food production or aesthetic enjoyment. It’s an activity that brings people joy, connects them to nature, and can be both therapeutic and practical. During the pandemic, we saw a significant surge in gardening interest as people spent more time at home.* > > *Meta’s mission is to build the future of human connection and the technology that makes it possible. Gardening represents a meaningful opportunity to leverage technology to connect people through shared interests and knowledge exchange. Currently, gardening enthusiasts are spread across various platforms and communities, making knowledge fragmentation a challenge.* > > *A gardening product within Meta’s ecosystem could leverage our social graph to connect gardeners with one another and with relevant knowledge, creating more meaningful connections through this shared passion.* > > *Let’s use this as a placeholder mission statement for Meta’s gardening product:* > > *“To empower people to connect and thrive through the shared experience of gardening, creating communities that grow together both online and in their gardens.”* #### **Question 3: Build a podcast product for Netflix.** > *Netflix’s mission is “to entertain the world,” primarily through video streaming, but audio content consumption has grown significantly, with hundreds of millions of podcast listeners in the U.S. alone.* > > *Most podcasts fall into the buckets of serialized entertainment, current events, or engaging conversations about interesting topics.* > > *While Spotify and Apple dominate the podcast space, Netflix has unique advantages: a massive entertainment-seeking user base, content creation expertise, and established creator relationships. Expanding into podcasts would deepen user engagement and create additional touchpoints when video isn’t practical (commuting, exercising, etc.).* > > *Netflix can differentiate its podcast offering by leveraging its recommendation engine and existing payment model, addressing the discovery and monetization challenges that plague current podcast platforms.* > > *Let’s use this as a placeholder mission statement for Netflix’s podcast product:* > > *“To enrich people’s entertainment experience beyond the screen by delivering captivating audio content that can be enjoyed wherever they go.”* ### Top pitfalls for product motivation Understanding what makes a great product motivation is only part of the equation. Just as important is knowing what to avoid. Here are the most common pitfalls I’ve observed candidates step into when discussing product motivation: - **Focusing only on features:** Describing what the product does without articulating why it matters misses the opportunity to demonstrate strategic thinking. Strong candidates connect functionality to deeper value and purpose. - **Ineffective mission statements:** Creating either overly vague missions (“help users be more productive”) or too narrowly defined purposes that don’t provide meaningful guidance for later decisions. Your mission statement should be specific enough to guide choices but broad enough to allow creative solutions. - **Strategic misalignment:** Either proposing a product direction that contradicts the company’s core strengths or focusing exclusively on company benefits without addressing user needs. The best product motivations balance user value with business strategy. ### Practice tips for product motivation To strengthen your product motivation skills, deliberate practice is essential. Here are a handful of exercises and techniques that can help you develop this critical area and deliver more impactful responses in your interviews: - **Connect products to human needs:** For everyday products you use, practice articulating not just what they do but why they matter on a deeper human level. This builds your ability to connect features to emotional and practical benefits, demonstrating that you understand that products must address fundamental needs beyond surface-level functionality. - **Study existing mission statements:** Analyze mission statements from successful products and identify what makes them effective guiding principles. Look for patterns in clarity, specificity, and purpose to understand how well-crafted mission statements balance aspiration with actionable direction. - **Time yourself:** Practice delivering complete product motivation sections in under 5 minutes. This develops the discipline to be concise yet comprehensive, ensuring that you have sufficient time for later sections of the interview while still fully articulating why the product matters. After establishing your product’s motivation and crafting a compelling mission statement, the next critical step is to identify and prioritize your target audience. This transition from “why” to “who” is where many candidates stumble, either by staying too high-level with generic user groups or getting lost in complex segmentation frameworks. Let’s explore how to approach segmentation in a way that impresses interviewers and creates a solid foundation for problem identification. ## Step 2: Segmentation The segmentation section demonstrates your ability to think both broadly about the ecosystem and deeply about specific user needs. Strong candidates methodically move from the big picture to a focused target, creating a foundation for identifying meaningful problems to solve. Generally there are three components to address in this section: 1. **Ecosystem analysis:** Identify all major stakeholders in the product ecosystem, demonstrating systems thinking beyond just end users. Then choose one ecosystem group to focus on, guided by your product mission and strategic priorities, with a clear rationale for this selection. 2. **Segment creation and prioritization:** Break down your chosen player into distinct segments based on behaviors, motivations, and context—not just demographics. Then evaluate these segments strategically using a reach vs. underserved degree framework to identify high-potential opportunities. 3. **Persona development:** Bring your chosen segment to life with a specific, relatable persona that embodies key characteristics and challenges. This bridges the gap between abstract segments and concrete problems for the next section. When developing segments, look for meaningful differences in behaviors, needs, and contexts. Start with understanding core motivations: why people use or would use your product. These motivations inform more specific segmentation criteria such as: - **Primary motivations:** What fundamental goals drive their behavior? - **Behavioral patterns:** How frequently and in what ways do they interact with similar products? - **Context of use:** Where, when, and how do they engage with the product/experience? - **Expertise level:** Are they novices, intermediate users, or experts? - **Resource constraints:** What limitations (time, money, knowledge, space) affect their usage? - **Goals and outcomes:** What specific results are they trying to achieve? A crucial test for effective segmentation is mutual exclusivity: each segment should represent users with distinctly different needs that couldn’t simultaneously belong to multiple segments. If someone could easily fit into several of your segments at once, your segmentation isn’t precise enough to guide clear product decisions. The key is creating segments that are meaningfully different from each other, where the needs of one segment would require substantially different solutions than another. Avoid segmentation that merely describes slightly different versions of the same basic user. ### Elevating segmentation What separates good segmentation from exceptional segmentation in these interviews: - **Motivation-based differentiation:** Go beyond demographics or simple usage patterns to create segments based on fundamentally different user motivations and goals, demonstrating deep understanding of what drives user behavior. - **Mutual exclusivity testing:** Create segments where users clearly belong to one group rather than potentially fitting multiple categories simultaneously, ensuring clear prioritization decisions and focused product direction. - **Vivid persona creation:** Transform abstract segments into specific, relatable individuals with detailed contexts, constraints, and behaviors that make their needs feel authentic and actionable. Let’s walk through segmentation for each of our example questions, starting with ecosystem players: ### Examples of articulating ecosystem players #### **Question 1: Tell me about a product you love and how you would improve it.** > *Major ecosystem players I would consider include:* > > - *Individual Pro users: Individuals who subscribe to Claude Pro* > - *IT decision-makers: Those who evaluate and approve AI tools for organizational use* > - *Anthropic product team: Those building and enhancing the Claude Projects feature* > - *Content creators: People whose content is being leveraged in a Claude Project* > > *From these ecosystem players, I’d like to focus on **Individual Pro users** as the primary group to target. They represent the direct consumers of the product and are the ones interacting with Claude Projects on a regular basis.* #### Question 2: You’re a PM at Meta. Design a product for gardening. > *Major stakeholders in the gardening ecosystem include:* > > - *Home gardeners: individuals who garden at home* > - *Plant nurseries and garden centers: businesses selling plants and supplies* > - *Gardening content creators and experts: individuals sharing knowledge* > - *Garden product manufacturers: tools, soil, fertilizers, etc.* > - *Local gardening communities and clubs* > > *For this exercise, I’d like to focus on **home gardeners** as the primary ecosystem player, since they align best with Meta’s mission of connecting people and would be the core users of our product.* #### **Question 3: Build a podcast product for Netflix.** > *Major stakeholders I would consider include:* > > - *Listeners: Netflix subscribers consuming podcast content* > - *Content creators: podcast hosts, producers, storytellers* > - *Advertisers: potential monetization partners* > - *Netflix internal teams: content, tech, marketing* > - *Competing podcast platforms: Spotify, Apple, etc.* > > *For this discussion, I’d like to focus on the **listeners** as our primary ecosystem group, as they represent the demand side of this product. Focusing on listeners aligns with Netflix’s customer-centric approach and would help us understand what type of podcast offering would drive the most engagement and value.* After identifying and selecting your primary ecosystem player, I recommend a quick check-in with the interviewer to prevent wasting time developing segments for an ecosystem player the interviewer might want you to reconsider: > *“I’d like to focus on [chosen ecosystem player] for these reasons [brief rationale]. Before I dive into defining segments within this group, I want to pause to make sure this line of thinking is making sense?”* Once you’ve received confirmation to proceed, the next critical step is to develop meaningful segments within your chosen ecosystem player. Strong segmentation goes far beyond basic demographics to uncover distinct user types with different needs and behaviors. Here are segmentation heuristics we can use to craft segments within a chosen ecosystem player: ### Examples of articulating segmentation heuristics #### **Question 1: Tell me about a product you love and how you would improve it.** > *Primary motivations for using Claude Projects might include:* > > - *Knowledge management: organizing and accessing personal or professional information* > - *Specialized assistance: getting domain-specific help in areas of expertise or interest* > - *Workflow automation: creating more efficient workflows* > - *Learning/education: using AI to help understand complex material* > > *Based on these motivations, I could define segments that factor in:* > > - *Technical sophistication (low to high)* > - *Frequency of AI assistant usage (occasional to power user)* > - *Document volume and complexity (small/simple to large/complex)* > - *Primary use context (personal vs. professional)* > - *Domain specialization (general vs. highly specialized)* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** > *Let’s consider the primary motivations that drive people to garden:* > > - *Functional gardeners: garden primarily to grow their own food* > - *Aesthetic gardeners: focused on beautifying their spaces with plants* > - *Wellness gardeners: garden for mental health and relaxation* > - *Social gardeners: see gardening as a way to connect with community* > > *Based on these motivations, I could define segments that factor in:* > > - *Experience level* > - *Available space* > - *Time commitment* > - *Urban vs. suburban vs. rural* > - *Digital savviness* #### **Question 3: Build a podcast product for Netflix.** > *Motivation groupings for listeners:* > > - *Education/learning* > - *Entertainment/narrative* > - *Community/fandom* > > *Based on these motivations, I could define segments that factor in:* > > - *Content types they consume (documentaries, narrative shows, etc.)* > - *Viewing behavior (binge-watching, casual viewing, rewatching)* > - *Engagement level with specific content* Considering the segmentation heuristics, candidates can then define concrete user segments to focus on for the rest of the interview. For time management and clarity, I recommend three segments. It’s better to present a thoughtful analysis of three meaningful segments than a shallow analysis of many. Once segments are defined, the next crucial step is to strategically prioritize one segment and bring it to life through a concrete persona. A well-crafted persona transforms an abstract user segment into a specific individual with context, constraints, and motivations. This helps focus our solution development and demonstrates to interviewers that we can make decisive product choices based on impact potential. Here are examples of how we can define and prioritize distinct segments, including a target persona. ### Examples of articulating segment prioritization with persona #### **Question 1: Tell me about a product you love and how you would improve it.** ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/3d3e9b51-d98c-47a0-8d46-e58695215862_2913x2381.jpeg) > *I’d like to focus on the **knowledge specialists** segment for several reasons:* > > - *They have the highest underserved degree, representing a unique opportunity.* > - *They align well with Anthropic’s mission to advance beneficial AI by supporting domain experts.* > - *They’re likely to be early adopters and provide valuable feedback for product improvement.* > - *Success with this segment could establish Claude Projects as the premium solution for specialized AI assistance.* > > *For our persona, let’s consider James, 37, a medical researcher specializing in rare genetic disorders. James has extensive research papers, clinical studies, and personal notes that he’d like to organize and query efficiently. He’s technically savvy but not a developer, and needs to quickly find connections across his extensive research library.* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/790edb7f-9805-43a6-bba9-b8b744e4ef16_2913x2205.jpeg) > *I’d like to focus on **novice urban gardeners**. They have high reach and are highly underserved, making them a strategic opportunity. This segment also aligns well with Meta’s strengths in connecting digitally savvy users and providing easy-to-use tools for sharing and learning.* > > *For our persona, let’s consider Casey, a 29-year-old marketing professional living in a city apartment with a small balcony. She’s excited about growing some plants but isn’t sure where to start and has limited space.* #### **Question 3: Build a podcast product for Netflix.** ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/918f9546-6db2-475f-9ad7-7b6cb1ff4d6d_2913x2242.jpeg) > *I’d like to focus on **educational-content extenders** because:* > > - *This segment leverages Netflix’s growing investment in documentary and educational content.* > - *They have a high underserved degree despite significant reach.* > - *It creates a natural extension opportunity for Netflix’s nonfiction content.* > - *It differentiates from competitors by focusing on knowledge-driven audio content.* > > *For our persona, let’s consider Alex. Alex watches Netflix documentaries about science, technology, and history regularly and often researches topics further after watching. They listen to educational podcasts during commutes and would value expert discussions that build upon the documentaries they’ve recently watched.* ### Top pitfalls for segmentation Beyond understanding what makes for strong segmentation, it’s equally important to avoid common mistakes. Here are the biggest pitfalls I consistently observe candidates falling into during the segmentation phase: - **Shallow segmentation:** Relying solely on demographic categories like “millennials in urban areas” creates weak foundations. Instead, combine multiple dimensions including behaviors, motivations, and contextual factors to create meaningfully different segments with distinct needs. - **Skipping ecosystem analysis:** Jumping straight to end-user segments without considering the broader ecosystem (supply side, demand side, partners, etc.) signals a lack of systems thinking and can miss critical strategic opportunities. - **Non-mutually-exclusive segments:** Creating segments where users could simultaneously belong to multiple categories leads to unclear prioritization and diluted product focus. Test your segments by asking if someone from one segment could be the same person as someone from another. ### Practice tips for segmentation To strengthen your segmentation skills and deliver more compelling responses in your interviews, consider these targeted practice exercises that build muscle memory for effective audience breakdown: - **Ecosystem mapping exercises:** Practice identifying all stakeholders by drawing ecosystem maps for 5-10 products you use regularly. For each product, identify at least 5 different ecosystem players. - **Mutual-exclusivity testing:** After creating segments for practice questions, explicitly test each set of segments by asking, “Could someone belong to multiple segments simultaneously?” If yes, refine your segmentation criteria until the answer is no. - **Reach-underserved matrix practice:** Practice placing your segments on a 2x2 grid of reach vs. underserved degree (just Low and High), forcing yourself to make clear distinctions between segments rather than conveniently picking the middle. After establishing your target audience through effective segmentation, the next critical step is to identify meaningful problems for your chosen persona. This transition from “who” to “what hurts them” is where many candidates demonstrate either surface-level thinking or exceptional user empathy. With a well-defined segment and vivid persona in mind, you’re now equipped to identify the specific pain points that prevent your users from achieving their goals. Let’s explore how to articulate problems that will resonate with interviewers. ## Step 3: Problem identification Problem identification is where you transform your understanding of the user segment into concrete pain points that need solving. This section demonstrates your ability to think critically about user experiences and identify meaningful challenges that could drive product development. Strong product managers don’t just build features—they solve real problems. In interviews, your ability to identify specific, impactful problems signals that you’ll prioritize user value over feature checklists. Generally there are three components to address in this section: 1. **User journey mapping:** Create a detailed visualization of how your persona interacts with the product or experience, focusing on specific contexts and scenarios in their daily life rather than generic stages. 2. **Problem discovery and articulation:** Identify pain points at each stage of the journey where users struggle, face uncertainty, or experience frustration. Frame these problems with specificity about context, emotional impact, and user outcomes. Clearly distinguish between needs (desires) and problems (obstacles). 3. **Problem prioritization and mission connection:** Once you’ve identified meaningful problems (I recommend three to avoid burning too much time), prioritize one based on two key dimensions: severity (how much pain the problem causes when it occurs) and frequency (how often it happens for your target user). Then explicitly tie your prioritized problem back to your mission statement. The best candidates think beyond generic pre/during/post journeys and instead visualize a “day in the life” of their persona. Remember that needs are desires (“I want beautiful flowers in my apartment”), while problems are specific obstacles (“It’s challenging to find flowers that thrive in my apartment’s limited lighting conditions”). ### Elevating problem identification What separates good problem identification from exceptional problem identification: - **Rich journey context:** Create a detailed user journey that captures specific scenarios in your persona’s life rather than generic stages applicable to any product. - **Emotional specificity:** Articulate problems with precise context, emotional impact, and tangible consequences that make them feel authentic. - **Systems-thinking depth:** Connect surface problems to underlying causes, showing deeper understanding beyond visible symptoms and creating opportunities for more impactful solutions. For each of our example personas, here’s how we might map their user journeys to identify potential problems and prioritize one to solve. ### Examples of articulating user journeys and problems #### **Question 1: Tell me about a product you love and how you would improve it.** > *User journey for James (knowledge specialist):* > > - *Discovery and setup: Learns about Claude Projects, creates account/upgrades to Pro, understands project creation process* > - *Document collection and organization: Gathers relevant research papers and notes, organizes documents for upload, decides what to include/exclude* > - *Project creation: Creates new project, uploads documents, sets project instructions* > - *Knowledge testing and refinement: Tests project with initial queries, evaluates response quality, refines project content and instructions* > - *Regular usage: Asks specialized queries, explores connections across documents, uses assistants for research synthesis* > - *Maintenance and updates: Adds new research as it becomes available, updates project parameters as needs evolve, manages multiple projects over time* ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/3b785da7-ec59-4424-a8d0-b94e01c3a088_2913x2205.jpeg) > *Based on frequency and severity, I’d like to prioritize solving the **context fragmentation** problem, as it has high scores on both dimensions and directly impacts the core value proposition of Claude Projects for knowledge specialists.* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** > *User journey for Casey (novice urban gardener):* > > - *Inspiration: Sees friend’s plants or online content that sparks interest* > - *Research: Searches for what plants might work in her space* > - *Planning: Decides what to grow and what supplies are needed* > - *Purchasing: Buys plants, containers, soil, and basic tools* > - *Setup: Sets up her gardening space* > - *Care and maintenance: Daily/weekly care of plants* > - *Troubleshooting: Addressing issues like pests or plant health problems* > - *Harvest/enjoyment: Enjoying the results of her efforts* ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/63dff8dd-03ba-49d5-a507-c0e9073b8a5e_2913x2326.jpeg) > *Based on both high frequency and high severity, I’d like to prioritize solving the **knowledge matching gap** problem. It occurs at a critical early stage that determines whether Casey will even get started successfully with gardening.* #### **Question 3: Build a podcast product for Netflix.** > *User journey for Alex (educational-content extender):* > > - *Discovery: Finding relevant educational podcast content related to documentaries they’ve watched* > - *Evaluation: Understanding if the podcast will provide new insights beyond what they’ve already learned* > - *Consumption: Listening to the podcast and absorbing the information* > - *Retention: Remembering key insights and information from the podcast* > - *Application: Using or sharing what they’ve learned* > - *Connection: Finding related content to continue their learning journey* ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/1f500089-1ac1-409f-96de-6ca0588c0520_2913x2202.jpeg) > *Based on both high frequency and high severity, I’d like to prioritize solving the **learning continuity gap** problem. It directly addresses the core need of our educational-content extenders and represents a clear opportunity for Netflix to create value by bridging video and audio experiences.* ### Top pitfalls for problem identification Understanding what elevates problem identification is only part of the equation. Just as important is knowing what to avoid. Here are the most common pitfalls I’ve observed candidates step into when discussing problem identification: - **Confusing needs with problems:** Listing what users want (“need better search”) rather than specific pain points they experience (“users struggle to find relevant content because the current search doesn’t understand domain-specific terminology”). This signals a lack of understanding about root causes. - **Shallow journey mapping:** Creating a generic journey that could apply to any product (like pre/during/post), missing the opportunity to demonstrate an understanding of specific contexts and scenarios relevant to your persona. - **Problem-solution conflation:** Jumping to solutions while describing problems (“users need a better recommendation algorithm” rather than “users struggle to discover relevant content”). This indicates difficulty separating problem spaces from solution spaces. ### Practice tips for problem identification To strengthen your problem identification skills and deliver more compelling responses in your interviews, consider these targeted practice exercises: - **Journey visualization exercises:** Practice “day in the life” visualization for different personas by sketching out their typical day, hour by hour, identifying all touchpoints where they might interact with your product. This builds the mental muscle for creating detailed, context-rich user journeys. - **Problem-need differentiation drill:** Take 5 product features you use regularly and practice converting them from needs (“I want X”) to problems (“I struggle with Y because Z”). This helps train your brain to think in terms of pain points rather than solution features. - **Cross-journey problem mapping:** For practice questions, identify at least one problem in each major journey stage, forcing yourself to think broadly across the entire user experience rather than focusing on only one phase. After establishing your target audience through effective segmentation and picking a meaningful problem to solve, the next critical step is developing solutions that address this pain point. The transition from “what hurts them” to “how we can help” is where you showcase your creative problem-solving and product thinking. Let’s explore how to develop and prioritize solutions that will impress interviewers. ## Step 4: Solution development This is the part of the interview where you can showcase creative problem-solving while maintaining structured thinking. After identifying and prioritizing a specific problem for your target persona, you’ll now demonstrate your ability to develop meaningful solutions that address these pain points. Strong candidates approach solution development methodically rather than jumping to the first idea that comes to mind. Start by taking a minute or two to brainstorm different approaches to your selected problem. This demonstrates breadth of thinking and creativity, showing the interviewer that you can consider multiple angles before making decisions. Generally there are three components to address in this section: 1. **Solution brainstorming and prioritization:** Generate multiple distinct approaches to solving your prioritized problem, exploring different angles and mechanisms rather than variations on the same idea. Then evaluate solutions using an impact vs. effort framework to identify high-value opportunities, providing clear reasoning for your assessment. 2. **V1 definition and go-to-market:** Outline a concrete v1 implementation of your chosen solution with sufficient detail to demonstrate feasibility. Explain how users would discover and engage with your solution, considering integration within existing product experiences and initial distribution strategy. 3. **Risk assessment and mitigation:** Identify potential challenges with your solution and how you might mitigate them, demonstrating strategic foresight and thoughtful planning. When developing solutions, consider approaches that leverage existing platform strengths and company capabilities, draw inspiration from analogous problems in other domains, apply emerging technologies thoughtfully rather than gratuitously, and range from practical near-term implementations to ambitious long-term visions. The best candidates don’t just describe what they’d build but also articulate how users would discover and engage with the solution, how it integrates with existing product experiences, and what success might look like. Finally, connect your chosen solution back to the product mission you established at the beginning, creating a complete narrative arc that demonstrates coherent thinking throughout the interview. ### Elevating solution development To deliver a strong response in solution development, focus on these key differentiators: - **Creative divergence and defensibility:** Generate truly innovative approaches that would surprise interviewers with their originality while articulating what makes your solution difficult for competitors to replicate. - **Ecosystem integration:** Demonstrate how your solution creates network effects or virtuous cycles within the company’s broader product ecosystem, showing systems thinking beyond the immediate problem. - **User transformation vision:** Describe how your solution fundamentally changes the user’s experience from their current state, with a compelling roadmap for how it could evolve beyond v1 to create even greater impact. Based on the problems we prioritized for each of our examples, here are solution approaches that address these key pain points. ### Examples of articulating solutions #### **Question 1: Tell me about a product you love and how you would improve it.** ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/b7448939-ecf1-42d6-9067-d3163efc291d_2913x2261.jpeg) > *I’d prioritize **smart document synthesis** for the following reasons:* > > - *Has high impact with moderate effort* > - *Directly addresses the core problem by creating connective tissue between documents* > - *Leverages Claude’s existing strengths in summarization and understanding* > - *Provides a good balance of automation and user control* > - *Creates immediate value without requiring users to learn complex new interfaces* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/1d1b6986-16a3-440c-998c-e51a4ea3b187_2913x2332.jpeg) > *I’d prioritize the **plant match AI assistant** for the following reasons:* > > - *Has high impact with moderate effort* > - *It directly addresses the knowledge gap with personalized recommendations, while leveraging Meta’s existing AI capabilities to make implementation feasible* > - *This solution would fundamentally change how novice urban gardeners like Casey approach plant selection, moving from overwhelming generic research to personalized, confidence-building recommendations* #### **Question 3: Build a podcast product for Netflix.** ![Image from The definitive guide to mastering product sense interviews](https://substack-post-media.s3.amazonaws.com/public/images/e0cf5826-f0f8-4bc3-9c71-d325ec72a545_2913x2332.jpeg) > *I’d prioritize **documentary audio extensions** for the following reasons:* > > - *Has high impact with moderate effort* > - *Creates a clear, direct connection between video and audio content* > - *Leverages Netflix’s existing content investments and creator relationships* > - *Provides a differentiated podcast experience that competitors would struggle to replicate* After prioritizing our solution based on impact and effort considerations, the next step is to articulate v1 scope. A well-defined initial version demonstrates your ability to translate strategic thinking into actionable plans while balancing ambition with feasibility. If time permits, identifying potential risks and mitigations shows interviewers that you think holistically about product launches. ### Examples of articulating v1 risks #### **Question 1: Tell me about a product you love and how you would improve it.** > *When users upload multiple documents to a Project, Claude would automatically create three helpful resources:* > > 1. *A brief summary highlighting main themes across all documents* > 2. *Simple text outlines showing how key concepts connect between documents* > 3. *A list of potential knowledge gaps where information might be missing or contradictory* > > *Users can review and edit these summaries before continuing. We’d start with supporting academic papers in medical research, then expand to other fields over time.* > > *Risk: Automatically generated syntheses might miss important connections or create incorrect relationships.* > > *Mitigation: Include a human review step where users can edit and refine the synthesis before finalizing.* > > *Risk: Processing large document sets could be computationally expensive and slow.* > > *Mitigation: Start with reasonable document limits and implement background processing with status updates.* #### **Question 2: You’re a PM at Meta. Design a product for gardening.** > *We’d build a simple questionnaire within Instagram and Facebook that asks users about:* > > 1. *Their location and available space* > 2. *Light conditions and time available for plant care* > 3. *Experience level and personal preferences* > > *The AI would then recommend 3-5 plants perfectly suited to their situation, with care instructions, growth timelines, and photos. Users could save recommendations to a wish list and share with friends. We’d start with the 50 most common houseplants and container-friendly edibles.* > > *Risk: Inaccurate plant recommendations could lead to plant failure and user disappointment.* > > *Mitigation: Partner with gardening experts to verify recommendations and implement a feedback system where users report successes/failures.* > > *Risk: Low initial engagement if users don’t discover the feature.* > > *Mitigation: Leverage Meta’s feed algorithms to showcase gardening content to those who’ve expressed interest in related categories.* #### **Question 3: Build a podcast product for Netflix.** > *We’d select 5-10 of Netflix’s most popular documentaries and create 3-5 audio episodes (30 minutes each) for each one. These episodes would feature the original directors and experts going deeper into topics from the documentaries.* > > *The content would be available on established podcast platforms while maintaining Netflix branding, with clear connections to the original documentaries. Users could discover these through in-app promotions and post-viewing recommendations.* > > *Risk: Audio extensions might not maintain Netflix’s high quality standards.* > > *Mitigation: Involve original documentary teams in production and establish clear quality guidelines.* > > *Risk: Difficulty tracking cross-platform engagement between Netflix video and external podcast platforms.* > > *Mitigation: Create custom tracking links and offer exclusive bonus content that requires Netflix authentication.* ### Top pitfalls for solution development Understanding what elevates solution development is only part of the equation. Just as important is knowing what to avoid. Here are the most common pitfalls I’ve observed candidates step into when discussing solution development: - **Solution jumping:** Rushing to a single solution without exploring multiple approaches, signaling narrow thinking and potentially missing more effective alternatives. Strong candidates demonstrate breadth by considering diverse approaches to the same problem. - **Feature obsession:** Focusing on listing features rather than articulating how the solution addresses the core user problem, demonstrating product building rather than problem-solving. The best candidates connect features directly to user pain points. - **Platform mismatch:** Developing solutions that don’t leverage the company’s unique capabilities or ecosystem advantages, missing opportunities for differentiation and implementation efficiency. Effective solutions should strategically tap into the company’s strengths. ### Practice tips for solution development To strengthen your solution development skills and deliver more compelling responses in your interviews, consider these targeted practice exercises: - **Solution brainstorming drills:** Practice generating 3 meaningfully different solutions to the same problem within 5 minutes, focusing on approaches that tackle the problem from completely different angles rather than variations of the same idea. - **Company capability matching:** For each solution you brainstorm, explicitly connect it to specific strengths or assets of the company you’re interviewing with (e.g. their technology, user base, or distribution channels), ensuring that the solutions leverage unique competitive advantages. - **Impact-effort prioritization:** Create a simple 2x2 grid with axes for impact and effort, forcing yourself to place each solution in a specific quadrant with clear reasoning for your placement, then practicing how you would verbally explain this prioritization. After developing and prioritizing solutions that address your target audience’s problems, you’ve now completed the core framework for product sense interviews. Taking this structured approach from product motivation through solution development demonstrates to interviewers that you can think strategically about products while remaining focused on solving real user problems. To wrap up, we’ll summarize the key components of successful product sense interviews and provide additional resources to help you continue your preparation. ## Summarizing the product sense interview framework Mastering product sense interviews requires both a structured framework and deliberate practice. The framework I’ve outlined provides a comprehensive approach that allows you to showcase strategic thinking, user empathy, and creative problem-solving—key qualities that top tech companies look for in product managers. To recap, the core framework consists of five essential components: 1. **Clear communication (3-5 minutes):** Begin with well-defined assumptions, articulate your game plan, and engage with the interviewer to establish a clear structure for the conversation. 2. **Product motivation (3-5 minutes):** Establish why the product matters to users and the company, creating a mission statement that guides your decisions throughout the interview. 3. **Segmentation (8-10 minutes):** Identify ecosystem players, create meaningfully different user segments, and prioritize one segment by evaluating reach and underserved dimensions. 4. **Problem identification (8-10 minutes):** Map the user journey for your chosen persona, identify specific pain points they experience, and prioritize one problem based on frequency and severity. 5. **Solution development (8-10 minutes):** Generate diverse approaches to solving the prioritized problem, evaluate options based on impact and effort, and outline a concrete v1 implementation. Throughout the interview, remember these key communication principles: - Use “waypointing” to signal transitions between key sections. - Take strategic pauses before each section to gather your thoughts before checking in with the interviewer. - Create a consistent narrative thread from product motivation to v1, connecting decisions back to your mission statement. While frameworks provide necessary structure, successful candidates also know when to adapt. Pay attention to interviewer signals—if they ask, “Are there any other problems you might consider?,” that’s often a cue that they’d like to see more breadth in your thinking. Questions about your rationale for prioritization choices offer opportunities to clearly explain your thought process. This balance between structure and flexibility demonstrates the adaptability that companies value in product managers. Interviewers are evaluating not just your final answer but your entire approach. They look for candidates who can: - Think systematically about product ecosystems - Break down ambiguous problems with clear structure - Identify specific user needs based on behavior and context - Develop creative solutions that leverage company strengths - Communicate complex ideas with clarity - Respond thoughtfully to feedback and new information Product sense interviews can initially seem daunting, but with deliberate practice and a structured approach, you can consistently demonstrate the thinking that top companies are looking for. Remember that interviewers aren’t expecting perfection—they’re looking for a thoughtful approach, solid reasoning, and clear communication. ## 📚 Extra resources to prepare for product sense interviews The most effective preparation combines understanding the framework with extensive practice on real interview questions. No one “wings” these interviews successfully! Deliberate practice makes all the difference. **To help you master product sense interviews, I’ve created several resources:** - I’ll be hosting [a free 30-minute lightning lesson on April 16th](https://maven.com/p/d8e439/practice-for-product-sense-pm-interviews-with-ai) where I’ll show you how to use Claude Projects to practice for your product sense interviews. - Maven course: [PM Interview Bootcamp with AI Copilot: Product Sense & Analytical Thinking](https://bit.ly/4iQm5xq). (Use code “LENNYSLIST” to get $100 off. Enrollment closes April 22nd.) - 📺 [How to Ace Product Sense PM Interviews](https://maven.com/p/af5908/how-to-ace-product-sense-pm-interviews): a free lecture I published in 2024 (and [Product Sense Interview Template](https://docs.google.com/spreadsheets/u/0/d/1-jdZAVtLuk8dUnf9peUKlNkguSOuAk5VdMPTmgr9Kns/edit) to structure your practice) - [Recorded mock of PS response](https://grain.com/share/recording/be7899c9-f09e-4a18-93ec-ef40e7a26fac/Sex5ycmtyfFMOytY9K16yYQAx5p7lEn7X8D0fy2b): watch me apply the framework to a question about designing a VR product for elderly users ([here’s a completed template](https://docs.google.com/spreadsheets/d/1WV2kQNdM_kKHvr6Wsq0Znx7-r1gLK0qTOcuZg3LPqwo/edit?usp=sharing) for the above mock interview) - [Lewis Lin’s PM Question Bank](https://docs.google.com/spreadsheets/d/1rz10oEeLx-eGnilahKczYPhGfCUzIEKL-xRnjoQ-SX4/edit?gid=1024620532#gid=1024620532): the best PM interview question bank, updated daily - My [AI Practice Copilot for Product Sense & Analytical Thinking Interviews](https://www.benerez.com/copilot) *Thank you, Ben! Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [14/46] Beyond vibe checks: A PM’s complete guide to evals I’m going to keep this intro short because this post is so damn good, and so damn timely. Writing evals is quickly becoming a core skill for anyone building AI products (which will soon be everyone). Yet there’s very little specific advice on how to get good at it. Below you’ll find everything you need to understand wtf evals are, why they are so important, and how to master this emerging skill. Aman Khan runs a popular [course on evals developed with Andrew Ng](https://www.deeplearning.ai/short-courses/evaluating-ai-agents/), is Director of Product at Arize AI (a leading AI company), and has been a product leader at Spotify, Cruise, Zipline, and Apple. He was also a [past podcast guest](https://www.youtube.com/watch?v=E_rNotqs--I) and is launching his first [Maven course on AI product management](https://maven.com/aman-khan/thriving-as-an-ai-pm) this spring. If you’re looking to get more hands-on, definitely check out Aman’s upcoming free 30-minute lightning lesson on April 18th: [Mastering Evals as an AI Product Manager](https://maven.com/p/e55ece/mastering-evals-as-an-ai-product-manager). You can find Aman on [X](https://x.com/_amankhan), [LinkedIn](https://www.linkedin.com/in/amanberkeley/), and [Substack](https://aiproductplaybook.com/). Now, on to the post. . . ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/d984d483-4383-4b24-83cd-5dea71e48d05_2912x1925.jpeg) After years of building AI products, I’ve noticed something surprising: every PM building with generative AI obsesses over crafting better prompts and using the latest LLM, yet almost no one masters the hidden lever behind every exceptional AI product: evaluations. Evals are the only way you can break down each step in the system and measure *specifically* what impact an individual change might have on a product, giving you the data and confidence to take the right next step. Prompts may make headlines, but evals quietly decide whether your product thrives or dies. In fact, I’d argue that the ability to write great evals isn’t just important—it’s rapidly becoming the defining skill for AI PMs in 2025 and beyond. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/f097be41-cdc9-42be-bf6f-7221b653b871_2913x4511.jpeg) If you’re not actively building this muscle, you’re likely missing your biggest opportunity for impact-building AI products. Let me show you why. ## Why evals matter Let’s imagine you’re building a trip-planning AI agent for a travel-booking website. The idea: your users type in natural language requests like “*I want a relaxing weekend getaway near San Francisco for under $1,000*,” and the agent goes off to research the best flights, hotels, and local experiences tailored to their preferences. To build this agent, you’d typically start by selecting an LLM (e.g. GPT-4o, Claude, or Gemini) and then design prompts (specific instructions) that guide the LLM to interpret user requests and respond appropriately. Your first impulse might be to feed user questions into the LLM directly to get out responses one by one, as with a simple chatbot, before adding capabilities to turn it into a true “agent.” When you extend your LLM-plus-prompt by giving it access to external tools—like flight APIs, hotel databases, or mapping services—you allow it to execute tasks, retrieve information, and respond dynamically to user requests. At that point, your simple LLM-plus-prompt evolves into an AI agent, capable of handling complex, multi-step interactions with your users. For internal testing, you might experiment with common scenarios and manually verify that the outputs make sense. Everything seems great—until you launch. Suddenly, frustrated customers flood support because the agent booked them flights to San Diego instead of San Francisco. Yikes. How did this happen? And more importantly, how could you have caught and prevented this error earlier? This is where evals come in. ## What exactly are evals? Evals are how you measure the quality and effectiveness of your AI system. They act like regression tests or benchmarks, clearly defining what “good” actually looks like for your AI product beyond the kind of simple latency or pass/fail checks you’d usually use for software. Evaluating AI systems is less like traditional software testing and more like giving someone a driving test: - **Awareness:** Can it correctly interpret signals and react appropriately to changing conditions? - **Decision-making:** Does it reliably make the correct choices, even in unpredictable situations? - **Safety:** Can it consistently follow directions and arrive safely at the intended destination, without going off the rails? Just as you’d never let someone drive without passing their test, you shouldn’t let an AI product launch without passing thoughtful, intentional evals. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/72e84e39-b50f-48ab-9e4f-1669abe5094e_2913x1771.jpeg) Evals are analogous to unit testing in some ways, with important differences. Traditional software unit testing is like checking if a train stays on its tracks: straightforward, deterministic, clear pass/fail scenarios. Evals for LLM-based systems, on the other hand, can feel more like driving a car through a busy city. The environment is variable, and the system is non-deterministic. Unlike in traditional software testing, when you give the same prompt to an LLM multiple times, you might see slightly different responses—just like how drivers can behave differently in city traffic. With evals, you’re often dealing with more qualitative or open-ended metrics—like the relevance or coherence of the output—that might not fit neatly into a strict pass/fail testing model. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/2ca5e38a-c452-4017-a9be-002ce885a3bc_2913x2166.jpeg) ## **Getting started** #### Different eval approaches 1. **Human evals:** These are human feedback loops you can design into your product (i.e. showing a thumbs-up/thumbs-down or a comment box next to an LLM response, for your user to provide feedback). You can also have human labelers (i.e. subject-matter experts) provide their labels and feedback, and use this for aligning the application with human preferences via prompt optimization or fine-tuning a model (aka [reinforcement learning from human feedback](https://en.wikipedia.org/wiki/Reinforcement_learning_from_human_feedback), or RLHF). - **Pro:** Directly tied to the end user. - **Cons:** Very sparse (most people don’t hit that thumbs-up/thumbs-down), not a strong signal (what does a thumbs-up or -down mean?), and costly (if you want to hire human labelers). 2. **Code-based evals:** Utilizing checks on API calls or code generation (i.e. was the generated code “valid” and can it run?). - **Pros:** Cheap and fast to write this eval. Some examples include simple checks (i.e. is this text string present in the paragraph), to more complex logic/system checks. This approach is often cheaper and faster to write on a first pass, as AI coding agents improve (and code logic is often faster to execute), compared to LLM-as-judge, which still requires LLM inference and calibration. - **Cons:** Not a strong signal for subjective or open-ended tasks. 3. **LLM-based evals:** This technique utilizes an external LLM system (i.e. a “judge” LLM), with a prompt like the one above, to grade the output of the agent system. LLM-based evals allow you to generate classification labels in an automated way that resembles human-labeled data—without needing to have users or subject-matter experts label all of your data. - **Pro:** Highly scalable (it’s like a human labeling your data but much cheaper) and uses natural language, so PMs can write prompts directly. You can get the LLM to generate explanations for its judgments, making them more reliable and explainable for debugging. While individual judgments may be subjective, they become empirically useful over large datasets—if a human can grade something, so can an LLM. Production systems often use techniques like confidence scores or panels of LLM judges to increase reliability. - **Con:** Requires initial setup of the LLM-as-judge system with some labeled examples to validate performance. Results are probabilistic rather than deterministic, so you need sufficient volume to trust the signal. Importantly, LLM-based evals are natural language prompts themselves. That means that just as building intuition for your AI agent or LLM-based system requires prompting, evaluating that same system also requires you to describe what you want to catch. Let’s take the example from earlier: a trip-planning agent. In that system, there are a lot of things that can go wrong, and you can choose the right eval approach for each step in the system. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/0226b638-2757-485f-9ded-8a8114c7b1a6_2913x3146.jpeg) ### Standard eval criteria As a user, you want evals that are (1) specific, (2) battle-tested, and (3) test for specific areas of success. A few examples of common areas evals might look at are: 1. **[Hallucination](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals/hallucinations):** Is the agent accurately using the provided context, or is it making things up? 1. Useful for: When you are providing documents (e.g. PDFs) for the agent to perform reasoning on top of 2. **[Toxicity/tone](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals/toxicity):** Is the agent outputting harmful or undesirable language? 1. Useful for: End-user applications, to determine if users may be trying to exploit the system or the LLM is responding inappropriately 3. **[Overall correctness](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals/q-and-a-on-retrieved-data):** How well is the system performing at its primary goal? 1. Useful for: End-to-end effectiveness; for example, question-answering accuracy—how often is the agent actually correct at answering a question provided by a user? Other common areas for eval would be: - [Code generation](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals/code-generation-eval) - [Summarization quality](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals/summarization-eval) - [Retrieval relevance](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals/retrieval-rag-relevance) Phoenix (open source) maintains a repository of off-the-shelf evaluators [here](https://docs.arize.com/phoenix/evaluation/how-to-evals/running-pre-tested-evals).\* Ragas (open source) also maintains a repository of RAG-specific evaluators [here](https://docs.ragas.io/en/latest/concepts/metrics/available_metrics/). *\*Full disclosure: I’m a contributor to Phoenix, which is open source (there are other tools out there too for evals, like Ragas). I’d recommend people get started with something free/open source, which won’t hold their data hostage, to run evals! Many of the tools in the space are closed source. You never have to talk to Arize/our team to use Phoenix for evals.* ### The eval formula Each great LLM eval contains four distinct parts: - **Part 1:** **Setting the role.** You need to provide the judge-LLM a role (e.g. “you are examining written text”) so that the system is primed for the task. - **Part 2: Providing the context.** This is the data you will actually be sending to the LLM to grade. This will come from your application (i.e. the message chain, or the message generated from the agent LLM). - **Part 3: Providing the goal.** Clearly articulating what you want your judge-LLM to measure isn’t just a step in the process; it’s the difference between a mediocre AI and one that consistently delights users. Building these writing skills requires practice and attention. You need to clearly define what success and failure look like to the judge-LLM, translating nuanced user expectations into precise criteria your LLM judge can follow. What do you want the judge-LLM to measure? How would you articulate what a “good” or “bad” outcome is? - **Part 4: Defining the terminology and label.** Toxicity, for example, can mean different things in different contexts. You want to be specific here so the judge-LLM is “grounded” in the terminology you care about. Here’s a concrete example. Below is an example eval for toxicity/tone for your trip planner agent. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/9cda3b44-8466-4878-b18d-197fc702688b_2913x1703.jpeg) ### The workflow for writing effective evals Evals aren’t just a one-time check. Gathering data to evaluate, writing evals, analyzing the results, and integrating feedback from evals is an iterative workflow from initial development through continuous improvement after launch. Let’s use the trip planning agent example from earlier to illustrate the process for building an eval from scratch. #### **Phase 1: Collection** Let’s say you’ve launched your trip planning agent and are getting feedback from users. Here’s how you can use that feedback to build out a dataset for evaluation: 1. **Gather real user interactions:** [Capture real examples](https://hamel.dev/blog/posts/field-guide/index.html#measure-alignment-between-automated-evals-and-human-judgment) of how users engage with your app. You can do this via direct feedback, analytics, or manual inspection of interactions within your application. 1. For example: Capture human feedback (thumbs-up/down) from your users interacting with the agent. Try to build out a dataset representative of real-world examples that have human feedback. 2. If you don’t collect feedback from your application, you can also take a sample of data and have subject-matter experts (or even PMs!) label the data. 2. **Document edge cases:** Identify the unusual or unexpected ways users interact with your AI, as well as any atypical responses from the agent. 1. As you inspect specific examples, you might want a dataset that is balanced across topics. For example: - Help booking a hotel - Help booking a flight - Asking for support - Asking for trip planning advice 3. **Build a representative dataset:** Collect these interactions into a structured dataset, ideally annotated with “ground truth” (human labels) for accuracy. I’d recommend having between 10 and 100 examples with human labels to start with, as a rule of thumb, to use as ground truth for evaluation. Start simple—spreadsheets are great initially—but eventually consider open source tools like [Phoenix](https://phoenix.arize.com/) for logging and managing data efficiently. I’m biased—I helped build Phoenix, but only because I was struggling with this myself. My recommendation would be to use a tool that is open source and easy to use for logging your LLM application data and prompts when getting started. #### **Phase 2: First-pass evaluation** Now that you have a dataset consisting of real-world examples, you can start writing an eval to measure a specific metric, and test the eval against the dataset. For example: You might be trying to see if your agent ever answers in a tone that reads as unfriendly to the end user. Even if a user of your platform gives negative feedback, you may want your agent to respond in a friendly tone. 1. **Write initial eval prompts:** Clearly specify the scenarios you’re testing for, following the four-part formula above. 1. For example, the initial eval might look something like: - **Setting the role:** “You are a judge, evaluating written text.” - **Providing the context:** “Here is the text : {text}” → In this case, {text} is a variable, where you will be providing the “LLM agent answer” in the [variable of the prompt](https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/prompt-templates-and-variables). - **Providing the goal:** “Determine whether the LLM agent response was friendly.” - **Defining the terminology and label:** “‘Friendly’ would be defined as using an exclamation point in response and generally being helpful. The response should never have a negative tone.” 2. **Run evals against your dataset:** You will run the eval by sending the eval prompt plus LLM agent answer variable to an LLM, and get back a label for each row in your dataset. 1. Aim for at least 90% accuracy compared with your human-labeled ground truth. 3. **Identify patterns in failures:** Where does the eval fall short? Iterate on your prompt. 1. In the example below: The eval disagrees with the human label in the last example. Our prompt above requires an exclamation point for an LLM agent response to be considered friendly. Maybe that requirement is too strict? ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/6c0f6186-adc4-41f8-b367-cc184a2bd9eb_2913x1830.jpeg) #### **Phase 3: Iteration loop** 1. **Refine eval prompts:** Continuously adjust your prompts based on results until performance meets your standards. 1. Tip: You can add a few examples to your prompt of “good” and “bad” evals to ground the LLM response, as a form of “[few-shot prompting](https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/multishot-prompting).” 2. **Expand your dataset:** Regularly add new examples and edge cases to test whether your eval prompts can generalize effectively. 3. **Iterate on your agent prompt:** Evals help you test your product when you make changes to the underlying AI system—in some ways, they are the final boss when A/B testing prompts for your AI system. For example, when you make a change to an agent (e.g. changing the model from GPT-4o to Claude 3.7 Sonnet), you can rerun the dataset of questions you collected *through your updated agent* and evaluate the new output (i.e. Claude 3.7) with your eval agent. The goal would be to improve on your initial agent (GPT-4o) eval scores, giving you a benchmark you can use for continual improvement. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/fa97cceb-0e2a-478e-b4c3-1f4f49c1e85d_2913x2538.jpeg) #### **Phase 4: Production monitoring** 1. **Continuous evaluation:** Set up evals to run automatically on live user interactions. - For example: You can continuously run the “friendly” eval on all your incoming requests and agent responses, to get a score over time. This can help you answer questions such as “Are your users getting more frustrated over time?” or “Are the changes we are making to our system impacting how friendly our LLM is?” 2. **Compare eval results to actual user outcomes:** Look for discrepancies between eval results and real-world performance (i.e. human-labeled ground truth). Use these insights to enhance your eval framework and improve accuracy over time. 3. **Build actionable eval dashboards:** Evals can help communicate AI metrics to stakeholders across your team, and they can even be tied to business outcomes. They can serve as proxy leading metrics for changes you make to your system. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/8ed7a15e-79ee-492e-a17e-7c1e8092a4d7_1232x720.gif) #### **Common mistakes I’ve seen teams make when adopting evals:** 1. Making evals too complex too quickly can create “noisy” signals (and cause the team to lose trust in the approach). Focus on specific outputs rather than complex evaluations—you can always add sophistication later. 2. Not testing for edge cases. Provide one or two specific examples of “good” and “bad” evals as part of your prompt—few-shot prompting—for increased eval performance. This helps ground the judge-LLM in what is considered good or bad. 3. Forgetting to validate eval results against real user feedback. Remember that you’re not just testing code, you’re validating if your AI can truly solve user problems. Writing good evals forces you into the shoes of your user—they are how you catch “bad” scenarios and know what to improve on. ## **What’s next?** Now that you understand the fundamentals, here’s exactly how to start with evals in your own product: 1. **Pick one critical feature** of your AI product to evaluate. A common starting point is “hallucination detection” for a chatbot or agent that relies on documents or context you provide it with to answer questions. Try to tackle evaluating a well-defined component in your system before evaluating deeply internal logic. 2. **Write a simple eval** checking whether the LLM output correctly references provided content or if it invents (hallucinates) information. 3. **Run your eval** on 5 to 10 representative examples from real interactions that you have collected or created. 4. **Review the results and iterate**, refining the eval prompt until accuracy improves. For a detailed example of how to build a hallucination eval, check out our guide [here](https://arize.com/llm-evaluation), as well as our hands-on course [on Evaluating AI Agents](https://www.deeplearning.ai/short-courses/evaluating-ai-agents/). ## **Looking ahead** As AI products become more complex, the ability to write good evals will become increasingly crucial. Evals are not just about catching bugs; they help ensure that your AI system consistently delivers value and delights your users! Evals are a critical step in going from prototype to production with generative AI. ![Image from Beyond vibe checks: A PM’s complete guide to evals](https://substack-post-media.s3.amazonaws.com/public/images/f6668c69-2f8d-433d-90f0-475630a8884a_2913x2110.jpeg) I would love to hear from you: How are you currently evaluating your AI products? What challenges have you faced? Leave a comment👇 [Leave a comment](https://www.lennysnewsletter.com/p/beyond-vibe-checks-a-pms-complete/comments) ### **📚 Further study** 1. [A conversation with OpenAI’s CPO Kevin Weil, Anthropic’s CPO Mike Krieger, and Sarah Guo](https://www.youtube.com/watch?v=IxkvVZua28k&t=1s) 2. Peter Yang + Aman: [The AI Skill That Will Define Your PM Career in 2025 | Aman Khan (Arize)](https://www.youtube.com/watch?v=u8lEDw7pOkE) 3. [DeepLearning.ai course on evals](https://deeplearning.ai) with Andrew Ng 4. Prompt optimization guide: 5. Arize eval hub (free!): 6. Peter Yang + Scott White: [Exclusive: Inside the Best AI Model for Coding and Writing | Scott White (Anthropic)](https://www.youtube.com/watch?v=ynWgKSF0TZY) *Thank you, Aman!* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [15/46] A new perk for annual subscribers: A free year of some of the world's most beloved products (while supplies last) **Important: This original offer has now been replaced with a new offer—click the image below to learn more 👇** ### **You can disregard everything else you see below this line….** ![Image from A new perk for annual subscribers: A free year of some of the world's most](https://substack-post-media.s3.amazonaws.com/public/images/12bf3ded-1a8f-4b84-bf45-df79a8a53a07_1456x994.jpeg) When you subscribe to Lenny’s Newsletter, you’ll get one free year of a rotating list of incredible products (while supplies last): 1. **[Bolt](https://bolt.new/)** 2. **[Superhuman](https://superhuman.com)** 3. **[Linear](https://linear.app/)** 4. **[Perplexity Pro](https://www.perplexity.ai/pro)** 5. **[Granola](https://granola.ai/lenny)** *Sold out: Notion, Cursor, v0, Lovable, Replit.* These companies have never offered anything like this before. You’re getting a $15,000+ value for the price of a yearly newsletter subscription. Using even just one of these products offsets the cost of the newsletter. And we’ll be adding additional world-class products to the bundle over time. This bundle is in addition to the existing benefits of being a paid subscriber: full access to every new newsletter post, along with five years’ worth of previous posts, and invites to a thriving members-only Slack community and local community meetups. A yearly subscription is an absolute no-brainer. Click here to grab this deal 👇 #### **Important deal details:** 1. You must be a **new paying customer** of the products to take advantage of the free year. If you’ve already paid for one of the products in the bundle before, you won’t be able to get a free additional year of that specific product. 2. You must have an **annual subscription** to Lenny’s Newsletter to be eligible for this bundle. Monthly subscribers do not have access to the deal. 3. **Both existing subscribers and new subscribers are eligible**. 4. Your one free year of the product begins **when you activate the product with your unique code**, not when you purchase this newsletter subscription. 5. **To secure your codes, you must claim them using the link below**. We may run out of codes for certain products over time, and this deal may end. 6. If you request an early refund or chargeback for your subscription, **your bundle codes will deactivate** (whether they’re active or unused)**.** #### **How to redeem the deal:** 1. Once you [become a paid yearly subscriber](https://www.lennysnewsletter.com/subscribe?), [click here](https://lennysbundle.com/) to redeem your codes. 2. If you’re on a monthly plan, [upgrade to yearly](https://www.lennysnewsletter.com/subscribe?), then [go here](https://lennysbundle.com/) to redeem your codes. 3. If you’re already a paid annual subscriber, [click here](https://lennysbundle.com/) to redeem your codes. #### **Specifics of what you get:** 1. **[Granola](https://granola.ai/lenny)**: One year of the Business plan for you and your team—up to 100 seats ($10,000+ value) 2. **[Linear](https://linear.app/)**: One year of the Business plan—two seats ($336 value) 3. **[Superhuman](https://superhuman.com)**: One year of the Starter plan ($300 value) 4. **[Perplexity](https://www.perplexity.ai/pro)**: One year of the Pro plan ($240 value) 5. **[Bolt](https://bolt.new/)**: One year of the Pro plan ($240 value) *Sold out: Notion, Cursor, v0, Lovable, Replit.* **If you have any trouble or questions, email support@lennyrachitsky.com.** Sincerely, Lenny --- ## [16/46] Make product management fun again with AI agents Tal Raviv’s [last guest post](https://www.lennysnewsletter.com/p/product-manager-is-an-unfair-role) on working “unfairly” as a PM is my fourth most popular of all time, and [his podcast episode](https://www.youtube.com/watch?v=wFhurV1l6Jk) is a huge fan favorite. Now he’s back with a guide to using AI agents to make PMing fun again, which I predict will be in the top 5 most popular posts of all time. [Editor’s note: this prediction proved to be true]. *For more from Tal, check out his [63 free video tutorials](https://talraviv.co/) on using AI agents at work, and his other guest post on [building your AI thinking partner](https://www.lennysnewsletter.com/p/build-your-personal-ai-copilot). You can also book Tal for an [AI build sprint](https://talraviv.co/build-sprints) with your team.* ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/70119d0a-d718-435b-9de9-86f47bc13309_2912x2130.jpeg) What’s beautiful about product management is that everything is our job. What’s maddening about product management is that *everything is our job*. But while we’re busy with draining-yet-essential tasks (penning updates, wrangling meetings, syncing sources of truth, or acting as mission control), we’re displacing critical time to get up to speed on new tech, immerse ourselves in customer conversations, analyze data, build trust, and be thoughtful about the future: the important parts of our job. [Productivity hacks](https://www.lennysnewsletter.com/p/product-manager-is-an-unfair-role) and [cultivating self-reliant teams](https://youtu.be/wFhurV1l6Jk?si=YFpjMnVgEl262Oi5&t=1672) can help, but with tech orgs flattening, more of these types of repetitive but necessary tasks are falling on fewer product managers. Enter AI “agents.” Unlike chat-based LLMs, agents can listen to the real world, make basic decisions, and take action. In other words, they’re becoming talented enough to take on our least favorite, least impactful—but still necessary to do—PM tasks. If you’re like me, you’ve heard the promises and proclamations about how agents will reshape productivity, but your workday hasn’t changed at all yet. It’s not you—operationalizing AI agents for product work is hard. *Where to start? What tools? What about security? Costs? Risks? And why is there such a #$@% learning curve?* After interviewing founders of AI agent platforms, running numerous usability sessions with PMs building their first agents, and gathering insights from a [hands-on workshop](https://maven.com/p/50b3a6/build-6-pm-ai-agents-in-56-minutes) for over 5,000 product managers, I’ve compiled their collective wisdom. This post shares their insights on what works—and what doesn’t—in the real world. We’re first going to learn how to build an AI agent, hands-on. Then I’ll share a unified framework for any PM to plan their second (and third) agent. We’ll cover best practices, pitfalls, powers, and constraints. While AI agents aren’t magic genies, they can take a lot of repetitive, energy-draining PM tasks off our plate, let us focus on the most important work, and even make our jobs a bit more fun. ## What is an AI agent? ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/312f3fd5-95f0-404c-83fd-c0d76f4f56de_556x500.png) The term “AI agent” is admittedly fuzzy. Instead of debating names, it’s more useful to identify behaviors. Think of the term “agent” as a spectrum, where AI systems become “agentic” the more of the following behaviors they exhibit: 1. **Acts proactively**,as opposed to waiting to be prompted. 2. **Makes a plan**,as opposed to being given instructions. 3. **Leverages context**,accessing an internal knowledge base about your company and team, pulling the most up-to-date information regularly. 4. **Draws on live data** such as a web search or a support queue (as opposed to relying on static training or manually uploading a file). 5. **Takes real-world action.** Updates a CRM, runs code, or comments on a ticket, as opposed to only making recommendations. 6. **Creates its own feedback loop.** Watches its own output and iterates without human assistance. With new startups launching weekly, this framework helps me appreciate each product’s place in the landscape. Each row here is a product category, and each column is a useful behavior. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/7a2f4d73-bb94-4d79-ada4-2d7ac3b6b7c8_2913x3100.jpeg) Notice how each row checks some, but not all, boxes. AI systems are still very early, and each category approaches the opportunity from a different starting point. Of all the flavors and approaches of agents today, the category I refer to as “AI automations” is currently the most practical for helping product managers with monotonous busywork. This includes tools like Zapier, Lindy AI, Relay App, Gumloop, Cassidy AI, and so on. In this post, we’ll focus on this category of agents, but however you define AI agents, the important thing is that they can help us spend more time with our customers, give more attention to our teammates, build better products, and have more fun. ## Launch your first AI agent right now Let’s quickly launch an AI agent that preps you for a customer call, and let it run in the background while you read the rest of this article. *These instructions are for Zapier Agents (note: I have no affiliation with Zapier, I’m just a fan of this tool) and will take just 10 quick steps.* ### Ingredients 1. Zapier Agents 2. Google Calendar (or Outlook) 3. Slack (or MS Teams) ### Instructions 1. [Create a Zapier account](https://zapier.com/sign-up) and navigate to “Zapier Agents.” *Note that Zapier Agents is a new, separate product from the classic “Zaps” you may be familiar with.* ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/7b0fcb3c-a21f-47e7-8b1c-929fe4675c48_1600x899.png) 2. Select “Create a custom agent,” give it a name, and click “Start from scratch.” ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/57a1aab8-ba53-40c7-ab75-f1f40b8ae395_1600x911.png) 3. Click “Configure” at the top and “Create behavior.” ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/897706f4-16fc-4730-9588-d6100c5ac190_1600x892.png) 4. Set our automation to run every day at 8 a.m. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/e1dc45e2-982f-48d6-a1fd-5bcb8244ad5f_1600x920.png) 5. Paste the following prompt into the instructions field: `Look at all my calendar events for today [CALENDAR FIND MULTIPLE EVENTS], find all the external participants (people who have a different email address domain than mine), and do a web search for each of those email addresses. Summarize who these people are (where they work, what title they have, how long they’ve been in that role, information on past roles, and anything else you deem relevant) and send me a message in Slack.` `[SEND DIRECT MESSAGE] for each meeting with external participants (skip meetings with only internal participants) and tell me:` `- Name of meeting` `- Time of meeting` `- Name and relevant information of each external participant (formatted on their own line)` ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/cc9ffbdd-cfe4-4fba-9e27-5e6ce9c0ec23_1600x887.png) 6. Delete the placeholder text “[CALENDAR FIND MULTIPLE EVENTS]” and, with your cursor still in that spot, click “Insert tools.” ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/969961f5-1f89-40d9-8d55-b25940a95f69_1600x897.png) If you’re using Google Calendar, choose “Find Multiple Events.” ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/66b0f44a-a34e-4567-9fcb-160af1350828_1600x912.png) 7. Link your calendar software, and select your personal calendar. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/1c6085b1-6b30-42de-b4c8-efce5c9cea83_1600x1122.png) You should see the calendar block inline with your prompt: ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/9dc850b4-db7f-4aaa-a13f-b2aa9770d0af_1600x891.png) 8. Delete the placeholder text “[SEND DIRECT MESSAGE]” and, with your cursor still in that spot, use the “Insert tools” menu to connect your favorite messaging service. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/6431bf3b-93e2-4b62-8e5b-45fc0fb3934c_1600x933.png) I’m using Slack, and I’ll set it to be able to send me a direct message. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/2264812f-56a1-4d87-ab65-696d13bdc9c8_1600x938.png) *Constraining the agent to only DM frees us for worry-free experimentation.* 9. Woo! We’re ready to run. Click “Save instructions & test.” ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/faf854ce-0d25-4f08-8501-0628c8917a24_1600x891.png) You’ll get a message like this: ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/0a88ed9c-1cfc-458f-b6d9-6b36fbbb8f10_1600x738.png) 10. If you’re happy with this, go ahead and turn it on. (Don’t sweat this decision, since the only action it can take is privately DMing you.) ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/42e5b32f-189d-422b-bef6-e3db8b88ff66_1600x898.png) Congratulations. You’ve set up your first PM AI agent, now running in the background while you sit back and read Lenny’s Newsletter. ## Plan your second AI agent With our new agent running in the background, let’s unpack what we did . . . by planning our second AI agent. The first step is deciding what problem we want to solve. For this exercise, we’ll focus on opportunities that pre-AI automations couldn’t address. Ask yourself: What ongoing work requires some judgment and writing abilities—but not your full expertise and intuition? Put another way, if my company assigned me a junior intern, what would I have them do? Below are examples of use cases where product managers have gotten a lot of value from AI agents. (If any of them jump out at you, feel free to copy.) ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/c347d339-d5da-45d9-916f-ca1d8149b286_2912x4362.jpeg) *Note: Try to phrase this goal in one or two sentences, exactly the way you would in a Slack message to a junior intern.* I recommend choosing **ongoing tasks that arrive continually**. AI automations shine in one-at-a-time, repetitive tasks. In contrast, I don’t recommend designing for a big, one-time “batch” task (e.g. sifting through dozens of emails that already arrived). For batch tasks, consider working directly from an AI system: - Export the data and manually upload to Claude, Gemini, or ChatGPT. - Use a built-in tool like Slack AI, Notion AI, Gemini for Workspace, or Microsoft Copilot. - Connect directly to software apps with MCP integrations. ## Design how your agent is going to work Now that we’ve chosen what to delegate, let’s design how it’s going to work. Below is a checklist for planning an agent, regardless of what platform we’ll choose. Keep your chosen task in mind as you work through the list: **☑️ Do** ***I*** **understand this task?** **☑️ Could I start even smaller?** **☑️ Can I keep the downside low?** **☑️ Am I giving enough context?** **☑️ Am I staying close to raw customer signals?** ### 1. Do *I* understand this task? Just as when delegating to people, the fanciest AI will perform only as well as the instructions it’s given. Are you clear on how *you* would accomplish this task manually, with mouse, keyboard, and coffee? Do you know where the key information lives? Can you clearly put it in words? The first step is looking inward. As Max Brodeur-Urbas, the CEO of Gumloop, put it, “Understanding a problem should be the only prerequisite to solving it.” The best way to gain this clarity is to do the task once or twice. That way, you can provide examples of what success looks like. In our customer prep call example, think of the last time you prepared for a customer call. What sources did you intuitively consult? What information were you primarily looking for? (And what wasn’t relevant?) And if you’ve already been doing this task manually for a while, you’ve got this step covered. For example, when I watched a colleague set up a “weekly updates” agent, he already had a channel full of examples he could copy and paste as templates. ### 2. Could I start even smaller? Since AI agents evoke the image of a magical genie, it’s tempting to ask for all of our wishes at once. It’s more realistic to approach our agent like a new product, or a new process. As PMs, we know that to make both successful, we need to start small and cut scope. (Ironically, as PMs, it’s hard to cut scope when it’s for ourselves.) Ask yourself, what’s the *worst* part? The step you dread the most? Let’s start by delegating only *that*. We’ll do the rest of the steps manually first. If your dream is to monitor five competitor websites, first launch with *one*. In our customer prep call example, it would have been tempting to have it scan the web, our Slack, Gong, Zendesk, Mixpanel, and HubSpot. However, we launched it with one data source to keep it simple to start, which allows us to build from there. ### 3. Can I keep the downside low? Murphy’s Law is as true for AI as it is for people: *Anything that can go wrong, will go wrong.* To sleep well at night, let’s ensure that any mistakes don’t really cost anything. Don’t try to predict how much a model will hallucinate (it will) or if your workflow will get it right the first time (it won’t). Instead, design your agent in a way that gives you all the upside and caps the downside. Examples of keeping a low downside, with high benefit: 1. Instead of pinging a Slack channel → **Send me a DM that I can copy-paste** 2. Instead of sending an email → **Create a draft and star the thread for my review** 3. Instead of making a decision → **Make a recommendation** 4. Instead of modifying a document → **Append suggestions at the bottom** Keeping a low downside is also a matter of physically restricting access with permissions (and even more granular). This is where agent platforms that have hard access constraints can really shine, because those physically limit the AI system’s behavior. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/027db2e8-326f-40d4-a2e1-cac596863314_1328x1284.png) ### 4. Am I giving enough context? Your AI agent most likely doesn’t need an in-depth competitive-landscape presentation, the three-year vision, or the entire company org chart. It *probably* does need to know where to access the right data, get your guidance for making decisions, and understand how to identify people on your team. Setting AI up for success looks a lot like what it takes to set up a human for success. For example, if you wanted an intern to “scan all Slack messages for feature requests about two-factor authentication shared by someone on the customer success team” . . . well, they’d need a way to know *who is on the customer success team*. If you expect your agent to make a decision on prioritization, *share your prioritization framework*. ### 5. (Danger zone) Am I staying close to raw customer signals? It feels obvious that an AI model with more raw training data will have better judgment. Yet when it comes to our own brains as PMs, it’s very tempting to fill our days with AI summaries—and deprive our own brains of “training data.” If I use AI to summarize everything, I’m gonna quickly degrade my customer intuition. In other words, product managers: AI isn’t endangering your job, but *letting AI* *read for you* is. What’s the middle ground? Use AI to traverse, roam, navigate, cluster, and clean up large amounts of data, but don’t let it blur your vision with summaries. Stay in the weeds by insisting on exact quotes and direct links to the original support tickets, sales call snippets, and screen recordings. Max Brodeur-Urbas shares another strategy. If you’ve ever tried to glean feedback from a support thread, you know that not only is it a ton of scrolling but also that the key insights are often at the end. Gumloop uses AI to get insights from their helpbot chat threads. But instead of summarizing, the role of AI is to *reason about the root cause* to better classify: “We use AI to analyze each chat and ask, ‘What is this person struggling with? What is the main complaint?’ We take those thoughtful analyses of the thread and we create a report that references the original issue, so we can go back and look at the raw conversation.” Now that we know *what* we want to build, let’s decide *how* we’ll build it. ## Construct your agent with a prompt (platform-agnostic) The elephant in the room around AI agents isn’t hallucinations, security, or costs (though we’ll address those soon). It’s the learning curve. As product managers, the last thing we need is one more hard thing to figure out. Many AI agent platforms require building blocks, subroutines, systems thinking. It may not be “coding,” but it sure is programming. At the same time, we’ve been spoiled with [AI prototyping tools](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product) that let us build with natural language conversations. When I was running AI agent usability sessions with PMs, several of the PMs had the same clever thought: “Why don’t I build this with an AI prototyping tool instead?” The short answer is “tempting, but not recommended.” While API integrations aren’t Nobel Prize–worthy work, you *are* building a third-party app. That creates bureaucracy for (1) registering your app with each SaaS service and (2) obtaining IT permission internally. (I haven’t even gotten to compliance or security reviews.) And real talk: using AI prototyping tools for production apps is still a lot of maintenance, babysitting bugs, and regressions. Just imagine building (and maintaining) our simple customer prep call agent, with Slack and Calendar integration, by vibe coding. It would take a lot of attention to get it working reliably over time. So how can we enjoy (1) building in plain English and (2) the pre-built integrations and security of existing platforms? The answer, of course, is to let AI help. Tools like Cassidy, Relevance, and Zapier Agents (in contrast to Zaps) have begun to let you prompt in natural language. Gumloop has “Gummie,” a chatbot that will guide you. Manus impresses at making a plan but lacks key workplace integrations or the ability to listen to triggers. None of these feel as smooth as giving instructions to a person, so I’ve created a prompt that transforms your natural language into step-by-step guidance. (File this under “prompts I hope will become obsolete very soon.”) Paste the following into your favorite LLM tool that has web search abilities and, ideally, reasoning. I recommend **OpenAI’s o3 Deep Research** or **Perplexity Deep Research** for this prompt. ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/1063e6ab-9aa5-4d73-aea4-2b517d639a6f_1416x390.png) ### AI agent builder prompt `Below are my objectives for an AI agent workflow. You’re an expert explainer of how to build an AI agent that is excellent at explaining to newbies. I want to build an agent in either Relay App or Lindy AI or Zapier Agents (not Zaps) or Cassidy AI or Gumloop or Relevance AI.` `For each platform, using only their official documentation or tutorials or videos, create step-by-step, explicit, hand-holding walkthroughs for me on how to create it in each platform (each one separately, without combining platforms).` `Keep it as simple as possible. No recommendation should ever require coding. (If you have no choice but to recommend a direct API call or webhook, make it super-clear and explicit how to make that work.) Use only the minimum necessary access permissions to achieve the task.` `Don’t gloss over any step; assume I don’t know anything and need even the smallest steps spelled out for me. If a step requires an LLM prompt, write the prompt for me. Same for any query strings (e.g. Gmail search query or otherwise).` `When you suggest a feature or functionality, ensure that this ability truly exists in that platform! If the web results don’t directly support your recommendation, It’s OK to still recommend it, but note your uncertainty inline in the step itself, using a “🚨” emoji, and provide alternatives in case it doesn’t exist, marked by “♻️.”` `For each service, spell out for me any components I’d need to have in place outside the platform I choose. Also, highlight any risks why these instructions might not actually work and questions to ask myself before I get started to save me time.` `If there’s a better tool you recommend for this job (AI agent, automation, or other no-code solution), repeat this process for that tool. If you recognize this is a batch task and not a continuous task (i.e. one-time vs. trigger-driven), then suggest better ways to do this potentially with an LLM even if it’s a bit more manual effort.` `🌅 OPPORTUNITY` `[If your boss gave you a junior intern (smart, motivated, zero experience), what would you have them do? Why is it valuable and impactful?]` `🪖 INSTRUCTIONS FOR AI AGENT TO FOLLOW` `[Tell your new junior intern how you would go about it at the level of naming the services, clarifying decisions, etc. that a human would need to perform the task. Use the format ❝Whenever ______ happens, I want you to decide ______, based on this data _______ and/or using web research, compare it to previous data from last time, and then go ahead and do ________.”]` `Ping me before the final action by [DMing me/Create a draft/etc.].` The results of this prompt are sometimes delightful, sometimes imperfect. But even when I need to fill in gaps, I’ve found it to be a much-needed boost over the intimidating learning barrier. As one founder candidly put it, “Today our platform makes smart people feel dumb.” With this prompt, I can focus more on *what* I want done and a little less on the *how*. I used it to review NPS survey responses that arrived in a Slack channel and decide if any of them hint at a technical issue that warrants proactively creating a Zendesk ticket. Here I’ve implemented it in Lindy AI: ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/b103d0f9-348c-4d84-b6ef-b2db6814c717_1600x1074.png) Not every tool will have what you need, but there’s usually a creative workaround. In Cassidy, there wasn’t a Zendesk “create ticket,” so I had it post in a channel instead: ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/a61c538f-210a-481b-8b8c-a7f84af7a506_1600x894.png)![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/3d376246-6f14-4183-a0d5-970067803d34_1160x230.png) ## Choose a platform The best AI agent platform is the one that your company already uses and trusts. If the marketing team has already connected something like Zapier to something sensitive like HubSpot, then your best bet is leveraging that instead of trying to pick a whole new tool. Another important criterion is whether it supports the integrations you need. It’s worth checking if a platform both integrates with the service you want to use *and* if it supports the action (e.g. several platforms can read Zendesk tickets but not create them). All else being equal, I recommend choosing the tool that can achieve your goal in the fewest moving parts. How to figure that out quickly? You guessed it: AI. Continue the same thread above: `Based on the platforms above and the methods chosen, please create a platform comparison table on all of the critical aspects required to make them happen. Each column should be a different platform, and each row should be a different functionality relevant to the steps you outlined. Focus the rows on concrete functionalities rather than abstract concepts or subjective traits or guesses. In each table cell, cite web sources where possible. Use emojis sparingly to draw attention to critical differences.` `Finally, if I could choose only one platform, based only on your answers above for this specific use case only and no web results or marketing claims, which platform do you recommend I choose to implement this specific objective based on simplicity and likelihood of not needing technical skills, fastest to set up, and easiest learning curve?` ## Releasing your agent into the wild By now, you’ve chosen a platform and built your first draft of an agent. Let’s test it out and give our junior intern a quick win. Since we’ve designed our AI agent with low downside (see above), launching this is not a dramatic decision. It’s just a test in a bigger environment. ### Be forgiving and compassionate This is good advice for AI as much as for people. If AI is acting weird in 2025, it’s worth searching for bugs in *our* inputs. One PM in an AI agent usability session I was running noticed the agent was hallucinating sprint numbers in Slack updates. He traced it back to the fact that his examples had “Sprint 5” in the titles but his project management board didn’t mention “sprints” at all. (It was an easy fix to remove “Sprints” from the examples.) The same PM noticed the AI was mislabeling tasks as done. He realized his project management board only had an estimated “end date” but there was no “task status.” This was fine when everyone accessing the board met daily. The AI (reasonably) assumed that anything past that date was “done.” (Again, easy fix: he added a “done” checkbox, which instantly improved the results.) Before getting frustrated with AI, ask yourself if you threw it any curveballs, or shared [enough context](https://www.svpg.com/lead-with-context-not-control/) to make the right decision. ### Trust is gained in drops After a few cycles, you’ll start to get a sense for how much you trust the AI. If the results generally look good, consider increasing its scope and responsibility. Dare you allow it to send that draft without approval? Perhaps add that additional data source to the flow? If things aren’t going well, be candid with your feedback. Process, people, and product have one thing in common: they require iteration. (Even the smartest person in the world needs feedback to know if they’re heading in the right direction.) Some tools can accept the feedback and translate it directly to their own instructions: ![Image from Make product management fun again with AI agents](https://substack-post-media.s3.amazonaws.com/public/images/313b6eb8-ae53-4f85-af33-7a1eb1467077_1312x1276.png) For other tools, what you can do is go back to the original AI thread above, and clarify and update your original prompt instruction (using the ✏️ icon). Then re-run it with the clarification. Another way to think of this is, you’re building a product with a customer of one. It’s natural for things not to work on the first shot. Fortunately, when they do, you’ll be able to push a lot of your least favorite work off your plate—it’s worth it. ## Optimizing costs PMs’ biggest concern about AI agents isn’t something we saw in the *Terminator* series. It’s cost. This is understandable—we’re creating a proactive, continuous process that decides when to run. At the same time, many of these platforms price based on usage. (Btw, Skynet’s monthly invoice must’ve been huge.) Like any business, AI agent companies aspire to price based on value. And hopefully the *value* of AI agents will grow past that of a junior intern. Flo Crivello of Lindy: “Companies have millions and millions of dollars in payroll and then they spend a thousand dollars on an AI. It’s like, you actually want these lines to cross. You should be happy to spend on the AI agent platform, because every dollar you spend on the AI agent platform is saving you $10 that a human would have done otherwise.” Meanwhile, competition will drive prices closer to cost. And the *cost* to run foundation models is steadily dropping. Setting down the economics textbook, here’s some super-practical advice on cost no matter where things go, from Jacob Bank of Relay.app: “When you’re working with an AI step, there are two ways to reduce cost. You pick a cheaper model or you feed it less data. I typically recommend first getting the thing working with a good model and lots of data. [Then] if it’s not that frequent of a use case, I just don’t even bother with cost optimization. If it’s something that’s going to run 100 times a day and cost $10 each time, then those are the two levers I pull.” ## Security Security was already on everyone’s mind with original chat interfaces, so understandably, security is a consideration with AI agents too. One top concern is around AI absorbing and learning from company data. Fortunately, every AI system provider has an option not to train with queries. (Whether that’s good enough depends on your company policy.) Another concern is having third-party SaaS accessing company data. I recommend evaluating this risk according to the alternatives: 1. Are you using third-party automations already (regardless of AI)? If so, how much more risk does AI itself add to your workflow? 2. How many *human* employees have been granted those permissions? Keeping recent scandals in mind, much more risk does AI truly add to the picture? To quote Jacob Bank again: “People are way too casual about hiring employees and giving them lots of information, and way overly strict about SaaS products that have security best practices.” All that said, I’m not a security expert. Consult your local CISO. ## Where might agents evolve next? So far, we’ve built one AI agent and a framework for creating more. Now let’s imagine how these tools might evolve from eager interns to capable colleagues. ### Closing the feedback loop *without a human* I’ve run agents that made an intelligent decision but then proceeded to implement it like a blindfolded kid hitting a piñata. When humans make a silly mistake (overwrite the headers of a spreadsheet or append an unreadable, ugly block of text to a Google doc), it’s instantly obvious to us. AI agents don’t have this instinct, and it’s not always for lack of intelligence. A big barrier is *permissions*. **Companies are more comfortable giving “write access” than they are allowing apps to view company data.** This may feel counterintuitive, but it makes sense that letting a third party “create a contact in Salesforce” is a safer permission than “view all my existing contacts.” Same for posting in a Slack channel vs. reading the contents of a Slack channel. Where AI cannot close the loop, humans bear the steps of (1) observing an error and (2) communicating that error to the AI system. In the “build/test/learn” cycle, “build” is way stronger than the other two. Solutions won’t come from better AI models or tools but rather from creative solutions on how to modularize and orchestrate them. Replit’s Agent recently became able to take screenshots of its own virtual browser and analyze what it sees, which saves a lot of error reporting for me as a user. For Slack and Salesforce, I can imagine an internal trusted AI agent that sits inside the destination app, has access to the same instructions, and gives feedback to the third-party agent. However this gets solved, having agents view results and improve their own instructions (much as people improve their operational checklists) will lift a big burden off humans, and be a giant step forward for AI. ### AI copilots, AI prototyping, and AI agents will start to blend Remember how this post began with a chart of “What is an agent?”? Each row had its unique gaps. I expect the rows to close the gaps and fill up with checkmarks. I expect that high-context thought partners (AI copilots), natural language app builders (AI prototyping), and proactive decision-makers (AI agents) will all start to borrow elements from each other. It only makes sense that the AI tools of the future will have a ton of context on your company and role, connect to real-world inputs and outputs, and operate by natural conversation. ## Putting it all together AI agents aren’t magical genies (yet), but they represent a genuine opportunity to reclaim our energy, focus, and ultimately our joy in product work. Taking a step toward using AI agents in your own work is doable if you take the right steps. To recap: 1. **Start small:** Pick a task you know well, and scope it down. 2. **Design for limited downside:** Create safety nets for inevitable mistakes. 3. **Build with words:** Leverage AI to leapfrog the learning curve. 4. **Iterate with compassion:** Even the smartest colleague needs feedback. 5. **Slowly build trust:** Gradually increase scope and responsibility. The most exciting part about AI is where we invest the time and energy we reclaim. We can immerse ourselves in customer problems, build deeper relationships with our teams, and enjoy *creating*. (You know—all the reasons we chose PMing in the first place.) So implement your ~~first~~ second agent today. Then your third. Not only will you build product intuition for agentic AI (a skill every PM needs in 2025), but maybe, just maybe, product management will get a little more fun again. ## Special thanks - PMs - Ursula Sage - Amir Klein - Tony Privitelli - Eliran Zagbiv - Aviram Marom - Orr Weil - AI agent product leaders - Jacob Bank ([Relay App](https://www.relay.app/)) - Justin Fineberg and Jake Rosenthal ([Cassidy AI](https://www.cassidyai.com/)) - Max Brodeur-Urbas ([Gumloop](https://www.gumloop.com/)) - Flo Crivello ([Lindy AI](https://lindy.ai/)) - Bryce Vernon ([Zapier Agents](https://zapier.com/agents)) *Thank you, Tal! For more, check out his [AI PM course](https://bit.ly/42Mfr4j) and his upcoming free 45-minute lightning lesson, “[How AI PMs Slice Open Great AI Products](https://maven.com/p/b1789c/how-ai-p-ms-slice-open-great-ai-products),” on May 13th. You can find Tal on [LinkedIn](https://www.linkedin.com/in/talsraviv/?originalSubdomain=il) and [X](https://x.com/talraviv?lang=bn).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [17/46] The ultimate guide to negotiating your comp I’m excited to share the most practical and specific guide to negotiating comp that I’ve ever seen. This approach is targeted at leaders (e.g., senior ICs and managers), but no matter your role, anyone can benefit from this incredibly pragmatic, direct, and practical collection of tactics and advice. A bunch of my trusted friends have been raving about [Jacob Warwick](https://www.thinkwarwick.com/), a full-time negotiation coach for executives. Over the last 10+ years, he’s guided more than 3,500 senior leaders through high-stakes comp negotiations. This post is a synthesis of everything he’s learned over the past 15 years. You won’t find anything like this elsewhere. For more from Jacob, check out his [free LinkedIn profile analysis tool](https://www.producthunt.com/posts/free-linkedin-profile-review), which just went live on Product Hunt today, his [interview preparation report](https://www.interviewreports.com) service, and his individual consulting services at [ThinkWarwick Global](https://www.thinkwarwick.com). You can also find him on [LinkedIn](https://www.linkedin.com/in/jacobwarwick/) and [X](https://x.com/jacobwarwick). *P.S. If you prefer, you can listen to this post in podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads)* ![Image from The ultimate guide to negotiating your comp](https://substack-post-media.s3.amazonaws.com/public/images/9b0b742f-6619-487f-8fe1-b12104d3513f_2912x1923.jpeg) Two equally qualified product leaders walk into the same negotiation. One leaves with a standard package—market-rate base salary, typical equity, standard benefits. The other walks away with double the compensation, accelerated vesting schedules, and a signing bonus that covers their mortgage for a year. The difference isn’t skill, experience, or even negotiation tactics. It’s a fundamental shift in approach. As a product leader, you already understand that successful products solve important problems for people, create emotional connection with their users, and meet real market needs. Exceptional negotiation works the same way. While most candidates ask, “What can you offer me?,” top earners demonstrate “Here’s how I’ll solve your most pressing challenges and create new possibilities for your business.” Demonstrating unmistakable value for the company you’re interviewing with, and building relationships with the people who matter, will create opportunities others never see and transform your compensation trajectory. This shift isn’t semantic—it fundamentally transforms how decision-makers perceive your value. When you make them feel confident, inspired, and excited about the future you’ll build together, compensation becomes a natural reflection of that value, not a negotiation point. Recently, I guided a Senior PM whose offer started with an advertised range of $185K-$285K. After our targeted interview and negotiation strategy, they walked away with a staggering $1.1M annual package—nearly four times the upper limit. This isn’t an isolated success. I’ve helped hundreds of leaders dramatically increase their compensation by refusing to accept published salary ranges as the final word. I’ll show you exactly how they do it. ## **Your negotiation playbook** Every successful negotiation starts with leverage. Whether you’re planning six months ahead or sitting in discussions right now, here’s the process I’ve developed through trial and error with clients over 15 years. ![Image from The ultimate guide to negotiating your comp](https://substack-post-media.s3.amazonaws.com/public/images/41507665-96a6-4da8-ba3e-f1dcdc8a4b4c_2912x1552.jpeg) **G: Gather intelligence.** Go beyond the obvious. Dig into the company’s real challenges, understand who truly makes decisions (hint: it’s not always on the org chart), and know their market better than they do. **A: Align with their needs.** Stop selling your resume. Start demonstrating how you’ll solve their specific problems for the company/team. When you position yourself as the solution to their challenges—not just another candidate—the power dynamic shifts immediately. **I: Influence key stakeholders.** Create champions throughout the organization, not just with the hiring manager. Show each stakeholder how you’ll make their world better, and they’ll fight for your compensation later. **N: Navigate complexity.** Master the delicate dance of pushing for what you’re worth without creating tension. Know exactly when to advance discussions and when to build relationships. Timing is everything. **S: Secure your value.** Get agreements right, start delivering value before day one, and build the foundation for your long-term success. Let’s break down exactly how to execute each step with real examples from leaders I’ve worked with who have used this approach to dramatically increase their compensation. ## **G: Gather intelligence that others miss** The most valuable information won’t show up in press releases or job descriptions. To build real leverage, spend time on three key intelligence domains: ### **1. Organization dynamics** Forget the org chart—real power flows through history, unspoken alliances, and relationships. **Approach:** - Identify who gets consulted before decisions are made (often not who you’d expect) - Learn which past failures still haunt leadership thinking - Discover which rising stars have the CEO’s ear - Uncover the true drivers that aren’t discussed openly **Real-world impact:** I watched a brilliant SVP candidate lose an $800K opportunity by making one critical mistake. They treated the Chief of Staff meeting as a formality while focusing all their energy on impressing the CEO. They never realized this “assistant” whispered in the CEO’s ear about every executive hire. Meanwhile, the candidate who got the job built genuine rapport with both, understanding the real power dynamic at play. **How to execute this:** Before any interview, ask your network, “Who really influences decisions at this company?” and “Whose opinion does the leadership team value most?” The answers might surprise you. During the interview, ask questions such as: - How are decisions typically made in this organization? - Who are the key people I will collaborate with? - What’s the history behind this position? Is it new or am I replacing someone? - How can I best show up for you? And how can I best show up for [name other team member(s)]? ### **2. Decision-maker psychology** ![Image from The ultimate guide to negotiating your comp](https://substack-post-media.s3.amazonaws.com/public/images/2371f48b-8273-4ac3-a41f-b87b29a15ffd_1456x888.jpeg) Decisions are made by people. Understand what drives them. **Approach:** - Research what motivates each key player both professionally and personally - Identify formative experiences that shaped their leadership philosophy - Discover what problems truly keep them awake at night - Learn how they measure success beyond the obvious metrics **Real-world impact:** My product-executive client secured a $1.8M package after we discovered that their future CEO consistently bet on leaders who had overcome significant failures. While other candidates showcased perfect track records, we strategically shared growth from setbacks—creating an immediate connection that moved beyond their credentials. **Script that worked:** When speaking with the CEO, my client shared, “That product launch failure taught me three lessons about market timing that completely transformed my approach. I’m curious—what experiences have most shaped your philosophy toward product leadership?” This led to an engaging brainstorming session with the CEO about product vision. Consider the following starting questions: - What tough lessons has the company learned that changed how you do things now? - Why did you decide to hire for this role right now vs. waiting? - Can you tell me about a time when things didn’t go as planned and how the team handled it? ### **3. Strategic pain points** In order to speak directly to how you can add real value to the company, figure out which divisions are missing targets, which roles have been open too long, and what shifts are being debated behind closed doors—and how you would solve those problems. **Approach:** - Track where the company is headed, not just where it’s been - Identify problems they’ll face six months from now - Understand which initiatives have underdelivered and why - Use your experience to form an opinion about what must be done **Real-world example:** A client targeting a VP of Product Design role used this approach perfectly. Instead of relying on her impressive portfolio alone, she dug deep into the company’s challenges through industry contacts. She discovered that while they consistently delivered on their roadmap, their products weren’t creating customer enthusiasm. Rather than focusing on her credentials in interviews, she “sold the vacation”—painting a vivid picture of their organization after her design transformation. She detailed exactly how she would restructure product development, rebuild customer experience, and deliver measurable outcomes. The leadership team bought into the future she helped them envision. While other candidates matched the job description, she addressed their unspoken needs and secured a $1.2M package—double the stated range that topped out at $600K. ### **Gathering self-assessment** Before your next opportunity, assess your intelligence-gathering with these questions: - **Organizational network analysis:** Can you name at least three influential people not on the leadership team? If not, your research is still surface-level. - **Decision-maker insight:** For each key stakeholder, can you identify their professional background, leadership philosophy, and one challenge they’re currently facing? [If gaps exist, deepen your research](https://www.execsandthecity.com/the-elite-executives-guide-to-interview-preparation/). - **Information-source diversity:** Review your research sources. If they’re limited to the company website, press releases, and LinkedIn, you’re missing crucial intelligence. Expand to industry forums, customer reviews, and first-degree connections. - **Strategic pain-point clarity:** Can you articulate the company’s three biggest challenges in their own language? If not, you haven’t yet uncovered what truly matters to them. - **Competitive-position knowledge:** Do you understand how they measure success against competitors? Can you name which metrics they’re winning on and which they’re trailing? If not, your market context remains incomplete. The difference between adequate preparation and strategic intelligence often comes down to this: Are you gathering facts anyone could find, or are you uncovering insights that create a genuine competitive advantage? ## **A: Align with advocates** Now that you know the importance of collecting information, it’s time to use that intel to align with key people at the company, rather than impressing them. This isn’t semantics—it completely transforms the power dynamic. Instead of proving yourself worthy, you’re evaluating mutual fit and demonstrating understanding. I recently helped a product leader transform a Senior Director role with a range of $143K-$285K into a $1.1M package by shifting one fundamental approach. Instead of selling his credentials, we engaged the leadership team in discussions about how investing in strong product leadership would position their company as an industry innovator. By focusing on their aspiration to be visionary, we changed how they perceived the role’s value, leading them to quadruple their initial compensation range. Consider these contrasting approaches: **Weak:** “I have 15 years of experience leading product teams and consistently exceeding targets. . .” **Strong:** “Your challenges with product-market fit remind me of a similar situation at [Company]. We discovered that the root cause wasn’t our process but our customer development approach. I’m curious, how aligned are your product and research teams on an ideal customer profile?” The weak response positions you as an applicant listing credentials. The strong response establishes you as a peer exploring solutions together. This transformation happens through three critical elements. ### **1. Storytelling** Forget polished success stories and rehearsed scripts. Focus on raw, relevant experiences that create immediate connection. **Approach:** - Connect your past challenges directly to their current problems - Share the messy middle of your experiences, not just outcomes - Focus on lessons learned that directly apply to their situation - Use stories as bridges to solutions, not as evidence of your greatness **Real-world impact:** The aforementioned product executive client landed their new role by sharing a story about a failed product launch—not to showcase resilience but to demonstrate how that experience shaped their approach to due diligence, which mapped directly to their prospective employer’s current challenges. The company recently went through a few failures of its own, and my client knew the CEO was motivated to avoid another mistake. They framed their experience learning from past missteps to ensure alignment with the CEO’s own goals. **Script that worked:** When building alignment with the CEO, my client used the following frame: “That failed launch taught me to question three assumptions we never validated. Looking at your current roadmap, I’m concerned you might be facing similar blind spots with your enterprise expansion. Have you tested the assumption that enterprise users need all these features?” Don’t be afraid to ask challenging questions to assess how well aligned you are with leadership. For example: - What’s a problem your team is stuck on right now that keeps coming back? - If you could solve one thing that’s broken in the company, what would it be? - What’s something you wish candidates would understand about this role that most don’t get? ### **2. Child-like curiosity** ![Image from The ultimate guide to negotiating your comp](https://substack-post-media.s3.amazonaws.com/public/images/3e91a67c-7e5c-463a-811a-36d2d16d0169_1456x888.jpeg) Big, bold questions that others may tiptoe around can demonstrate both competence and humility. **Approach:** - Ask questions that reveal your strategic thinking - Challenge assumptions respectfully - Inquire about root causes rather than symptoms - Demonstrate you’ve done your homework through the specificity of your questions **Real-world impact:** A veteran product leader with 20 years of experience spanning well-known startups and Fortune 500 tech approached an interview differently. Though a VP by title, she applied for a Director role but thoroughly researched the company’s products, analyzed growth patterns, tracked leadership changes, and connected with former employees before meeting the CEO. Rather than simply trying to impress, she entered with a genuine evaluation mindset—was she the right fit for this company’s challenges? This balanced approach gave her the confidence to ask tough, insightful questions that ultimately led to the role being upleveled from Director to VP, with compensation doubling from $600K to $1.2M. She flipped the power dynamic by asking the CEO questions no other candidate dared: “How much of the team’s failure to grow do you take ownership of, what are you doing to remedy that, and how can I best show up for you to ensure it never happens again?” That conversation led to a leadership role created specifically for her at double her initial target. **Follow-up technique:** After asking a bold question, resist the urge to fill silence. Let them process and respond. The quality of their answers will tell you volumes about their self-awareness and leadership style. ### 3. “**Sell the vacation”** Engage in real-time thinking about the company or team’s future challenges to demonstrate that you already understand how the company could succeed. **Approach:** - Paint a vivid picture of what success looks like after you join - Get specific about the changes you’d implement in the first 30, 90, and 180 days - Connect your vision to metrics they already care about - Make them feel the relief of having their problems solved **Real-world impact:** One product leader transformed a panel interview by sketching how the company’s product organization would function six months after his arrival—complete with restructured processes and improved metrics they’d achieve together. He didn’t just answer questions; he invited the leadership team to mentally experience working with him. Instead of saying, “I could help,” he painted the picture of a new reality that would exist when they solved current challenges together. **Technique in action:** Here’s how he sold the vacation: “Based on what you’ve shared about your roadmap challenges, here’s how we would restructure your process in the first quarter. We’d start by implementing a two-track system that separates innovation from maintenance, then create a weighted scoring model that aligns with your business objectives. By month three, we’d see a 30% reduction in scope creep and a clearer connection between product work and revenue growth.” ### **Alignment self-assessment** Before your next conversation, review these warning signs that you’re still in “applicant mode”: - **Talk-to-listen ratio:** Record your next mock interview. Are you speaking more than 60% of the time? Scale back and ask more questions. - **Past vs. future focus:** Review your prepared examples. For every achievement you share, add a specific connection to their current challenges. - **Authenticity test:** Ask a trusted colleague, “Do my interview responses sound like me at my best, or like a polished candidate?” If it’s the latter, rewrite them in your natural voice. - **Connection signals:** After your next interview, note whether the conversation stayed formal or evolved into collaborative problem-solving. If it remained formal, you likely stayed in “impress mode.” - **Power position check:** Did you find yourself waiting for their approval or were you evaluating them as much as they evaluated you? Equal evaluation indicates partnership. The difference between a standard interview and a transformative opportunity often comes down to this: Are you trying to impress because you don’t think you belong in the room—or are you creating alignment because you know you belong there? ## **I: Influence momentum by adding value** Successful negotiation requires more than convincing just the hiring manager—you need to understand how power flows through organizations and build champions across functions and levels. Remember that **influence isn’t about manipulation but about creating real value in every interaction**. Here’s how to master three critical elements of influence: story selection, relationship building, and value demonstration. ### **1. Story selection** Your stories must resonate with everyone involved. Tailor them to each person’s priorities while demonstrating both functional expertise and broad business understanding. **Approach:** - Customize your examples for each stakeholder’s role and priorities - Reference other people in the room to create connections - Bridge functional expertise with business impact - Make technical concepts accessible to non-technical stakeholders **Real-world impact:** When discussing a product-market fit challenge in a panel interview, one client used this approach: “As product leaders, we often try to solve problems with feature development—and I’m glad Kara is on the call today because she’ll immediately recognize this.” This simple acknowledgment transformed the dynamic from a one-to-many interview into a collaborative discussion where each person felt seen and valued. **Technique in action:** Before any group interview, map each attendee’s role, likely priorities, and how your experience connects to their world. Prepare at least one tailored story or question for each person. **Pro tip:** People often [view your LinkedIn profile before interviews](https://www.execsandthecity.com/leaders-share-selectively-linkedin/). Keep it simple and focused instead of sharing too much. This helps you control your story during interviews and tailor what you share based on who you’re talking to. Your profile should work for you, not box you in. ### **2. Relationship building** Every interaction is an opportunity to build support and rapport. **Approach:** - Identify potential blockers early and transform them into advocates - Ask for input rather than trying to impress - Follow up individually with key stakeholders after group meetings - Create connection through shared professional challenges **Real-world impact:** One client turned a potential blocker into her strongest champion by understanding his concerns about the product roadmap and incorporating his insights into her strategy. She didn’t just listen—she made him part of the solution by asking, “As we solve this together, what will we tackle next?” That relationship proved crucial later when compensation discussions hit a snag. This once-skeptical stakeholder advocated for her value, helping secure a 40% increase over the initial offer. **Relationship-building script:** “I appreciated your perspective on [specific point they raised]. It’s made me think differently about [aspects of the role]. Could I get your input on how you’d approach [related challenge]? Your experience with [their background] would be incredibly valuable.” ### **3. Value demonstration** Instead of telling people how great you are or asking for what’s next, show them how you’ll improve their work. **Approach:** - Deliver small pieces of value before you’re hired - Share relevant insights that connect to their current challenges - Follow up with thoughtful analysis rather than generic thank-you notes - Create momentum through continuous engagement **Real-world impact:** After an initial interview, one product leader I worked with noticed a UX issue on the company’s website. Instead of just mentioning it, he created a quick prototype showing an alternative approach and shared it in his follow-up email. This unexpected value demonstration led to an accelerated hiring process and a 30% higher offer than initially discussed. **Follow-up template that works:** “Thank you for our conversation about [specific challenge they mentioned]. I’ve been thinking about your approach to [topic], and I wanted to share an insight from my experience with [relevant situation]. [Brief valuable insight.] I’m already preparing a few ideas for our next chat. I’ll keep you informed as I speak with others on the team. Are you open to debriefing next Thursday after 2 p.m. PT?” ### **Influence self-assessment** Before your next interview process, evaluate your influence strategy with these questions: - **Stakeholder mapping:** Have you identified every decision-maker and influencer in the process? Can you name what each cares about most professionally? - **Value creation audit:** Review your last three follow-up communications. Did each one provide specific value or insights relevant to their business, or were they generic updates? - **Champion development:** In your current process, have you converted at least one person from neutral to advocate? If not, who has the potential to become your internal champion? - **Cross-functional appeal:** Can your examples and stories resonate with people in Product, Engineering, Marketing, Sales, and Executive leadership? If they’re too functionally specific, broaden your narratives. - **Momentum check:** Does each interaction build upon the previous one, or are you starting fresh each time? If conversations feel disconnected, create clearer continuity between touchpoints. The difference between landing a role and creating a groundswell of support comes down to this: Are you trying to get through the interview process, or are you building momentum by providing value that makes you the obvious choice? ## **N: Navigate complexity** Negotiation is not a single conversation about compensation but rather an intricate interplay of influence, timing, and strategic pressure. You’ll always encounter complexity in negotiations. Better to anticipate it than be surprised. I’ve watched talented product leaders destroy multimillion dollar deals by rushing to close, compromising too quickly, missing signals, and avoiding uncomfortable conversations. Here’s how to master three critical elements of navigating complexity simultaneously. ### 1. **Manage the power dynamic** Success is about creating mutual value while steadily building leverage. **Approach:** - Signal excitement while creating space for discussion - Use questions instead of demands - Focus on mutual value, not just your compensation - Keep conversations positive even when pushing for more **Real-world impact:** A product executive client received what seemed like a strong offer. Instead of accepting immediately, we responded: “I’m excited about the opportunity and ready to get started. I’d like to understand more about how you see the role evolving as we grow—and how we might make this deal better for both of us. I’m available for coffee early next week. Work for you?” This response maintained momentum and signaled interest while creating space for value creation. We ultimately secured a package 35% above the initial offer. **Power-dynamic starter phrases:** 1. **For showing enthusiasm while creating space:**“I’m genuinely excited about this opportunity and can see myself making an impact. I’d love to discuss a few details to make sure we’re aligned on expectations.” 2. **For turning demands into questions:**“What’s the chance that we could explore a slightly different structure for the compensation package?” 3. **For focusing on mutual value:**“I’m thinking about how I can deliver the most value in this role. Would it make sense to discuss how the compensation structure could align with those key objectives?” 4. **For keeping negotiations positive:**“I appreciate the strong offer. I’m wondering if we could find a way to bridge the gap between this and what would make it a complete win for both of us.” 5. **For creating space without losing momentum:**“This looks promising, and I’m eager to move forward. Could we set up a quick call to discuss a few adjustments that might work better for everyone involved?” **Why this works:** These phrases transform potential confrontation into collaborative problem-solving. They maintain enthusiasm while creating the psychological space needed for negotiation. By framing requests as questions that explore mutual benefit, you shift the conversation from adversarial to partnership-focused, often resulting in significantly improved offers without damaging relationships. ### 2. **Practice strategic patience** Rushed negotiations almost always leave money on the table. **Haste equals risk.** **Approach:** - Resist the urge to accept the first good offer - Create a competitive dynamic without being manipulative - Be transparent about your timeline when appropriate - Never let artificial deadlines force decisions **Real-world impact:** A product executive was negotiating with three companies simultaneously. Instead of jumping at the first solid offer, he said: “I want to be transparent—I have a few opportunities I’m considering right now. I’m most excited about our discussions and want to ensure we find the right fit for both. Let’s align on a timeline that works for everyone.” This approach kept all options alive while building trust through transparency. The result? A final package with a multi-seven-figure signing bonus to convince my client to choose them over the competition. **Timeline management script:** “I’m making a significant career decision and want to be thoughtful. I have a few conversations in progress, but I’m particularly excited about this opportunity. I expect to be ready to make a decision by [specific date]. Does that timeline work with your hiring process?” ### **3. Master stakeholder management** Negotiations involve multiple stakeholders with different priorities and powers. While HR must follow company guidelines, hiring managers can often bend rules for candidates they truly want. Your job is to identify and negotiate with whoever has the real decision-making authority. **Approach:** - Map each stakeholder’s priorities and concerns - Tailor your communication to each person’s perspective - Keep everyone appropriately informed without overplaying your hand - Build support across the organization, not just with the decision-maker **Real-world impact:** One product leader negotiating a CPO role at a software company masterfully managed five stakeholders: the CEO, COO, board chair, HR head, and external recruiter. Rather than viewing this as a simple two-party negotiation, she recognized each stakeholder had different priorities: - CEO cared about product vision and strategic alignment - COO focused on team integration and operational fit - Board chair prioritized growth metrics and ROI - HR head needed to maintain compensation precedents - Recruiter wanted to close quickly for their fee We navigated this complex situation by: - Having targeted conversations with each stakeholder about their specific concerns - Sharing relevant updates with each party without revealing confidential details from other conversations - Demonstrating market value through competing interest without appearing solely money-motivated - Building genuine relationships focused on future collaboration, not just negotiations - Maintaining a patient, strategic approach even when pressured to decide quickly This approach resulted in a package exceeding the company’s initial band by 40% and included severance provisions that insulated her risk in joining. **Stakeholder management technique:** Create a simple matrix listing each stakeholder, what they care about most, what objections they might raise, and how your value proposition addresses their specific concerns. Reference this before every conversation to stay strategic. ### **Navigation self-assessment** Before your next negotiation, evaluate your strategy with these questions: - **Patience quotient:** In your last negotiation, how many days passed between receiving the offer and accepting it? If less than three to five business days, you likely moved too quickly. - **Leverage inventory:** Can you name at least three specific forms of leverage you currently hold? If not, you need to build more before advancing discussions. - **Stakeholder map:** Have you identified all decision-makers in your current process and what each cares about most? Missing stakeholders means missing opportunities to build support. - **Power position:** In your recent interactions, did you find yourself apologizing, over-explaining, or accepting constraints too quickly? These are signs of negotiating from weakness rather than partnership. - **Timeline control:** Who’s driving the pace of your current process—you or them? If you’re not actively managing the timeline, you’re likely missing opportunities to build leverage. The difference between good and extraordinary outcomes often comes down to whether you’re improvising or planning multiple moves ahead. Here’s a short video I made specifically for this post, sharing six negotiation techniques that transform high-pressure conversations into successful outcomes: [Watch on YouTube](https://www.youtube.com/watch?v=o76AQ57L2eA) ## **S: Secure your success** The secure stage isn’t just about getting the offer—it’s about setting yourself up for continued success. Product leaders on average have a shorter tenure (2.6 years) than typical vesting schedules (4 years). While not always preventable, I suspect you aim to stick around long enough to maximize your compensation—and with successful products to show for it. I’ve watched too many candidates fumble at the finish line because they focus so intently on immediate gains that they compromise future wealth opportunities. Here’s how to master three critical elements that ensure long-term success. ### **1. Agreement engineering** This goes beyond basic offer acceptance. It’s about creating clear alignment that prevents future friction. **Approach:** - Involve an employment attorney to review all agreements - Question vague terms and get specific definitions - Document verbal promises in writing - Address potential issues before signing, not after **Real-world impact:** Imagine winning the startup lottery only to lose everything on a technicality. It happens more often than you’d think. One client shared a painful story with me about how he lost millions because he rushed through final terms. The error? He didn’t get proper documentation for early stock payouts. This mistake cost him over $20 million. Taking time and getting legal support could have prevented this. While slowing down feels uncomfortable in the moment, it’s not as uncomfortable as losing eight figures in payout. **Clarification script:** “I’m excited about finalizing our agreement. I noticed the term ‘significant equity’ in our discussions. To ensure we’re aligned, could you help me understand the specific percentage or number of shares that represents? I want to make sure we’re both clear on expectations.” ### 2. **Value acceleration** Start delivering before day one to create leverage for future negotiations and set yourself up for immediate impact. **Approach:** - Share relevant industry insights during final discussions - Connect key stakeholders to valuable resources - Identify immediate impact opportunities - Set up informal conversations with future team members **Real-world impact:** A recent product head demonstrated this by: - Sending the CEO a competitive analysis they’d created in their spare time - Introducing their future marketing leader to a valuable industry contact - Suggesting three quick wins they could implement in their first month - Meeting informally with key team members to build rapport before starting Their value was evident before officially beginning. And the employer couldn’t imagine hiring anyone else. **Value acceleration tactic:** Even before accepting a new position, send your new boss, “I've been thinking about our conversation regarding [specific challenge]. Here are three resources that might be helpful as we prepare to tackle this together, and I’d love to set up a brief call next week to discuss initial ideas before my official start.” ### **3. Term definition** Never assume shared understanding of corporate language—it’s a minefield of misaligned expectations. **Approach:** - Question every vague or ambiguous term - Get written clarification on all key expectations - Define metrics for success explicitly - Document verbal agreements promptly **Real-world impact:** One client delayed their start date by three weeks to document performance metrics, reporting structure, and future compensation triggers; 30% of their compensation was aligned to getting it right. That patience paid off with a $1.2M acceleration in equity vesting when the company was acquired two years later. Because they had clearly defined what constituted “successful integration” after acquisition, there was no ambiguity about whether they’d met the requirements. Misunderstanding happens constantly with terms like: - “Significant equity” - “Performance-based bonus” - “Growth opportunity” - “Strategic leadership role” - “Competitive package” Each party interprets these through their own lens of experience. **Definition technique:** For any vague term, respond with “I want to make sure I’m delivering exactly what you expect. When you say [vague term], what specifically would success look like in six months? What metrics would indicate I’ve achieved that?” **Examples from the field:** - A client recently discovered that “fully remote” actually meant “remote until our new office opens next year.” - Another found that when one CEO said, “I want you to be a change agent,” they actually meant “Please don’t fire my friends.” ### **Secure success self-assessment** Before finalizing your next opportunity, evaluate your secure strategy with these questions: - **Agreement review:** Have you had an employment attorney review your offer? If not, you’re gambling with your financial future. - **Ambiguity audit:** Circle every vague term in your offer letter or contract. Have you secured written clarification for each one? If not, these are potential land mines. - **Value initiation:** Have you demonstrated value between acceptance and start date? If not, you’re missing a crucial opportunity to strengthen your position. - **Success metrics:** Can you clearly articulate how your performance will be measured in this role? If not, you’re setting yourself up for misaligned expectations. - **Relationship foundation:** Have you begun building connections with key team members before your start date? If not, you’re delaying your ability to create impact quickly. The difference between securing an offer and securing success often comes down to this: Are you focusing solely on getting the job, or are you meticulously engineering the foundation for long-term achievement? ## **Bring the entire negotiation playbook together** ![Image from The ultimate guide to negotiating your comp](https://substack-post-media.s3.amazonaws.com/public/images/b27a4152-2c55-4027-a9a8-25c3f56dae37_1456x892.jpeg) To illustrate how these principles work together in real life, let me share the story of “Carol,” a client whose strategic patience transformed her career. When Carol first approached me, she was a Senior Director at a Fortune 500 company facing significant leadership changes. A new CEO had arrived, creating waves of uncertainty throughout the organization. Work had become increasingly difficult, and Carol was contemplating quitting altogether. Her direct boss, a Vice President, was being considered for promotion to a more senior executive role. But having guided many leaders through similar transitions, I recognized something Carol couldn’t yet see: when new CEOs arrive, they typically bring their own leadership team. Her boss’s promotion was far from guaranteed. I also suspected something else—her boss, being savvy, likely had backup opportunities lined up. This insight became the foundation of our GAINS approach. Instead of helping Carol craft a hasty exit strategy that would leave equity and opportunities on the table, we developed a strategic plan with no predetermined timeline. We simply didn’t know how long it would take for our strategy to unfold—it ultimately stretched to 16 months. We began with the **Gather** phase, mapping the political landscape within the company during this transition. Rather than focusing only on her immediate situation, Carol invested time understanding the new CEO’s priorities and how her boss fit into that vision. She gathered intelligence about which executives were likely to stay and which might depart. The most critical insight came from our observation of her boss’s behavior—subtle indicators that suggested contingency planning was already underway. For the **Align** phase, Carol shifted her approach dramatically. Instead of focusing on her own frustrations or career path, she positioned herself as the ideal successor to her boss. She began demonstrating not just competence in her current role but readiness for the next level. During a pivotal conversation with her boss, Carol shared her observations about the organizational changes and offered specific ways she could support her boss’s objectives. She wasn’t just seeking a promotion; she was creating value as the perfect number two who could ensure continuity regardless of what happened next. Through the **Influence** phase, Carol methodically built relationships across departments. She created champions in Finance, Technology, Operations, and HR by anticipating how her department could remove friction for their teams. One leader later remarked, *“Carol was the only leader thinking about how these changes affected everyone else, not just her own team.”* When her boss began interviewing at another major company, as we suspected, Carol demonstrated what true strategic **Navigation** looks like. Instead of panicking about her own future or trying to protect her position, she offered genuine support. *“I support whatever decision you make, even if you go somewhere else,”* she told her boss. *“I’ll always want to work with you.”* This wasn’t just relationship maintenance; it was strategic positioning that created multiple paths forward. When the predicted scenario unfolded—her boss not receiving the internal promotion and deciding to move to another company—Carol had two valuable opportunities: potentially stepping into her boss’s vacated role at her current company, or following her boss to the new organization in an elevated position. In the **Secure** phase, Carol masterfully managed both conversations. Rather than immediately jumping at either opportunity, she thoughtfully engaged with both companies, creating a transparent but strategic timeline that respected both organizations while maximizing her options. The deliberate approach created a controlled competitive dynamic. The new company recognized they weren’t just hiring Carol’s technical skills but also the continuity and institutional knowledge she brought from working with her former boss. The result was remarkable: an $800,000 cash signing bonus and a $500,000 initial equity grant to convince her to join the new company with her former boss. The package represented more than twice her previous compensation and a first-year win of $1,400,000. But the financial outcome, while impressive, wasn’t the most valuable aspect of Carol’s journey. She entered her new role with clear expectations, strong relationships, and a foundation for long-term success. What makes Carol’s story so powerful isn’t just the compensation increase, but how she transformed potential career disaster into extraordinary opportunity through strategic patience. While others might have reacted impulsively to organizational turmoil, Carol’s 16-month approach to building leverage, creating options, and advancing relationships exemplifies how the GAINS framework works best. ## Bottom line GAINS isn’t a collection of negotiation tricks—it’s a step-by-step framework for creating advantages through consistent and intentional action. The strongest product and growth leaders don’t just negotiate better—they create conditions where traditional negotiation isn’t necessary. Their value is so clear, their relationships so strong, and their options so abundant that compelling opportunities naturally flow their way. Negotiation isn't magic. It’s methodical. It’s about showing up differently every single day: - **Gather** intelligence before you need it. - **Align** through every interaction. - **Influence** across the team. - **Navigate** complexity with patience. - **Secure** outcomes that compound over time. The question isn’t whether you can negotiate better—it’s whether you’re willing to work to create genuine negotiation power. To continue mastering the art of negotiation, I share strategic influence frameworks in [my weekly newsletter](https://www.execsandthecity.com), offer a complimentary [LinkedIn profile analysis](https://www.producthunt.com/posts/free-linkedin-profile-review), and provide [comprehensive interview preparation reports](https://www.interviewreports.com) that transform ordinary conversations into significant outcomes. Stay fearless, friends. *Thanks, Jacob!* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [18/46] State of the product job market in 2025 Last year, I did a two-part series on the state of the product job market (see [part 1](https://www.lennysnewsletter.com/p/new-data-on-the-product-job-market), [part 2](https://www.lennysnewsletter.com/p/state-of-the-product-job-market-part)), where we found: 1. A shift toward hiring senior candidates 2. The decline of Scrum 3. The share of remote roles shrinking 4. The share of PM roles based in the Bay Area growing I**t’s a great time to revisit the state of the product job market, and there’s a lot to be optimistic about! We’re seeing:** 1. **Open PM and engineering roles are trending up** 2. **AI PM roles (and AI-related roles in general) are exploding** 3. **Layoffs are slowing** 4. **Capital investment is increasing** 5. **The Bay Area continues to be the center of opportunity** 6. **Remote work opportunities are declining—and are probably near a new baseline** 7. **There are lots of open PM roles around the world** A big thank-you to [TrueUp](https://trueup.io/) for making this analysis possible through their incredible dataset. They continuously scan **every open job at every big tech company and every top startup,** and then categorize and tag each role. You can browse all open PM roles [here](https://trueup.io/product), and sign up to get job alerts when new product roles are posted. Now, on to the data. *P.S. If you prefer, you can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ## **1. Open PM and engineering roles are trending up** In spite of what you’re hearing about recessions and AI taking jobs, open roles for product managers and engineers at tech companies are *increasing*. There are over 6,000 open PM roles globally right now—53.6% above the bottom we saw in 2023, and already up 11% since the start of the year. This is the most open PM roles we’ve seen in over two years. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/6c8bdf74-b339-4a01-8bd4-d05c7b04ec1f_2913x2491.jpeg) Engineering role growth looks very similar—just scaled up 10x. Open engineering roles saw an uptick earlier this year, and we’re currently seeing the highest number of open roles since late 2022. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/024a1995-988c-4f70-bc59-f757fc723b2d_2913x2491.jpeg) Designer roles have been more stable over time than PM and engineering. They do seem to be dipping a bit recently, but they also never saw as deep a dip as PM and engineering did over the past few years. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/4b195c04-6a89-46a0-89c4-ba130785a6d8_2913x2491.jpeg) ## **2. AI roles are exploding** When you look at just AI PM roles—which includes roles at both AI companies (e.g. OpenAI, Anthropic, Scale) and roles specifically focused on AI products/features at any tech company (e.g. AI/ML PM at Salesforce), the growth we’re seeing is insane. [There are 688 open AI PM roles today](https://trueup.io/ai-product). ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/6f8715a7-df99-43eb-94f0-f549c2dd4d46_2913x2411.jpeg) Zooming out further, here are all “AI” open roles globally. This includes all open roles at what we consider AI companies (e.g. OpenAI, Anthropic, Scale), roles at companies that are heavily AI-driven (e.g. Cursor, Harvey, Lovable), and AI-specific roles at non-AI companies (e.g. AI PM at Figma). ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/e1d51c6c-0707-40f6-9ea9-a671e3bce398_2912x2305.jpeg) ## **3. Layoffs are slowing** More good news: We’re on track to see fewer layoffs this year than in the past four years. However, the possibility of future cuts is ever-present, even at the most respected companies. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/12c582e6-a67f-4739-9272-78f376814cf9_2913x2454.jpeg) ## **4. Capital investment is increasing** We’re seeing more capital investment (e.g. GPUs, data centers, etc.) this year than ever (+29% from last year, +227% from five years ago). It’s still unclear if this only means more hardware or whether it will also lead to more jobs, but I’d rather see this than have it going in the other direction. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/15ac2f42-1f5c-480c-8ab9-8bdd011001e4_2913x3191.jpeg) ## **5. The Bay Area continues to be the center of opportunity** For most tech roles, the Bay Area continues to have the most open roles of any city. For PMs, nearly 20% of all open roles are based in the Bay Area. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/8119b3f9-ff30-4dc2-864a-d9941ea02a6d_2913x2446.jpeg) [Since our last analysis](https://www.lennysnewsletter.com/i/149172750/which-cities-have-the-most-open-pm-roles), Berlin and Austin entered the top 10 locations with the most open PM roles, and Boston and L.A. fell out of the top 10. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/39fffad0-2136-4e17-a480-795462e17a7a_2913x2584.jpeg) For engineering, over 17% of all open roles are based in the Bay Area—and that share is increasing. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/2d867446-b362-4b0d-ae66-48e4cd2377d9_2913x2446.jpeg) Even more so for design roles (assuming that recent dip is temporary). ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/0747016a-e6ec-4fbf-a5a6-a714e953ec8d_2912x2446.jpeg) The share of roles based in the Bay Area is even more interesting for “AI” jobs—almost a third of all open AI-related roles globally continue to be based in the Bay Area. 😮 ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/42535ae1-79ee-4e4a-b2d3-acca19b8ff02_2913x2446.jpeg) ## **6. Remote work opportunities are declining** The number of roles that allow you to work remotely continues to decline. At remote work’s peak in late 2022, we saw about 35% listed open PM roles as remote-friendly. Today, only 23% have a remote option. The trend is clear, and we’re probably near the new baseline of around 20% of roles being remote-friendly. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/439de8fe-9c97-47ab-af32-c8ee276720d4_2913x2445.jpeg) That trend is identical for engineering and designer roles. ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/184095e9-4219-4a61-b82c-dfe80fa006f2_2913x2445.jpeg)![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/d949a6bb-395c-45c9-9f5c-0f37f36d7ee2_2913x2445.jpeg) ## **7. There are lots of open PM roles around the world** Finally, if you’re looking for a PM gig and having a hard time, keep going. The roles are out there. Here’s a list of the top 20 companies that have the most open roles. [You can browse all the open roles here](https://trueup.io/product). ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/8f1f5268-ef61-495f-8bef-31570d3cfaf3_2913x4698.jpeg) If you’re looking for an AI PM role specifically, here’s who’s hiring: ![Image from State of the product job market in 2025](https://substack-post-media.s3.amazonaws.com/public/images/bf1dfaf7-1b13-4c51-8dea-99f00d81b274_2913x4698.jpeg) ## Other trends we’re following 1. The rise of the product engineer: We’re seeing 415 open Product Engineer roles currently (e.g. [v0](https://job-boards.greenhouse.io/vercel/jobs/5466858004), [Amplitude](https://job-boards.greenhouse.io/amplitude/jobs/7897466002), [n8n](https://jobs.ashbyhq.com/n8n/3075bca2-cfaf-474a-9161-f0d190fbad53)), and this seems to be growing. 2. How AI copilot tools change PM interviewing: We’re seeing anecdotal evidence that this is becoming an issue and will change how PMs interview. Overall, there’s a lot to be optimistic about! I’m excited to continue to follow changes in the job market for product roles. Stay tuned for more updates. Have a fulfilling and productive week 🙏 **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [19/46] Five principles for successfully managing managers *👋 Welcome to a **✨ free monthly edition ✨** of my weekly newsletter. Each week I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lennybot](https://www.lennybot.com/) | [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny) | [Swag](https://lennyswag.com/)*** *If you aren’t a paid subscriber, here’s what you missed this month:* 1. *[Make product management fun again with AI agents](https://www.lennysnewsletter.com/p/make-product-management-fun-again)* 2. *[State of the product job market in 2025](https://www.lennysnewsletter.com/p/state-of-the-product-job-market-in)* 3. *[The ultimate guide to negotiating your comp](https://www.lennysnewsletter.com/p/the-ultimate-guide-to-negotiating)* *Annual subscribers also get a free year of [Perplexity Pro, Notion, Superhuman, Linear, Granola, and more](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** [Saumil Mehta](https://www.linkedin.com/in/saumilmehta1/) is the former CPO and Head of Business for Square. He most recently ran Product Management, Marketing, and Partnerships across Square while reporting to the CEO. Over a nine-year tenure, he also held various General Manager roles with full P&L responsibility, including the flagship Point of Sale business. Along the way, he helped Square scale from a late-stage pre-IPO startup to $3.6 billion in annual gross profit, expand globally, and build out a full suite of products across payments, banking, and SaaS. Prior to Square, Saumil was a startup founder for LocBox, a marketing automation company that was acquired by Square in 2015. After stepping down a few weeks ago, he’s spending his newfound time crystallizing his most important leadership lessons learned over the past decade. I’m excited that he is sharing one that he considers maybe his most valuable—the criticality of highly effective “skip leads” to help organizations scale across countries, products, and revenue. In a world where middle managers are derided and founder mode is considered de rigueur, this essay offers a nuanced path forward, with concrete best practices on how these roles can help products and companies scale. You can follow Saumil on [LinkedIn](https://www.linkedin.com/in/saumilmehta1/) and [X](https://x.com/saumil?lang=en) and read his personal writing about company-building on [Medium](https://medium.com/@saumil). *P.S. If you prefer, you can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/0afba8e7-e98c-40d9-89c3-b9a1c50d5730_2912x2129.jpeg) Have you heard that old corporate chestnut “People don’t quit their jobs, they quit their managers”? Most of us have had the experience of hearing a friend or colleague vent about their manager’s lack of support, micromanagement, missing product sense, unfair feedback, or some other relationship problem that is as common as the day is long. As I got more senior at Square—eventually leading all product, marketing, and partnerships, spanning more than 800 colleagues and over 100 managers—I heard these types of frustrations all the time, both at work and with friends. It happened with PMs, engineers, designers, and marketers at all levels of seniority and tenure. Over time, while the conversation stayed the same, my advice changed significantly. **Instead of focusing on the individual’s manager, I now always say, “Tell me about your manager’s manager.”** This person is also referred to as the “skip lead.” ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/4d0446ab-6d5b-4e9d-9b29-662dce5df238_2913x2234.jpeg) My focused questioning on skip leads usually surprises people. But there is a reason for it, born of hard-won experience. In fact, I believe strongly that **all manager challenges are either directly or indirectly about the skip lead**. Instead of placing all the burden on the manager, we need to shift our analytic lens upward, onto the skip. After all, the skip lead hires the manager. They review the manager’s performance. They coach them—or fail to. Over the years, I’ve concluded that a lot of organizational leverage comes from helping newly minted managers of managers become more effective. Through conversation, talent reviews, and direct experience leading large organizations, I’ve developed a few key principles to help new skip leads better understand and perform their role. If you are a new skip lead or hope to be one soon, this is for you. ## 1. Recognize that you are in a different role First, simply take a step back to recognize (and accept) that managing managers is a very **different set of skills and responsibilities than directly managing ICs**.Most people intuitively understand that the transition from IC to line manager is a role change with a steep learning curve. But the notion that the learning curve from line manager to manager of managers is just as steep—if not steeper—is highly counterintuitive. After all, they are *already* “in management.” Isn’t this just the same bag of tricks applied to a larger group? ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/14be7bf4-bfbf-4a3e-8e93-13f431b5b630_2913x1226.jpeg) Nope. Not even close. It’s a very different responsibility set, because instead of directly working with ICs, who have all the context since they are doing the work themselves, the role requires working *through* an abstraction layer—the line manager—and doing so across a half dozen or more managers. Any notion of staying on top of every piece of work goes out the window. Also blurred is the formerly clear division of responsibilities between ICs as makers and managers as supporters, helpers, or coaches, because you now have the situation of managers reporting to *other* managers. And the ability to get unfiltered information straight from the IC on the ground? That’s gone too. Sadly, the organizational and senior leadership failure to even recognize the major role change is *rampant*. And as we know, wherever there is a major lack of self-awareness, mediocrity is sure to follow. In my own org, I was so clear about the under-recognized nature of this problem that for years I insisted on personally signing off on each case of a line manager becoming a manager to other managers. Occasionally I would even set up a one-on-one with the new skip to set clear expectations about the major change in role. It was perceived as overkill at times, but I knew that if the new skip could start from self-awareness about the steep learning curve for a new role, the entire org was already starting out ahead. ## 2. Conduct the API endpoint experiment APIs are wonderful because they allow engineers to get tasks done by simply relying on the interface and ignoring all the underlying complexity behind it. The analogy can be applied to managing managers by engaging with the following thought experiment: “If I had zero access to any IC in my organization and could only push or pull information from my managers to make decisions, how would I go about being effective?” This type of introspection is powerful in the same way as lifting heavy weights at the gym or taking a cold plunge: doing something uncomfortablein an unnatural environment produces growth. Most new skip leads who were until recently line managers are most comfortable with dealing directly with ICs, getting detailed context from them and giving them direction. This provocative question forces them to grapple with how they’d create value when that core context-building and decisioning tool in their toolbox is gone and only the managers directly reporting to them are available. Immediately it forces the new skip lead to confront a panoply of follow-up questions, like: 1. How do I help line managers spot problems on their team without talking to their ICs? 2. How do I help line managers deal with IC performance issues on their team instead of dealing with them myself? 3. When do I defer to the line manager’s judgment if it runs counter to mine? 4. When do I insist that my point of view carries the day even though the manager has talked to all the ICs and has more firsthand information? 5. How do I review work coming out of a line manager’s team in a way that is additive or complementary to what the manager is already doing? 6. How do I give input early enough that a line manager can course correct before it’s too late? 7. What are the areas that I fully delegate to my line managers and what are the areas where I retain final sign-off, and why? Running through this exercise has two key caveats. First, answering these questions at a level of instinctive competence takes years of practice. (But I find that new skip leads who *ask* these questions accelerate their learning in their role.) And second, this is a thought experiment, to be taken seriously but not literally. In real life, skip leads *should* of course talk with their skip reports on a regular basis. But how they engage with their skip reports and how that interplays with their direct reports can benefit from an exercise in extremes. Let’s now consider how to interact with direct-report managers and their reports (i.e. skip reports) with the next principle. ## 3. Never undermine At Square, I had recurring one-on-ones with dozens of skip reports, weekly office hours available for anyone in the 800-person org, and lots of ad hoc conversations across the org over Slack and in person. If you are a new skip lead, this should be common practice for you too. These conversations are extremely valuable because they provide unfiltered feedback that your direct-report managers may have missed or even intentionally filtered out. They also provide a crucial source of context on how well your direct-report managers are doing. But these conversations are also risky. The natural instinct for new skip leaders is to offer strong opinions, solve problems, and make decisions on the spot without getting full context from anyone else in the management chain. This feels efficient! It may even feel like engaging in “founder mode.” But in general, off-the-cuff solutions and strong opinions actually undermine your direct-report managers in the management chain and disempower them in front of their own team. You rob your direct-report managers of the opportunity to get your coaching on how to handle specific situations, learn your line of reasoning, understand your context—and then apply all of this themselveswith their team in an ongoing and scalable way. In addition to undermining the direct-report manager, you can easily confuse the skip report with contradictory directions between you and their direct manager. Jeff Weiner, the former CEO of LinkedIn, articulated this problem in a great [essay](https://www.linkedin.com/pulse/20140602024642-22330283-avoiding-the-unintended-consequences-of-casual-feedback/) on the topic: “Years ago, a former direct report of mine helped bring this point home. While he and his team welcomed my input, he observed that oftentimes what I thought was a take-it-or-leave-it remark would create a massively disruptive fire drill. Up until that moment, I had no idea my opinion was being weighted so heavily.” The far better option, instead, is to listen carefully, ask lots of follow-up questions, and then actively work with your direct-report manager to do all the follow-up and actual direction-setting. ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/19a3a9d4-1755-4fcd-9c9f-5cbbcc7e3a62_2913x1512.jpeg) None of the above is in conflict with the recent exhortations to founder CEOs to engage in founder mode. Founder mode, done well, results in the founding CEO staying highly engaged and hands-on while also coaching and empowering the management chain. Founder mode done poorly is simply a recipe for undermining, disempowering, and creating chaos. ## 4. Never cover The flip side of undermining your direct-report managers as a new skip lead is *covering* for your direct-report manager’s underperformance with your own bosses. Consider the following visual, which will feel very common to any PM who has worked in a large organization. ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/57becc07-bcd1-4755-8665-e41e2531b732_2913x2428.jpeg) Even though the skip leader subconsciously (or even explicitly) knows that the line manager is struggling, confronting the issue head-on is hard—especially as a new skip. So the natural inclination is to quietly solve the issue themselves, pull double duty by taking on part of the line manager’s job, and then provide cover for the direct-report manager’s performance with others. In fast-growing tech companies, this happens often. I’ve lost track of the number of times I’ve sifted through the wreckage of a project gone sideways and found exactly this issue. But the instinct to cover for bad news is deeply rooted in human emotions like shame and embarrassment of failure, fear of confrontation, and insecurity. The better approach for the new skip leader is to (1) clearly report the underperformance and a plan to course correct up the management chain, (2) solicit feedback on the course correction from leaders up the management chain, and (3) set clear performance expectations with the direct-report manager. ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/a6f45c32-17a8-435c-8e53-f6d0c3a2947e_2913x2428.jpeg) ## 5. Use task-relevant maturity (TRM) to delegate and review work As we see from the principles so far, skip leads can be most successful when they learn how to get the most out of their direct-report managers. Broadly speaking, this is about delegating responsibilities across direct-report managers to maximize the output of the overall team. Sounds simple enough, right? But in practice, delegating well is very difficult. In most high-growth technology companies, the skip leader navigates a constantly changing landscape. Every few weeks, new projects are planned. A previously planned project that is now underway goes off the rails when engineers find unexpected problems. A large customer complains loudly to senior leadership, resulting in a fire drill. A senior IC in the midst of a critical project threatens to quit. An engineer and a designer get into a heated argument over a prototype. A sudden regulatory change threatens an upcoming country launch. As you can see, each of these projects—and the situations that surround them—varies widely in the level of effort and time required, the level of ambiguity and technical complexity, the number of teams involved, and potential impact to the company. ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/bf8fe209-68d8-439d-bb4c-ddf65ac91da7_2913x2106.jpeg) To further complicate matters, projects and situations are only one half of the ledger. The skip lead also has to figure out which of their direct-report managers is best suited to take them on, because just as every project or situation is different, so is every line manager.. In any skip leader’s organization, line managers will vary widely in skill level, institutional knowledge, emotional maturity, ability to handle ambiguity, and other unique strengths and weaknesses. In product development, managers may even specialize in engineering sub-disciplines, like mobile, and PM leads may specialize in domains like developer tools or consumer social media. Just as every project or situation is different, so is every line manager. Delegating well is so difficult because it requires constant matchmaking between a variety of projects and a variety of people in an ever-changing environment. Thankfully, Andy Grove’s classic *[High Output Management](https://www.amazon.com/High-Output-Management-Andrew-Grove/dp/0679762884/ref=sr_1_1)* gives us a key management concept framework for this kind of matchmaking: **task-relevant maturity (TRM)**. Task-relevant maturity asks skip leads to evaluate the particulars of each project or situation—its scope, size, risk, duration, ambiguity, and so on—and then consider which manager is best suited based on their skills and experience (specifically for that project or situation). When a project or situation arises, the skip leader can ask questions to determine which manager would be the best match for it: - Which of my managers has dealt with a very similar project before? - Which of my managers has the right level of technical skill required by this situation? - Which of my managers has asked me to be stretched into a project like this? - Which of my managers is senior enough to deal with a high degree of ambiguity? - Which of my managers do I trust the most with this very delicate interpersonal situation? - Which of my managers is best suited for a mobile-heavy project? - Which of my managers can do this if I personally lean in a bit more than usual? - Who can I hand this off to such that I can have this off my plate altogether? Skip leads who are elite at delegation continually filter projects, situations, and people through the lens of task-relevant-maturity to maximize team output and velocity. But delegation is only the first critical step. As Andy Grove reminds us in the book, delegation does not mean abdicating responsibility. It is still the skip leader’s job to ensure that they deeply review work on an ongoing basis—and micromanage when necessary—to keep things on track. Ultimately, the buck stops with the skip leader. This 2-by-2 rubric can help skip leaders with post-delegation review across two key dimensions: uncertainty and impact. By clearly articulating the uncertainty of a project (e.g. a new 0-to-1 product offering, which by definition is highly uncertain) and the project’s impact (e.g. a key launch in the core product that can make or break the financial plan), skips can easily adapt their own review cadence after they initially delegate the project to a line manager. ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/911008db-aac0-42e1-88b5-3f4a1e492c09_2913x2069.jpeg) For example, a project with low impact and low uncertainty can be easily delegated to a newer line manager and “forgotten” about. But a project with high impact and high uncertainty should be assigned to someone much more capable (i.e. with high TRM for the situation) *and* reviewed closely and continually. A common category of high-uncertainty, high-impact project involves multiple teams working together, cutting across line managers and maybe even across skip leads. Given the inherent difficulties of inter-team communication, competing priorities across teams as well as the number of line managers involved, it is the skip leader’s responsibility to continually review work after delegating to find problems as early as possible—and fix them. ![Image from Five principles for successfully managing managers](https://substack-post-media.s3.amazonaws.com/public/images/46e8f913-47b4-4dc8-b89d-79ce19757e01_2913x1589.jpeg) ## In closing A new manager of managers can start to apply these five principles in their first week in the role, and hopefully for years after. It starts with recognizing the change in role and the steep learning curve. If new skip leads use cognitive hacks to focus attention, are careful about interactions with ICs and the accidental impact on line managers, deeply understand task-relevant maturity, and use inconsistency as a signal to resolve conflicts, they can become experts at managing managers over time. But if you are new to this role, I hope you will also give yourself the grace of patience. Getting to elite status is hard and takes many years. After over a decade of managing managers, I’m still learning, improving, and refining my pattern-matching instincts. You will too. And maybe, in a few years, when you get that prototypical complaint from a friend, you too will ask, “Who’s the skip?” *Thanks, Saumil! You can follow Saumil on [LinkedIn](https://www.linkedin.com/in/saumilmehta1/) and [X](https://x.com/saumil?lang=en) and read his personal writing about company-building on [Medium](https://medium.com/@saumil).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [20/46] How tech workers really feel about work right now Today, with insights from over 8,200 of you (thank you to everyone who participated!), we’re excited to share the results of our first-ever large-scale tech worker sentiment survey. It’s hard to imagine a more intense time to be working in technology, so Noam and I wanted to capture a rich picture of tech workers’ sentiment—at this moment, and going forward. Our analysis examines several dimensions of tech worker sentiment, including career outlook, role satisfaction, burnout, quitting intentions, career path clarity, leadership effectiveness, work setup preferences, and more. What we discovered is that tech professionals are experiencing a fascinating mix of emotions about their careers in 2025. *P.S. If you prefer, you can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* # Our biggest takeaways 1. **Burnout is at critical levels:** Almost half of our respondents are experiencing *significant* burnout. 2. **Tech workers are more optimistic than we expected—but optimism is declining:** 58.5% of tech workers remain optimistic about their roles, and 54.8% remain optimistic about their careers. However, there has been a significant negative sentiment shift over the past year. 3. **Startup founders are the happiest people in tech:** They’re the only group growing more optimistic while consistently outranking everyone else in workplace well-being. 4. **Managers need help:** Only 26% of tech workers consider their managers highly effective, while over 40% view them as ineffective. 5. **Where people work makes little difference in how they feel about work—on the surface.** But dig deeper, and hybrid workers are the happiest, remote workers are doing well, and in-office workers are experiencing hidden frustrations. 6. **Small-company employees are doing the best:** They outperform their large-company counterparts on nearly every work sentiment measure, from job enjoyment to sense of belonging. 7. **The mid-career slump:** Mid-career workers are struggling the most with burnout, lower job enjoyment, and the most pessimism about the future. 8. **A widespread gap in career clarity:** Many tech workers don’t know what they should be doing to continue developing in their careers. ## How we ran the survey and who responded ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/700cf469-4192-4cfa-aa6b-d86ebf8e175d_1600x1379.jpeg)![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/fbd6be0f-0bba-49fc-8cbe-2d5c7d66ff91_1600x1381.jpeg) We started by asking about people’s optimism (or pessimism) regarding the future of their *role* and *career*. We also asked people to reflect on the past year in tech and how their feelings had changed. We then dug into people’s work experiences, including enjoyment, burnout, quitting intentions, career development thoughts, and manager effectiveness. Finally, we explored how people feel about their work environments, how engaged they are at work, and how much belonging they feel in their organization. ## Takeaway 1: Burnout is a critical issue facing the tech workforce ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/7693a33e-5d97-42da-96ee-9aa5332b8a08_2912x2261.jpeg) - 44.67% of all respondents are currently experiencing *significant* burnout (meaning they said they’re “moderately,” “very,” or “completely” burned-out). - Burnout doesn’t just affect tech workers’ experience in their current job—it’s also correlated with people’s optimism or pessimism about the future of their career. 60% of extremely *pessimistic* respondents reported high burnout. Only 8% of extremely *optimistic* respondents reported that they’re highly burned-out. - Interestingly, workers at smaller companies report significantly lower burnout levels than those at larger companies. However, there’s a dramatic spike in burnout within midsize companies (500 to 1,000 employees), perhaps because they’re large enough to have corporate bureaucracy but not large enough to have well-developed employee support systems. - Full-time employees experience nearly twice the burnout rate of self-employed workers and contractors. More autonomy and control over work arrangements equals significantly lower burnout rates. - Hardware is indeed hard. Tech workers in hardware and infra roles show the highest burnout rates, whereas people are (moderately) less burned-out working on emerging technology (e.g. AI) and specialized sectors (e.g. fintech, healthtech). - Burnout varies by role, and there's a clear split - founders and engineers are dealing with way less burnout compared to everyone else. ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/a1eaf977-b81b-4d2c-aa83-6b9343bb3a45_2912x2288.jpeg) - Not surprisingly, there’s a clear business impact of burnout: - Burnout strongly correlates with quitting intentions—67.8% of workers who are actively exploring new opportunities or considering quitting are highly burned-out. - Burnout also strongly correlates with worker engagement. High burnout dominates among workers with low engagement, with close to 30% (!) of not-engaged workers saying they’re *completely* burned-out. ## Takeaway 2: Tech workers are more optimistic than we expected—but optimism is declining We asked tech workers how optimistic or pessimistic they are about: 1. The future of their *career* 2. The future of their *job function*, in light of the rise of AI Broadly speaking, tech professionals maintain a positive outlook, with 54.8% expressing optimistic *career* sentiments, versus 33.3% expressing pessimistic views. *Job function* sentiment is even rosier, with 58.5% reporting optimistic feelings and only 25.1% pessimism about their specific role. In both cases, the rest of the respondents selected “neutral.” We were curious to understand how each respondent views the future of their career compared with the future of their role. So we dug deeper into the comparison between current roles vs. long-term career prospects: - 45.2% of respondents have aligned career and role sentiment, meaning they’re equally optimistic (or pessimistic) about their career *and* specific role. - 32.0% are more optimistic about their current role than their overall career. - 22.8% are more optimistic about their career than their current role. This means that while many tech workers feel good about their immediate job situation, they have more concerns about long-term industry trends and career progression. But optimism seems to be declining. We asked tech workers how their feelings have *changed* over the past year, and there’s been a significant negative shift in sentiment: ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/e51e74ff-4575-423a-99a3-2f786bf6bcc9_2912x1891.jpeg) This could be due to the economic climate, the rise of AI, or other factors we’re still investigating. Only two segments have become more optimistic over the past year: founders and new college grads. Conversely, designers and researchers had the largest negative sentiment change. Employees at large companies feel more pessimistic, as do fully remote workers. ## Takeaway 3: Startup founders are the happiest people in tech One of the most striking patterns is that founders have the most positive sentiments across nearly all metrics compared with other tech roles: - The highest career optimism - The highest job function optimism - More optimism compared with last year (which means they’re *way* more optimistic about the future than everyone else in this sample) - The highest job enjoyment - The lowest burnout levels - Perhaps less surprisingly, founders are the most committed to their roles, with the lowest quitting intentions - Finally, founders feel much more engaged and have a significantly stronger sense of belonging than all other roles ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/42b0dedc-bbc0-4515-89fb-51f425df6dd1_2926x1878.jpeg) This remarkable pattern suggests that ownership, autonomy, and purpose are potent drivers of work satisfaction and well-being. ## Takeaway 4: Managers need help Fewer than 1 in 3 tech workers (26.6%) rate their managers as highly effective, while more than 4 in 10 (42.3%) rate their managers as ineffective. **But the mind-boggling insights are the correlations (which don’t imply causation!) between direct-manager effectiveness and worker sentiment.** ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/66336b6d-160e-4921-b2dd-d33940542a45_2912x2113.jpeg) Leadership impacts *virtually all worker sentiment dimension*s. For most sentiment metrics, there’s a massive gap between those who rated their managers as ineffective and those who rated them as extremely effective. - People with great managers are 48% more engaged than those with poor managers. - People with great managers also feel 63% (!) more belonging, 31% less burnout, and 62% (!!!) more job enjoyment compared to those with ineffective managers. Retention is also deeply impacted by manager ratings: - Workers with ineffective leadership are 4.3 times as likely to be at risk of leaving (72.6% vs. 16.9%). - Workers with extremely effective leadership are 8.2 times as likely to be committed to their role (61.6% vs. 7.5%). - The most dramatic difference is in workers “actively looking” for new opportunities, who are 5 times as likely to rate their manager poorly. Leadership quality may be the single most important lever for retention. It’s tied to substantial increases in engagement, belonging, and enjoyment, and notable decreases in burnout. Interestingly, remote leadership was rated slightly more effective than in-office leadership, challenging assumptions about remote management challenges. Leadership effectiveness ratings gradually decline as company size increases, with small companies showing modestly better leadership perception than large enterprises. People in product and design roles have the most negative leadership perceptions. The leadership effectiveness gap represents a major opportunity. With only 26.6% of workers rating their manager as highly effective, improving leadership quality represents one of the highest-leverage investments organizations can make. ## Takeaway 5: Hybrid workers feel better about their job; in-office workers feel better about their career Over 50% of tech workers in this sample are primarily working remotely (37.7% fully remote, 16.9% mostly remote). Only about 5% of respondents are in an office every day (a ratio that’s in line with current industry data). At first glance, the data seemed to suggest that tech workers are feeling just about the same, no matter where they work: - **Job enjoyment is the same for everyone.** Apparently, on-site restaurants, free dry-cleaning, and video games don’t drive higher job enjoyment. - **Burnout is also the same across work setups.** Remote workers aren’t experiencing more burnout because they are physically separated from their teams. - **Work setup isn’t related to quitting intentions.** - **Engagement levels are remarkably consistent across work arrangements.** Tech workers’ interest, involvement, and connection to their work aren’t related to where they do it. - And contrary to common assumptions, **remote workers feel just as much belonging as in-office workers do**. But when we dove deeper, we revealed some unexpected patterns: - We asked respondents directly how their work setup has impacted their feelings about their job. **People working hybrid (mostly from home) felt** ***much*** **better about their job than those who work in-office, a 21% difference in sentiment.** - **However, when it comes to feelings about the** ***future*****:** - In-office workers were slightly more optimistic about the future of their *career* compared with fully remote or even hybrid workers. - In-office workers were also slightly more optimistic about the future of their *job function* in light of the rise of AI. In other words, hybrid workers are feeling *a lot* better about their job, but in-office workers feel better about their *career prospects*. ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/61641e55-6fb2-4e94-9a97-6e5a1440a234_2912x2295.jpeg) **What this all means:** 1. Hybrid workers are the happiest, but remote workers are surprisingly happy as well. These findings challenge the standard narrative that in-office work inherently creates stronger engagement, belonging, or job enjoyment. 2. In-office workers are doing OK on the surface, but when asked directly, they share struggles that are likely to manifest later on. 3. Flexible work arrangements that prioritize remote work with some in-person interaction may provide the optimal balance for most tech workers. It’s also important to note that several confounding variables may be in play regarding work location. Age, company size, and role distribution are all related to work setup. It’s possible that some differences are due to those factors rather than the work location itself. For example, remote workers skew older (45.1% ages 35 to 44, vs. 30.8% for in-office). Or hybrid arrangements are more common in larger companies (30% at 1-to-50-employee companies, compared with 46.6% at 1,000-plus-employee companies). ## Takeaway 6: Small-company employees are doing the best Company size emerges as one of the strongest predictors of positive sentiment, with statistically significant differences across organization sizes for 9 out of 11 sentiment metrics. ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/160ab535-5b0f-47b9-9725-e634d3c94002_2912x2287.jpeg) **People in smaller companies are:** - More optimistic about their career - More satisfied in their role - Experiencing less-pronounced negative sentiment change - Enjoying their job more - Less burned-out - Less eager to quit - Rating their managers more highly - More engaged - And we saved the strongest correlation for last: They feel more belonging. These findings reinforce the importance of creating tight-knit teams and smaller communities within large organizations. Big companies must foster a sense of connection and purpose that’s clearly lost when they grow beyond their first 50 to 100 employees. ## Takeaway 7: The mid-career slump is real Career optimism declines with experience, and the decline is substantial, with early-career workers scoring 15% higher than their later-career counterparts. But across multiple other sentiment variables, we observed a U-shaped pattern with age and experience: ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/df096cfc-441d-40c7-8bb9-a0663446c751_2912x2015.jpeg) - Early-career professionals (0 to 6 years) generally report high optimism (about 62% optimistic), low burnout (47% not or only slightly burned-out), and high job enjoyment (71% enjoying their work). - Mid-career professionals (7 to 14 years) report the most negative sentiment patterns, with higher burnout (41% not or only slightly burned-out), higher quitting intentions (28% committed to their roles), and lower job enjoyment (66% enjoying their work). - Late-career professionals (15-plus years) report higher job enjoyment (75% enjoying their work) and lower quitting intentions (34% committed to their roles) than mid-career professionals. Engagement and belonging remain fairly consistent from early to mid-career stages. Late-career professionals report the highest levels of engagement and belonging, and also enjoyment, clarity, and commitment. Apparently, if you make it through the mid-career slump and find your niche, you become more content compared with earlier-career workers. This pattern suggests that organizations should pay special attention to mid-career professionals, who may be experiencing a “career crisis” when early enthusiasm has waned but long-term career resilience and clarity haven’t yet developed. ## Takeaway 8: There’s a widespread gap in career clarity We asked tech workers how clear they are about the skills and career development paths they should pursue to reach their professional goals. What we found is widespread uncertainty: ![Image from How tech workers really feel about work right now](https://substack-post-media.s3.amazonaws.com/public/images/dc04a3c9-64dd-43dd-9e93-e44a1929e455_2912x1793.jpeg) Career clarity was strongly related to our other key sentiment metrics, particularly engagement and leadership perception. It’s hard to be an engaged worker when you’re not sure what to engage *in*. And managers play a critical role in setting expectations and guiding workers along their career path to secure their future. This uncertainty around career development likely contributes to career anxiety and the negative shifts in future outlook we’ve observed, and it may partly explain the high percentage of professionals considering quitting (about 40%). ## Bonus takeaway #1: Women are more burned-out (but more engaged) than men In this sample, more women reported being burned-out—or burned-out to a greater degree—than the male respondents did. We’re not talking about massive gaps but, rather, consistent patterns that show up across large samples: - Women’s burnout scores are roughly 3% higher than men’s. - Women’s role and career optimism are about 3.3% lower. Women also reported slightly lower enjoyment and more negative emotional impacts from their work setup. Now, these might seem like small differences (and statistically speaking, they are small effect sizes), but they’re significant. Small sentiment gaps can have outsized impacts on retention and performance over time. But interestingly, women respondents said that they’re also more engaged and have a stronger sense of belonging than the men we studied. What does this tell us? Women in tech are feeling more connected and committed to their work (and work communities), while simultaneously feeling more strained and less optimistic about their future in the industry. The differences may be small, but it’s important to be mindful of trends like these that can hurt women’s equity and inclusion in the industry. ## Bonus takeaway #2: AI is keeping tech workers up at night About 1,750 of our respondents shared more about how they’re feeling about working in tech. Two themes stood out (and they’re both about AI): ### 1. Tech workers are navigating existential anxieties and the practical realities of AI Many respondents expressed deep uncertainty about the future of their roles and the tech industry as a whole: > *“What’s the endgame? Many articles have a fear-mongering message that every tech role is replaceable by AI in the foreseeable future. Is the future just a handful of ‘founders’ with AI agents for every existing role? Are we going to have AI sales agents selling AI-made products to other founders with their own AI-made products and AI-agent teams? AI-generated consumer products with AI-generated content and AI-generated comments? It’s hard to feel motivated for the long term when the current product culture is so heavily invested in talking about AI that the craft of product management is falling by the wayside.”* > *“Work is also now all about how many AI agents you can use and deploy. What will happen to people’s thinking skills? Look at what highly processed, optimized foods have done to our bodies. What will highly optimized, efficient LLMs and agents, which can do your thinking, writing, and execution, do to our brains? I’m not optimistic about all of this.”* And there’s a real struggle to distinguish AI hype from the current practical realities of the technology: > *“It generates pessimism to see how little many tech leaders understand ‘AI’—when we’re largely still dealing with ML and LLMs, and where LLMs are largely still chatbots. We’re not addressing harms, we’re forgetting about hallucinations, and we’re not aligning the use of these proto-AI tools to user needs. . . We’ve got a glut of founder-vision-driven AI-hyped feature factories in early startups, and the setting in of enshittification of older startups.”* ### 2. The AI era is a time of career transformations A significant number of our respondents are questioning their career path and fundamentally reconsidering their profession, not just their current job. Some are making deliberate career moves to “future-proof” themselves, focusing on whatever they perceive as more “AI-resistant.” > *“Most often I find myself wondering not about whether I have a path for growth in product but if I want to stay in product in the first place, and what I might pivot to if not. I have zero ideas where to start.”* However, there are plenty of barriers to these strategic repositioning attempts, in particular those we’ve already called out—lack of managerial support, uncertainty about which skills to develop, and limited internal mobility. > *“I’ve been in a multi-year cycle of re-orgs, manager changes, and empty promises of promotions that never happen. Losing motivation since I don’t see a path for professional growth for myself.”* The AI era has also bolstered entrepreneurial ambitions: a notable number of our respondents expressed interest in building their own products as a response to industry uncertainty. Some have already gone for it. > *“I’m very positive because I quit the company I was working for the last 10 years to build my own product. Create a path for yourself.”* ## Where do we go from here? The data paints a clear picture of how tech workers are feeling. Burned-out, but somewhat optimistic. Seeking more autonomy and control, but also guidance and inspirational leadership. Searching for clarity, and hoping it comes with flexibility and an environment where they can feel belonging. It’s a complicated moment to be working in tech for sure. **To give you something tactical to take away immediately, here’s what the data tells us you can do if you want to be happier at work:** 1. Monitor your burnout. [Take our burnout survey](https://theburnoutcheck.com/) and consider making changes if you don’t like what you see. 2. Try working at a smaller company. 3. [Or start your own company!](https://www.lennysnewsletter.com/t/startups) 4. Become better at [managing up](https://www.lennysnewsletter.com/p/managing-up) (or become a manager and [learn to do it better](https://docs.google.com/document/d/1pQei75-auE2c2bxHkOuIRBC9Udd2oOzAEo--m0KbhP8/edit?usp=sharing)). 5. Find a more flexible work setup (e.g. a hybrid-friendly role). 6. If you’re lacking career clarity and especially if you’re mid-career, [“offsite” with yourself](https://docs.google.com/document/d/1xzOCH0Tnyg8kPoFa_811SJL0rpKJeMA57QjBMR2apew/edit?usp=sharing) and figure out your future. And here are the five highest-leverage moves companies can make to give tech workers what they seek right now: - **Invest in leadership development.** With only 26.6% of tech workers rating their managers as highly effective, this is your biggest opportunity. Great leadership strongly correlates with higher engagement, stronger belonging, and less job-seeking. Nothing else moves the needle this dramatically. - **Inspire and motivate “founder mode” for everyone.** Founders showed the most positive sentiment across nearly every metric. Enable your teams to have maximal autonomy, ownership, and purpose—these are the secret ingredients that make founders 3x to 4x more optimistic, 2x more engaged, and almost 3x happier than everyone else. - **Embrace flexible work.** The data challenges the standard narrative about in-office work. Core job metrics don’t significantly differ between setups, but hybrid arrangements are viewed most positively. Focus on creating connection regardless of physical location. - **Think small, even when you’re big.** Company size was one of the strongest predictors of positive sentiment. People at smaller companies reported higher optimism, less burnout, and dramatically stronger belonging. Large orgs must create tight-knit teams and communities to combat the “big-company blues.” - **Target mid-career support.** The data shows a clear U-shaped pattern in sentiment. Your mid-career folks (7 to 14 years of experience) are struggling most with career clarity, showing higher burnout and lower enjoyment. They need your attention before they head for the exit. While executives fret over AI disruption and remote work policies, our data reveals more layers to what’s going on with tech sentiment. The companies that will dominate the next decade won’t be those that stop at developing the most advanced AI or the fanciest offices, but those that recognize that treating people like founders rather than functions may be the ultimate competitive advantage. Special thanks to the more than 8,200 tech professionals who shared their insights for this survey. Your candid feedback will help us track sentiment in the tech industry and identify areas for improvement. Have a fulfilling and productive week 🙏 **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny and Noam 👋 --- ## [21/46] Taking the week off I’m taking this week off to recharge ([per my PTO policy](https://www.lennysnewsletter.com/about#%C2%A7posting-schedule)). If you’re looking to learn something in the meantime, I encourage you to play around with [Lennybot](https://www.lennybot.com/). It’s trained on every single one of my newsletter posts and podcast interviews. It’s good. And totally free. Some questions to [ask it](https://www.lennybot.com/) to get you going: - How do I find an idea for a new product? - How do I write a strategy doc? - What can I do to get promoted faster? - How do I know if I should be a founder? - What are some great interview questions for product managers? - How do I help my team move faster? - How do I get better at managing up? - Burnout. What can I do about it? Pro tip: Try the “Voice” mode. ![Image from Taking the week off](https://substack-post-media.s3.amazonaws.com/public/images/dca89170-5e3e-4f73-b2d3-10ccb341f3e7_4032x3024.jpeg) See you next week! --- ## [22/46] How to get your entire team prototyping with AI [Colin Matthews](https://www.linkedin.com/in/colinmatthews-pm/) holds the distinct honor of having not one but *two* posts in my top 10 [most popular posts of all time](https://www.lennysnewsletter.com/?sort=top). Today he’s back with a post that I suspect will make it three. I love Colin’s stuff because it’s always rooted in the real-life problems that product builders actually run into with AI tools—and he articulates the solutions so simply and practically. I’m excited to keep collaborating with him to help us all learn how to go from hype to reality. Let’s get into it. *For more from Colin, check out his [free 45-minute live online class](https://maven.com/p/becb6f/product-development-lifecycle-with-ai-prototyping) coming up on July 8th, and his course [AI Prototyping for Product Managers](https://maven.com/tech-for-product/ai-prototyping-for-product-managers?promoCode=LENNYSLIST) (which kicks off on June 30th; use code LENNYSLIST to get $100 off). You can also [book Colin for one-day team workshops](https://teams.techforproduct.com/) with your team.* *P.S. If you prefer, you can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/2e8ff549-677d-4cce-b38c-8176c8c461dc_1456x1098.png) Since I first shared a [guide to AI prototyping for product managers](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product) in January, AI prototyping tools have become immensely popular. It’s now more common than not for product teams to use v0, Bolt, Replit, or Lovable to build a quick prototype to share their ideas and get feedback. After having taught more than 500 PMs over the past five months how to get the best out of these tools (along with another 10,000 in numerous free online lightning lessons), I’ve seen two questions come up most often when people try to use these tools in real life: 1. How do you make your prototypes look good enough to show customers or senior stakeholders? 2. How do you successfully adopt these tools as a team, instead of a lot of individuals working in silos? These common questions are the reasons product, design, and engineering teams are struggling to figure out team workflows for AI tools. Without handoff guidance or high-quality prototypes, AI prototyping is left to explorations and experiments, which are hard to operationalize and scale. In this post, I’ll walk you through three ways to create component libraries that consistently create great mocks; team workflows to reduce rework; and a step-by-step guide on how prototypes should be used. I’ll also give you plenty of time-, token-, and headache-saving tricks for most popular AI prototyping tools. I’ll be including [v0](https://v0.dev), [Bolt](https://bolt.new), [Cursor](https://www.cursor.com/), and [Magic Patterns](https://www.magicpatterns.com/?via=tfpl1) in this post, but these techniques work with any AI prototyping tool. To give you a quick example of what’s possible, here’s an Airbnb prototype I built with a single prompt: “*Build a homepage and a listing details page.”* ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/83eb4332-a6d7-430c-a570-c8b023ac0681_810x453.gif) Let’s get started! ## **Component libraries** Component libraries are the first big improvement you can bring to your team. They allow you to maintain branding and consistent styling without having to manually clean up each prototype to look like your product. Building a component library will take some up-front effort, but you’re also building reusable assets that will greatly improve the quality of your prototypes going forward and save you time after the initial investment. There are three different approaches to building component libraries with AI prototyping tools, with different advantages and tradeoffs in effort and output: 1. Screenshots 2. Chrome extensions 3. Code ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/34d32270-55dc-4a08-8b81-9fbf7c058613_2912x2140.png) Let’s go through each and discuss their tradeoffs. ### **Method 1: Screenshots** Screenshots are the easiest way to start building a component library. You don’t need any technical expertise, and this approach works with any tool. Start by prompting your favorite AI prototyping tool with the following: `You are tasked with creating a component library based on a screenshot using React, and Tailwind CSS.` `All components should be custom-made to match the screenshot as closely as possible.` `Follow these instructions carefully:` `1. Analyze the provided screenshot.` `2. Identify distinct UI components in the screenshot. These may include, but are not limited to, ◦ Buttons ◦ Input fields ◦ Navigation bars ◦ Cards ◦ Modals ◦ Typography elements` `3. For each identified component:` `a. Create a React functional component.` `b. Use Tailwind CSS classes to style the component, matching the visual design in the screenshot.` `c. Ensure the component is responsive and accessible.` `d. Add any necessary props for customization.` `e. Include a brief comment describing the component’s purpose.` `4. After creating all individual components, create an index page that imports and displays each component with example usage.` `Remember to use only custom-made components and Tailwind CSS classes. Do not use any external libraries or pre-built components.` `Strive to match the visual design in the screenshot as closely as possible while maintaining good coding practices and component reusability.` Let’s give this a try with v0, re-creating a Google Calendar UI. Remember to include a screenshot! ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/f6b3e2f9-7e1c-4d0f-868c-8a78df59e155_796x613.jpeg) Here’s our initial result: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/883135d3-0c4a-4dda-9fc0-de2d932f25d9_1334x748.gif) One quick note: You may end up with a reproduction of the screenshot instead of a list of components. These AI tools are very tuned to reproduce screenshots, so you may need to push it with one more prompt, like *“Instead of showing me a page that re-creates the UI, the index page should be a list of components in the component library.”* From here, you can continue adding more components by pasting in additional screenshots with the prompt *“Add these components as well.”* If you want to refine the components, I recommend starting with this prompt: `“List the differences between the screenshot and your implementation. How can you match the design more exactly? Don’t code.”` This technique (called reflection) makes it easy to leverage the AI to quickly make UX improvements. Once you have your component library complete, pause! You’ll want to create a fork of the project and start to use your components in a new project. Your forked project will automatically reuse your components to assemble your design like Legos. Again, this technique works in any of your favorite AI prototyping tools. (Here’s how to create a fork in each AI prototyping tool: [Bolt](https://support.bolt.new/building/using-bolt/projects-files#duplicate-projects), [v0](https://www.perplexity.ai/search/how-to-create-a-fork-in-v0-m6V2aWHwQZqx8UCQZWzEgg), [Lovable](https://docs.lovable.dev/user-guides/quickstart#remix-an-existing-project), [Replit](https://docs.replit.com/getting-started/quickstarts/remix-an-app), [Magic Patterns](https://www.perplexity.ai/search/how-to-create-a-fork-in-magic-tGrNnh7RRWOZvNSDMrNeyg).) For example, we can prototype an AI agent in Google Calendar that blocks deep-work slots with a simple prompt like *“Create a Google Calendar main page view that has an AI scheduling assistant to recommend deep-work blocks.”* Here’s the result: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/2e5aab6f-96f0-4f41-8056-5b3d6e9404ae_1456x695.gif) Spending time to improve your components is worth it. Every prototype made using them has better visual quality, is less distracting to stakeholders, and allows conversations to focus on the specific elements that are being tested. ### **Method 2: Chrome extensions** The next option for creating component libraries is to use Chrome extensions. As of today, [Magic Patterns](https://www.magicpatterns.com/) is the only tool that supports extracting components from a webpage directly. Using the extension is simple: Select a UI element, extract its styling, and turn it into a reusable component. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/b9eb6753-fee8-49a7-8701-43708be6b3a0_810x444.gif) I like to fire off multiple of these extracts at the same time to create as many components as I can in a short period of time. Once they land in Magic Patterns, you can continue to prompt your way through refinements. The real magic comes in using these components. Let’s try the prompt *“Build me a channel view of YouTube.*” Here’s the result from a single prompt: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/b4479d89-0d28-4082-aeb6-8bc7c6233495_810x405.gif) ### **Method 3: Code** The last way to build a component library is with code, using your real components. This approach requires the most effort but will provide you with prototypes that are indistinguishable from your real product. You’ll need assistance from engineering, and the overall effort and complexity will depend on your codebase and company size. This option is most realistic in smaller companies or those with more technical designers and PMs. To start, your engineering team will need to set up the repository so you can run the front-end application locally without having to connect to any back-end services or databases. This can be accomplished by creating a new entry point for the client-side app that mocks up the API responses. Once you have your local environment running, you can use an AI code editor like Cursor or Windsurf to prototype. Here’s an example of an app I’m working on. All of the information you see is mocked up client-side, with no servers or databases connected. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/4f3d0de7-2dac-4017-a94f-3074ae6f16ec_1600x876.gif) This approach works best if you’re comfortable with GitHub, branches, and running basic terminal commands. Another way to leverage code to get more accurate components is with the new [Figma MCP server](https://www.figma.com/blog/introducing-figmas-dev-mode-mcp-server/) (which just launched last week!). In case you’re unfamiliar, MCP is a protocol that allows AI agents to retrieve information or take actions in another application. In this case, we can use the Cursor agent to make calls to Figma’s MCP server to retrieve detailed styling information, and then transform that into components. Figma currently supports four MCP actions: 1. Get Code 2. Get Variable Definitions 3. Get Image 4. Get Code Connect This effectively gives Cursor the ability to autonomously take screenshots, extract design tokens, and get CSS from Figma’s Dev Mode. Here’s a quick side-by-side of a Figma [design library](https://www.figma.com/community/file/1267195373409722424/design-system) and the same components reproduced in code. Figma: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/69d956a0-c7af-400f-bd7f-250504bb946f_1128x922.png) Code: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/e59bc1bb-ef87-42d9-8064-532dcc3c6d21_810x468.gif) Here are the steps to build your components using Figma MCP: 1. Enable the Figma MCP server for the design you want to work with (Preferences, then Enable Dev Mode Server). This will provide you with a URL to connect to. 2. In Cursor, go to Settings, MCP Tools, and “Add a new MCP server.” Paste in the URL provided by Figma. 3. Copy a component URL from Figma by right-clicking and selecting “Copy as URL.” 4. Ask Cursor to generate this as a component, and to make your index page a list of components. So far, I’ve found this works best if you have *existing* components Cursor can mimic, or provide clear instructions for how the code should be formatted. This option is very new but also very exciting for design teams who want to leverage their existing component libraries with AI prototyping tools. To summarize, component libraries give your team a set of shared artifacts that make it easy for anyone to create realistic prototypes with AI. Your component library can be shared across your entire team, making it easy to onboard new team members and maintain high-quality visuals for testing directly with customers or in meetings with senior stakeholders. ## **Team workflows** Now that you have components to help you create realistic prototypes in a flash, let’s talk about team workflows. There are two workflows I want to cover: 1. Baselines and forks 2. Product development lifecycle ### **Workflow 1: Baselines and forks** Baselines and forks are an extension of the component libraries strategy. With components, you’re giving team members a head start by providing building blocks they can easily assemble into good-looking prototypes. “Baselines” go one step further, giving you a high-quality reproduction of your current product experience, allowing your team members to more easily experiment with new ideas using simple prompts. Forks build on this further, allowing you to duplicate a project and all of its contents without using any tokens, and keep you from screwing up your baseline. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/353ed3d8-4f36-492d-92fa-24131b3855ad_2912x2140.png) This is best explained with an example. Let’s say I’m the PM at Airbnb working on the new Experiences product. This is a product that features the most unique and interesting experiences around the world. Using baselines and forks, I’ll start by reproducing the current page and then explore a few new ideas. Creating the initial page with my component library took me about 20 minutes, mostly ironing out all the little details to make the prototype really look like Airbnb (see my [previous guide](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product) for advice on fine-tuning your prototypes). Here’s how it looks: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/e57b11d4-19a4-4279-9f3b-597c3abe7583_1456x722.jpeg) By the way, here’s the real site: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/036f085e-5a93-4a30-bf79-b3b85de845cb_1456x742.jpeg) Now that I have this page, I’ll treat it as my “baseline.” I won’t make any more changes and instead create a fork. On the fork, I’ll enter a single prompt: “*Modify this page so that it runs through a questionnaire to determine experiences that I would like before showing me experiences. Maintain the Airbnb branding.*” Here’s the result: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/3c86b02d-4bcd-49c4-a5bc-0098e3a5721f_1456x725.gif) Now let’s test another idea: using the customer’s past travel history to suggest relevant experiences. Again, I create a fork from the baseline. On the new copy, I prompt, *“Modify this page so it shows me experiences I would like based on my past travel history with Airbnb. For each recommendation, add a UI element that says ‘based on your trip to. . .”* Here’s the result: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/05a81208-a8fd-4be6-bdf0-6046efb2afdc_1456x797.jpeg) Baselines and forks are best used when you want to explore multiple different ideas without having to rebuild the starting page each time. They provide your team with another leg up in getting high-quality results quickly when using AI prototyping tools. ### **Workflow 2: Product development lifecycle** If you implement everything we’ve covered so far, you’ll be dramatically ahead of the average AI prototyping noob. But it’s very possible that your PMs or designers will still prototype in isolation. Creating a shared understanding for your team for when to use AI prototypes is just as important as knowing how to get good results. Before we get into the product development lifecycle, I want to briefly touch on fidelity. For a long time, we’ve had low-fidelity scratch-pad versions of UIs, and high-fidelity mocks for the polished finished product. AI prototyping introduces medium fidelity—better than a napkin drawing but still not as good as your finalized Figma mocks. Here’s an example of a mid-fi mock of Reddit. At first glance, it seems correct, but you’ll notice lots of imperfections, like two buttons for upvoting, too much padding on the left navigation, and the top right navigation items being pushed in too far from the right edge. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/2d3b5e7c-0dfe-4bd4-8143-9965e00dce24_1338x834.png) You can also definitely create high-fi mocks with AI prototyping tools—the key is how much time and effort you put into them. It’s critically important to choose the appropriate fidelity for your prototype based on the context, and then set clear expectations with team members around that level of fidelity. For example, if you’re explaining an interaction to an engineer, mid-fi is totally fine, and it’s a waste of time to push past that. Whereas if you’re pitching your CEO on funding a new team that will require millions in new investment, a functional high-fi mock is probably worth the time. Below, I’ve outlined the six main steps in the product development lifecycle and how prototypes can be leveraged, as well as what fidelity they should be and who is responsible for the prototype. I encourage you to take this as general guidance and apply what makes sense in your own company. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/afb538be-9085-4b2e-a14c-0466b0ffc881_2912x1654.png) Let’s walk through an example of the full development lifecycle for a new feature at Reddit called Reddit Answers. This will be a gen AI feature that allows users to ask questions and get quick answers based on past Reddit posts and comments. #### Step 1: Discovery Starting in Discovery, you may choose to use an AI prototyping tool to quickly express an idea in medium fidelity. This is typically used only internally, and may just be between product, design, and your engineering lead. We might start out with something like this, which should take only about 20 minutes: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/47b20c8f-ec09-4662-9eeb-6cf36716690b_810x514.gif) This is a great asset for internal communication, but we probably wouldn’t want to show this off to any executives or customers. By the way, here’s the prompt that generated this prototype: `# Reddit Answers PRD` `## Overview` `Reddit Answers is a new Q&A feature that leverages Reddit’s vast community knowledge to provide generated, fact-based answers to user questions, with references to specific posts and comments. It prioritizes highly upvoted, less controversial content and supplements Reddit information with external sources to enhance credibility.` `## User Stories` `As a user, I want to ask questions and receive comprehensive answers based on Reddit’s collective knowledge.` `As a user, I want to see which specific Reddit posts and comments contributed to the answer.` `As a user, I want to understand how reliable the Reddit-sourced information is compared with external sources.` `As a user, I want to access additional external links that provide more context to the answer.` `As a user, I want to navigate easily to the Reddit discussions referenced in the answer.` `As a developer, I want to access Reddit Answers data via API to enhance AI applications with Reddit’s knowledge.` `## Implementation Phases` `### Phase 1: Core Q&A Experience` `Question input interface in left nav` `Answer generation from Reddit content` `Basic reference linking to source posts/comments` `Simple reliability scoring` `Local answer history` `### Phase 2: Enhanced Credibility & Context` `Sophisticated reliability-scoring algorithm` `External source integration` `Content moderation systems` `Improved reference highlighting` `Personalized answer recommendations` `### Phase 3: Ecosystem Integration` `API access for AI companies` `Analytics dashboard for Reddit team` `Community feedback mechanisms` `Mobile optimization` `Enhanced search integration` `## Design System` `Colors:` `Primary: Reddit Orange (#FF4500)` `Secondary: Periwinkle Blue (#9494FF)` `Background: Light Gray (#F8F9FA)` `Text: Dark Gray (#1A1A1B)` `Typography:` `Reddit’s IBM Plex font family` `Spacing:` `Consistent 8px grid system` `##Components:` `Question Input` `Answer Card` `Reference List` `Reliability Score Badge` `External Link Cards` `Source Attribution Tag` `## Data Model` `### Question` `id: unique identifier` `text: the user’s question` `timestamp: when question was asked` `user_id: who asked the question` `category: auto-categorized topic area` `### Answer` `id: unique identifier` `question_id: reference to question` `text: generated answer content` `reliability_score: 0-100 score comparing Reddit info with external sources` `timestamp: when answer was generated` `### Reference` `id: unique identifier` `answer_id: reference to answer` `post_id: Reddit post referenced` `comment_id: specific comment referenced (optional)` `relevance_score: how relevant this reference is to the answer` `controversy_score: measure of how controversial the content is` `upvote_count: number of upvotes on referenced content` `### ExternalSource` `id: unique identifier` `answer_id: reference to answer` `url: link to external source` `title: title of the external content` `relevance_score: how relevant this source is to the answer` `domain_authority: credibility rating of the source domain` #### Step 2: Roadmap and alignment Once we’ve firmed up the idea and want to build more buy-in internally, we likely want to put something together in high fidelity that we can show stakeholders and customers. We’ll refine our initial prototype, use our component library, and come out with a much more polished prototype. This can take anywhere from 20 to 60 minutes, depending on the complexity of your prototype. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/f9e3a0d9-a2ab-456f-a53c-39e59900633c_810x533.gif) #### Step 3: PRD and mocks Now that we have buy-in to build this new feature, we need to figure out exactly how it will work. We can include the prior high-fidelity prototype inside the PRD and demo it live to the team to start generating questions and working out edge cases. We might hear things like: - What does the accuracy score mean? - What external sources should be used? - Is there any content within Reddit that should be excluded from answers? - How should answers be vetted before being included in the AI response? Using the prototype for communication is a much faster and easier way to drive these important conversations than hoping an engineer reads your whole PRD and adds a comment or two. This conversation with engineering will likely take two hours or more across a few discussions. #### Step 4: User interviews As we’re continuing to iterate on the idea, we can bring our high-fidelity prototype with us to user interviews. We can get feedback on the exact user flows very early in the development cycle instead of waiting for the feature to be ready in production. User interviews will likely take you three to five days to collect enough feedback. Although direct user interviews are likely the best way to gather feedback, we can also scale our feedback collection with surveys. You can embed your favorite survey tools directly into the prototype to collect structured feedback by sharing it with actual users. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/9af48762-269f-460d-a67d-ed188fefc7ff_810x533.gif) #### Step 5: Engineering scoping and delivery It’s important to remember that the code generated by AI prototyping tools is mostly useless to your engineering team (unless your components are built using your existing code). It doesn’t follow any of the existing patterns, use any of the same libraries, or may not even be written in the same programming language. Some ways the prototype can be useful are for documenting very specific interactions, like animations. If you want the loading animation to be exactly what you created in the prototype, your engineering team can likely use that as a head start during implementation. Throughout the building process, the high-fidelity prototype can still continue to be a useful communication tool any time there are questions about aspects of the product or feature. Prototyping doesn’t really impact development timelines, so expect two to six weeks, depending on the complexity of the feature. ## **Bonus tip: Using your actual logo in prototypes** This section is quick but super-important, because I see a lot of people run into this challenge. Most AI prototyping tools would rather draw an uncanny-valley version of your logo than use the real thing. With all the other small UI mistakes they make, this often makes the prototype feel unusable when presenting to customers or senior stakeholders. The simple solution to this is to paste in a link to the exact image you want to use. There are two ways to find that logo URL: 1. Right-click on the image, inspect element, and grab the URL or SVG for the image. 2. Use a logo repository, like Brandfetch, that hosts most company logos. Find yours and grab the embed link. Starting by grabbing the image directly, we can inspect the site and click on the logo. This image is an SVG, meaning there is no link to copy. We can instead copy the SVG content and paste that into our preferred AI tool. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/cde8d4c6-75bc-4f5a-ad49-c1daf45d6da5_264x249.png) Here’s the result in Bolt with a simple request to build a homepage: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/6ddc2bd3-9228-461d-96e9-436995afce5d_1263x237.png) To include your logo, simply paste the link or SVG text with a prompt like *“Add my logo to the header using this link.”* The second option is to look up the logo on a site like Brandfetch and copy the embed link. Paste that directly into your AI prototyping tool and you’ll get a clean logo instantly. ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/bfc51610-0712-447a-a4b9-f4bc6adc8d80_1373x623.png) As you’d expect, this looks identical: ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/2790a01c-717a-4e7a-9234-3ce32216b2d7_1271x257.png) You can use these approaches for any images you want to include—just make sure you’re linking to the image itself, not a webpage. The URL will typically end in PNG or SVG. One last tip: If you need some realistic images for your prototype, just ask the AI tool to grab them from [Unsplash](https://unsplash.com/), which does a reasonably good job at picking relevant images. Here’s the Airbnb example once more, where the AI decided on these images based on the prompt *“Add images from Unsplash that make sense given the context.”* ![Image from How to get your entire team prototyping with AI](https://substack-post-media.s3.amazonaws.com/public/images/46318531-4992-4093-b086-ab4d69afc502_1456x797.jpeg) # **Where to go from here** If you want to drive adoption on your team, start with component libraries to establish visual consistency, implement baselines and forks to accelerate iteration, and align your team on when and how to use different fidelity levels throughout your product development lifecycle. These aren’t merely nice-to-haves—this is what separates teams just poking at the surface of what AI can do for them from those fully integrating AI tools into their best practices and daily workflows to supercharge their products. Give these workflows a try, and feel free to reach out with your progress. I’m excited to see you and your team level up with AI prototyping! *Thanks, Colin!* *For more from Colin, check out his [free 45-minute live online class](https://maven.com/p/becb6f/product-development-lifecycle-with-ai-prototyping) coming up on July 8th, and his course [AI Prototyping for Product Managers](https://maven.com/tech-for-product/ai-prototyping-for-product-managers?promoCode=LENNYSLIST) (which kicks off on June 30th; use code LENNYSLIST to get $100 off). You can also [book Colin for one-day team workshops](https://teams.techforproduct.com/) with your team.* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [23/46] How tech’s most resilient workers handle burnout Three weeks ago, we shared the results of our first-ever large-scale [tech worker sentiment survey](https://www.lennysnewsletter.com/p/how-tech-workers-really-feel-about). The post generated [tons of conversations](https://www.linkedin.com/search/results/content/?keywords=lenny+tech+worker+sentiment+burnout&origin=SWITCH_SEARCH_VERTICAL&sid=AF5), in particular about burnout. Readers asked, “Who *are* these people who *aren’t* burned-out?” **So we decided to do a timely follow-up investigation to find out the common threads among respondents who:** 1. Haven’t felt much (or any) burnout in recent months/years 2. *Have* felt burned-out but were able to deal with it *extremely effectively* 3. Feel like they’ve cracked the “secret code” for dealing with burnout ![Image from How tech’s most resilient workers handle burnout](https://substack-post-media.s3.amazonaws.com/public/images/11ddfea8-947c-46b6-8f21-6146f5137c70_1600x1242.png) We ran open-ended, video-based surveys with about 175 respondents and interviewed about 15 people who reported low to no burnout in the original survey. Our respondents revealed that dealing with burnout isn’t simply about resilience or stress management. **The key is to systematically design a career and lifestyle that make burnout structurally unlikely.** We call tech workers who manage to do this “burnout conquerors.” *P.S. If you prefer, you can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* # Introducing ARMOR: five shields against burnout ![Image from How tech’s most resilient workers handle burnout](https://substack-post-media.s3.amazonaws.com/public/images/d965e5d4-23ba-47f0-af77-1a826b9b4212_2913x1991.png) My investigation revealed five core strategies for dealing with burnout. These strategies are the armor that tech workers are donning to protect themselves from burnout. ## **A: Autonomy** Many people we heard from immediately identified that autonomy is a critical ingredient to being happy at work. ![Image from How tech’s most resilient workers handle burnout](https://substack-post-media.s3.amazonaws.com/public/images/f9fe8588-bea0-4f43-bb41-dab4a6eeb394_1200x1374.png) We also saw this again and again in our research. > *“I’ve had strong leaders who believed in my development and let me run with high autonomy to solve challenging problems. I’m happy more than 95% of the time.”* > *“I intentionally seek out environments where I have autonomy and variety in my day-to-day work, which keeps things fresh and prevents stagnation. Ultimately, choosing a path that matches my strengths and curiosity has made all the difference in sustaining my motivation and well-being.”* Burnout conquerors continuously strive to create as much autonomy over their work lives as possible. Here’s how: #### **1. Proactively take control of your time and efforts** Burnout conquerors keep a keen eye on how they spend their time and energy at work to make sure they’re working smarter, not harder. An open block in their calendar to do deep work isn’t just going to drop in from the sky. Their most productive hours aren’t going to magically become available for priority work. Burnout conquerors make their days their own. > *“For me, burnout happens when I become ineffective at my job, no matter how many hours I put in. When I start to feel like I’m not making progress, I immediately start examining why. A perfect example is right now—I’ve not moved the needle with my priorities in weeks. There’s been a lot of firefighting, but nothing proactive. I started to examine what I’m spending time on and how I can pivot to spending time on what I should be doing. Calendar audit was step #1, declined meetings I really didn’t need to be in, blocked off heads-down time, and, in a matter of days, I’m back on track.”* #### **2. Hit your goals to build trust** Rather than expecting trust in the workplace, burnout conquerors earn it by consistently delivering regardless of the external circumstances. The more you deliver on what you promised, the more everyone will trust you and leave you alone to do what you need to do. But that trust is not granted; it’s earned through behavior and successes (which is accomplished by taking control of time!). > *“I control what I do; I control my success. I’m responsible for my failures. Some days are hard, and some people are difficult, but I choose how to respond and handle the situation to deliver on my goals.”* To get this done, burnout conquerors focus on small sets of *quality* outcomes driven by highly targeted goals, which they *communicate up, down, and all around*. #### **3. Overcommunicate your needs to others—and yourselves** Burnout conquerors understand their own needs and communicate them clearly and appropriately to the people around them. They know that when people know what to expect of you, they are rarely caught off-guard and feel more comfortable letting you take the wheel. A common tactic for burnout avoidance we heard is to “overcommunicate” about everything. That includes far more than just what you’re working on and what resources you need to get it done. > *“Most of my peers focus on communicating what they’re working on. They forget about [communicating] the ‘hows’: their boundaries, their constraints, and their strengths.”* Burnout conquerors overcommunicate how to best work with them and how they like to communicate, receive feedback, and make decisions. Set the foundations of those conversations with a “How I Work” document. Find a template [here](https://lg.substack.com/p/the-looking-glass-a-user-guide-to). (Also, learn to [manage up](https://www.lennysnewsletter.com/p/managing-up) generally.) ## **R: Rock-solid boundaries** Burnout conquerors know that to ruthlessly protect their time and energy, they need to erect boundaries that can withstand the pressure of a fast-paced and results-driven workplace. Here’s how: #### **1. Set, automate, and, yes, overcommunicate clear work-life boundaries** Burnout conquerors create consistent time-based (e.g. “I’m ending the day at 6 p.m. to be with my kids”) or task-based (e.g. “I’ll focus on 3 to 5 essential and urgent tasks each day”) hard stops. Anything beyond that time or list gets pushed to another time (or [delegated to AI agents](https://www.lennysnewsletter.com/p/make-product-management-fun-again)). > *“I try to be strict about when I’m* not *going to do work. More than generic advice about ‘disconnecting,’ setting really clear micro-rules for myself has helped me prevent work stress seeping into the rest of my life. I know it might sound sad to even need these micro-boundaries, but it was necessary, and it’s been incredibly helpful.”* To ensure that those boundaries hold, the people we heard from automate their after-hours responses and set clear calendar blocks, so people get notified when they’re unavailable. They also share their working schedule with their colleagues up front, including daily constraints and personal needs, and then as needed when those boundaries get tested (which they always do). #### **2. [Get comfortable saying no](https://www.lennysnewsletter.com/p/how-to-say-no)** Boundaries will always get tested by new requests or projects—some of which are exciting and seriously appealing! **But there’s** ***always*** **an opportunity cost.** Burnout conquerors understand that saying yes to something means saying no to something else. They reframe no’s as tradeoffs—to themselves and others. > *“I don’t frame responses to colleagues with open-ended questions that block my ability to define a ‘ready’ state. To overly simplify, I don’t say, ‘I have options A or B. Which do you want?’ I say, ‘I have option A and B. Here’s how we got to these options. Here are the tradeoffs. Please let me know your thoughts. I’ll move forward with A if I don’t hear any new feedback.’”* > *“I read this book called* You Have a Choice *by Eric Nehrlich. One of the big unlocks for me was ‘You don’t have to do anything you don’t want to . . . as long as you’re okay with the consequences.’”* #### **3. Take** ***real*** **breaks: rhythmic, reflective, restorative** Breaks are necessary and reasonable for the roles and responsibilities we all have in this industry. Performing at a high level is essential at work, but turning off is too. > *“Someone once told me, ‘We’re not performing brain surgery,’ and it was such a good reality check. Unless you’re a literal brain surgeon, it’s OK to not be perfect and available all the time. When I’m on, I’m on 100%. And when I’m off, I’m off 100%.”* We heard a lot about the types of breaks that are successful in helping people prevent burnout. We categorized them into three types. ![Image from How tech’s most resilient workers handle burnout](https://substack-post-media.s3.amazonaws.com/public/images/e7cb9bf1-68a9-44bd-a273-8d436a19b90e_2631x3934.png) ## **M: Maintenance rituals** Burnout conquerors treat their health and wellness as critical professional infrastructure. We’ve all seen the advice on walking 10,000 steps a day or trying to get 7 to 8 hours of sleep. But the people in tech who are beating burnout go one level deeper. Here’s how: #### **1. Develop powerful “early warning” abilities** Burnout conquerors are incredibly in tune with the physical and emotional signals of burnout. They’re listening to the body’s stress indicators, aware of energy levels and mood changes, and carefully log sleep disruptions, tension, and other manifestations, like low energy. > *“One thing that I learned in the mental health program during my leave of absence was how to check in daily with my emotions. This sounds pretty basic, but it isn’t a practice people are typically skilled at. Paying attention to our bodies and feelings is critical to well-being and spotting early signs of burnout. I also created a coping ladder of different tactics to use when I noticed my emotions were elevated.”* When people understand their baseline, they can detect when something’s off internally sooner than others might. The faster they know something’s wrong, the faster they can make it right. #### **2. Monitor your health signals, maintain your wellness** To keep that early-warning system sharp, burnout conquerors track all the things, like their sleep, nutrition, and exercise. But interestingly, our participants don’t get overly technical about tracking. They use *heuristics*: **Sleep:** > *“I track the number of nights with disrupted sleep. One night of occasional insomnia is OK, two nights is a sign of trouble, three nights is a major problem.”* **Nutrition:** > *“Tracking nutrition can be challenging, so I use LLMs to build my weekly menus, and it gives me information on macros. All I have left to do is count how many times I cheated. . .”* **Exercise:** > *“If I’m not getting enough steps in a day, I know something is amiss. It’s happening right now and I’m actively tackling it.”* Wellness is treated as a non-negotiable business need, not optional self-care. No weeks off, scheduled sleep times, and a strong focus on consistency and intensity. Many of the exercise routines participants shared felt more like Dwayne “The Rock” Johnson’s than a tech worker’s exercise routine: weights 5 or 6 times a week, marathon training, mindfulness retreats, biohacking. Burnout conquerors are spending *significant* time outside of work in training. > *“I go to bed at the same time every night. We get up at the same time every morning.”* > *“I cook all our food. The biggest difference for us was when we stopped eating out during the week. Eating out is a weekend treat.”* > *“Turn off. And by turning off I don’t mean scroll social media—instead, every day, meet people, do something manual (cook, build, iron your clothes), read a paper book.”* > *“I train 5 to 6 times a week (gym, run, padel, etc.). I have a fixed sleep habit, from 10 p.m. to 6-7 a.m. I eat simple. The intensity of my regimen helps to balance the intensity and speed of work.”* #### **3. Be open to all forms of therapy** It really stood out in the data that burnout conquerors seem comfortable connecting with their emotions, going to therapy, and taking leave when necessary. Many of our participants consider therapy and mental health support as preventive care. They ignore stigmas around such support and leverage their companies’ mental health benefits to the fullest. *“My secret weapon against burnout is rest and time away from work. I requested a mental health leave of absence this fall . . . I sought out additional mental health support. One thing that I learned how to do in the mental health program was to check in daily with my emotions.”* ## **O: Original thinking** Burnout conquerors don’t accept conventional organizational expectations of employees. They don’t buy into the common corporate narrative that puts the onus on employees to deal with burnout. And they don’t feel guilty if they’re feeling bad or in need of a break. The people who are actively preventing burnout have a strong belief system that insulates them from falling victim to burnout-related organizational mind games. Here’s how: #### **1. Understand that burnout isn’t your fault** While burnout conquerors take actionable steps toward preventing burnout in their own lives, they also know that the kind of deep exhaustion so many in the industry are feeling should not be blamed on the people suffering from it. Burnout is “organizational warfare,” not a personal responsibility. > *“I’m super-annoyed by the common expectation that you have to deal with your burnout on your own, as if it’s a personal problem and not an organisational problem. Burnout is not a personality problem; it’s a result of broken processes, unclear expectations, bad incentives (e.g. dumb performance review cycles and expectations), and poor work conditions . . . and no, those free-pizza Fridays and free mental health sessions are not company benefits. They are band-aids and signs of serious organisational dysfunction.”* The system is broken, and it’s not your job to fix yourself to fit a toxic system. #### **2. Don’t assume that working harder always yields better results (karma’s involved)** Burnout conquerors reject the fundamental meritocracy belief of tech—that effort directly correlates to outcomes. Instead, they embrace uncertainty, karma, and trusting that “things work out” regardless of individual grinding. > *“I think the conventional wisdom for many people is that if you work harder, you achieve greater results. I also believed this when I was younger. And then, with a lot of different cycles in life, I realized there are a lot of other things that play a role. There’s some karma, you know . . . I believe things will get right in the long run if my intention, my overall direction, is correct.”* #### **3. Feel feelings instead of optimizing them away** Burnout conquerors don’t buy into the optimization culture of tech. When bad feelings arrive—and they always do at some point—they know it’s never more efficient to push them away to try to maximize consistent output. Instead, as one respondent says, “sometimes you need to just be human and feel terrible for a while.” > *“Generally, all advice sounds really good in theory, but it’s really hard to put it into practice, especially if people are already under pressure. Telling someone who needs rest to exercise regularly can just add to their stress . . . I think sometimes we just need to also fully feel what we feel—if we’re stressed, acknowledge that; if you need to cry, cry; if you need to take a couple of days to just lie on the couch and eat bad food, do it. As long as you set a light at the end of the tunnel for yourself.”* ## **R: Role architecture** Burnout conquerors design their careers strategically from day one. That means choosing the right *company* and the right *role.* To make these decisions, burnout conquerors have rubrics for evaluating the best match for their needs. Here’s how: #### **1. The relationship test** Burnout conquerors realize that they can only thrive in an environment where they’re surrounded by people they want to be surrounded by. > *“Who you work with matters more than almost anything else about the job.”* > *“Choose your job based on who you work with. Period.”* > *“I prioritize who I work for and work with above all else.”* #### **2. The Sunday test** They follow their gut feelings and take the “Sunday scaries” seriously. If they’re excited for the challenge of the next week and looking forward to working with colleagues, those are great signs. If they’re not . . . it’s not. > *“A great litmus test is how I feel on Sunday.”* #### **3. The energy test** They optimize for “energy production.” Burnout conquerors seek meaning, challenge, and balance in their jobs and careers. > *“I’ve intentionally pursued work that gives me energy. I exit situations that drain my energy.”* > *“If I’m not passionate about the thing that is interrupting my personal life, then I find something I am passionate about.”* #### **4. The values alignment test** They reflect on what they really, really want, applying the “Five Whys” framework to their career. They don’t get enamored by shiny objects but, rather, focus on how a company’s culture aligns with their core values. > *“Is there anyone around me who is burned-out as well? Are there people who don’t respect my boundaries?”* > *“Work cultures where you have ownership. Being trusted to make decisions and do good work is energizing.”* ## A final piece of advice: Your work reality can only be as good as your relationships The last question in my survey or interview was *“What’s* THE *career choice that’s helped you avoid getting burned-out?”* One set of responses stood out to me: how critical it is to prioritize *people and relationships* at home and at work. In one of my interviews, I spoke to [Joshua Herzig-Marx](https://www.linkedin.com/in/joshuaherzigmarx/). He told me that for him, partner or spouse involvement is crucial for managing burnout. His partner is a sounding board and collaborator. To him, solo burnout management is like trying to product manage without partnering with stakeholders. Partnership enables perspective sharing, empathetic listening, and brainstorming solutions. There was also a consensus among our burnout conquerors that, at work, who you surround yourself with matters more than anything else. > *“Don’t work with assholes, because one way or another, assholes lead to people around them burning out, either directly or indirectly.”* ## Check your own ARMOR strength To find out how protected you are from the burnout that’s affecting the majority of tech workers, answer these questions for yourself now and any time you feel those feelings creeping in: ![Image from How tech’s most resilient workers handle burnout](https://substack-post-media.s3.amazonaws.com/public/images/2d14d6c3-5da3-46b4-b428-65fd5ee70765_2631x5907.png) If you’d like to assess your burnout, take our burnout check [here](https://theburnoutcheck.com/). And if you answered no to the above questions, tried the suggestions from our burnout conquerors, and are still dragged down by that wiped-out feeling, it may be time to consider a bigger change. Special thanks to the nearly 200 tech professionals who shared their insights *again* for this follow-up research. Your candid feedback will help us deal with burnout in the tech industry. 🙏 *Thank you, Noam! You can find Noam on [LinkedIn](https://www.linkedin.com/in/noamsegal/) and [X](https://x.com/noamseg).* *Have a fulfilling, productive, and non-burnout-inducing week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [24/46] An AI glossary You’re probably hearing a lot of AI jargon, and you probably *sort* *of* know what some of it means . . . but not *really*. Below is an “explain it to me like I’m 5” definition of the 20+ most common AI terms, drawn from my own understanding, a bunch of research, and feedback from my most AI-pilled friends. If you already know all this, no sweat, this post isn’t for you. For everyone else, keep the following list handy next time you’re in a meeting and you’re struggling to keep up with all the AI words flying around the room. I’ll continue adding to this list as new buzzwords emerge. *P.S. If you prefer, you can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from An AI glossary](https://substack-post-media.s3.amazonaws.com/public/images/3ded72ed-e8da-43dd-ba53-50eb67ca8518_2912x1938.png) ### **Model** An AI model is a computer program that is built to work like a human brain. You give it some input (i.e. a prompt), it does some processing, and it generates a response. Like a child, a model “learns” by being exposed to many examples of how people typically respond or behave in different situations. As it sees more and more examples, it begins to recognize patterns, understand language, and generate coherent responses. There are many different types of AI models. Some, which focus on language—like [ChatGPT o3](https://openai.com/index/introducing-o3-and-o4-mini/), [Claude Sonnet 4](https://www.anthropic.com/claude/sonnet), [Gemini 2.5 Pro](https://deepmind.google/models/gemini/pro/), [Meta Llama 4](https://www.llama.com/), [Grok 3](https://grok.com/), [DeepSeek](https://www.deepseek.com/), and [Mistral](https://mistral.ai/)—are known as large language models (LLMs). Others are built for video, like [Google Veo 3](https://deepmind.google/models/veo/), [OpenAI Sora](https://openai.com/sora/), and [Runway Gen-4](https://runwayml.com/). Some models specialize in generating voice, such as [ElevenLabs](https://elevenlabs.io/), [Cartesia](https://cartesia.ai/), and [Suno](https://suno.com/). There are also more traditional types of AI models, such as classification models (used in tasks like fraud detection), ranking models (used in search engines, social media feeds, and ads), and regression models (used to make numerical predictions). ### **LLM (large language model)** LLMs are text-based models, designed to understand and generate human-readable text. That’s why the name includes the word “language.” Recently, most LLMs have actually evolved into “multi-modal” models that can process and generate not just text but also images, audio, and other types of content within a single conversational interface. For example, all of the ChatGPT LLM models natively support text, images, and even voice. This started with GPT-4o, where “o” stands for “omni” (meaning it accepts any combination of text, audio, and image input). [Here’s a really good primer on how LLMs actually work](https://every.to/p/how-ai-works), and [also this popular deep dive by Andrej Karpathy](https://youtu.be/zjkBMFhNj_g): [Watch on YouTube](https://www.youtube.com/watch?v=zjkBMFhNj_g) ### **Transformer** The transformer architecture, developed by Google researchers in 2017, is the algorithmic discovery that made modern AI (and LLMs in particular) possible. Transformers introduced a mechanism called “attention,” where instead of only being able to read text word‑by‑word, sequentially, the model is able to pay attention to all the words at once. This helps the models understand how words relate to each other, making them far better at capturing meaning, context, and nuance than earlier techniques. Another big advantage of the transformer architecture is that it’s highly parallelizable—it can process many parts of a sequence at the same time. This makes it possible to train much bigger and smarter models simply by scaling up the data and compute power. This breakthrough is why we suddenly went from basic chatbots to sophisticated AI assistants. Almost every major AI model today, including ChatGPT and Claude, is built on top of the transformer architecture. [This is the best explanation of transformers](https://ig.ft.com/generative-ai/) I’ve seen. Here’s also a more technical and visual deep dive: [Watch on YouTube](https://www.youtube.com/watch?v=KJtZARuO3JY) ### **Training/Pre-training** Training is the process by which an AI model learns by analyzing massive amounts of data. This data might include large portions of the internet, every book ever published, audio recordings, movies, video games, etc. Training state-of-the-art models can take weeks or months, require processing terabytes of data, and cost hundreds of millions of dollars. For LLMs, the core training method is called “next-token prediction.” The model is shown billions of text sequences with the last “token” (roughly analogous to a word, see definition of *token* below) hidden, and it learns to predict what word should come next. As it trains, the model adjusts millions of internal settings called “weights.” These are similar to how neurons in the human brain strengthen or weaken their connections based on experience. When the model makes a correct prediction, those weights are reinforced. When it makes an incorrect one, they’re adjusted. Over time, this process helps the model improve its understanding of facts, grammar, reasoning, and how language works in different contexts. [Here’s a quick visual explanation](https://www.youtube.com/watch?v=rEDzUT3ymw4). If you’re skeptical of next-token prediction generating novel insights and super-intelligent AI systems, here’s Ilya Sutskever (co-founder of OpenAI) explaining why it’s deceptively powerful: [Watch on YouTube](https://www.youtube.com/watch?v=YEUclZdj_Sc) ### **Supervised learning** Supervised learning refers to when a model is trained on “labeled” data—meaning the correct answers are provided. For example, the model might be given thousands of emails labeled “spam” or “not spam” and, from that, learn to spot the patterns that distinguish spam from non-spam. Once trained, the model can then classify new emails it’s never seen before. Most modern language models, including ChatGPT, use a subtype called “self-supervised learning.” Instead of relying on human-labeled data, the model creates its own labels, generally by hiding the last token/word of a sentence and learning to predict it. This allows it to learn from massive amounts of raw text without manual annotation. ### **Unsupervised learning** Unsupervised learning is the opposite: the model is given data without any labels or answers. Its job is to discover patterns or structure on its own, like grouping similar news articles together or detecting unusual patterns in a dataset. This method is often used for tasks like anomaly detection, clustering, and topic modeling, where the goal is to explore and organize information rather than make specific predictions. ### **Post-training** Post-training refers to all of the additional steps taken after training is complete to make the model even more useful. This includes steps like “fine-tuning*”* and “RLHF.*”* ### **Fine-tuning** Fine-tuning is a post-training technique where you take a trained model and do additional training on specific data that’s tailored to what you want the model to be especially good at. For example, you would fine-tune a model on your company’s customer service conversations to make it respond in your brand’s specific style, or on medical literature to make it better at answering healthcare questions, or on educational content for specific grade levels to create a tutoring assistant that explains concepts in age-appropriate ways. This additional training tweaks the model’s internal weights to specialize its responses for your specific use case, while preserving the general knowledge it learned during pre-training. Here’s an awesome technical deep dive into how fine-tuning works: [Watch on YouTube](https://www.youtube.com/watch?v=eC6Hd1hFvos) ### **RLHF (reinforcement learning from human feedback)** RLHF is a post-training technique that goes beyond next-token prediction and fine-tuning by teaching AI models to behave the way humans want them to—making them safer, more helpful, and aligned with our intentions. RLHF is a key method used for what’s referred to as “alignment.” This process works in two stages: First, human evaluators compare pairs of outputs and choose which is better, training a “reward model” that learns to predict human preferences. Then, the AI model learns through reinforcement learning—a trial-and-error process where it receives “rewards” from the reward model (not directly from humans) for generating responses the reward model predicts humans would prefer. In this second stage, the model is essentially trying to “game” the reward model to get higher scores. [Here’s a great guide](https://www.labellerr.com/blog/reinforcement-learning-from-human-feedback/), plus this technical deep dive into RLHF: [Watch on YouTube](https://www.youtube.com/watch?v=T_X4XFwKX8k) ### **Prompt engineering** Prompt engineering is the art and science of crafting questions (i.e. “prompts”) for AI models that result in better and more useful responses. Like when you’re talking to a person, the way you phrase your question can lead to dramatically different responses. The same AI model will give very different responses based on how you craft your prompt. There are two categories of prompts: 1. Conversational prompts: What you send ChatGPT/Claude/Gemini when you’re having a conversation with it 2. System/product prompts: The behind-the-scenes instructions that developers bake into products to shape how the AI product behaves Here’s a podcast episode from just last week where we cover this and much more: [Watch on YouTube](https://www.youtube.com/watch?v=eKuFqQKYRrA) ### **RAG (retrieval-augmented generation)** RAG is a technique that gives models access to additional information at run-time that they weren’t trained on. It’s like giving the model an open-book test instead of having it answer from memory. When you ask a question like “How do this month’s sales compare to last month?” a retrieval system is able to search through your databases, documents, and knowledge repos to find pertinent information. This retrieved data is then added as context to your original prompt, creating an enriched prompt that the model then processes. This leads to a much better, more accurate answer. One common source of “hallucinations” is when you don’t give the model the context it needs to answer your question through RAG. Broadly, to summarize: - **Pre-training:** Teaches the model general knowledge (and language) - **Fine-tuning:** Specializes the model for specific tasks - **RLHF:** Aligns the model with human preferences - **Prompt engineering:** The skills of crafting better inputs to guide the model toward the most useful outputs - **RAG:** A technique that retrieves additional relevant information from external sources at run-time to give the model up-to-date or task-specific context it wasn’t trained on Here’s a great overview of fine-tuning vs. RAG vs. prompt engineering: [Watch on YouTube](https://www.youtube.com/watch?v=zYGDpG-pTho) ### **Evals** Evals (short for “evaluations”) are structured ways to measure how well an AI system performs on specific tasks, such as correctness, safety, helpfulness, or tone. They define what “good” looks like for your AI system and help you answer the question: Is this model doing what I want it to do? Evals are basically unit tests or benchmarks for your AI product. They run your model through predefined inputs and compare its responses against expected outputs. This helps you quantify progress, catch regressions, and guide you toward improvements. For example, here’s what an eval to measure the toxicity and tone of a model’s response might look like. Your model output would be inserted into the {text} variable: ![Image from An AI glossary](https://substack-post-media.s3.amazonaws.com/public/images/9cda3b44-8466-4878-b18d-197fc702688b_2913x1703.jpeg) Evals are often described by top product leaders as the most critical, yet overlooked, skill for building successful AI products: ![Image from An AI glossary](https://substack-post-media.s3.amazonaws.com/public/images/f097be41-cdc9-42be-bf6f-7221b653b871_2913x4511.jpeg) [Don’t miss this excellent guest post by Aman Khan](https://www.lennysnewsletter.com/p/beyond-vibe-checks-a-pms-complete) that teaches you how to craft evals for your product. ### **Inference** Inference is when the model “runs.” When you ask ChatGPT a question and it generates a response, that’s it doing inference. ### **MCP (model context protocol)** MCP is a recently released open standard that allows AI models to interact with external tools—like your calendar, CRM, Slack, or codebase—easily, reliably, and securely. Previously, developers had to write their own custom code for each new integration. MCP also gives the AI the ability to take *actions* through these tools, for example, updating customer records in Salesforce, sending messages in Slack, scheduling meetings in your calendar, or even committing code to GitHub. It’s still early in the definition of AI protocols, and there are other competing proposals, like A2A from Google and ACP from BeeAI/IBM. Here’s a really nice in-depth explanation of MCP: [Watch on YouTube](https://www.youtube.com/watch?v=N3vHJcHBS-w) ### **Gen AI (generative AI)** Gen AI refers to AI systems that can create new content, such as text, images, code, audio, or video. This is opposed to models that just analyze or classify data, such as spam detection, fraud analysis, or image recognition models. ### **GPT (generative pre-trained transformer)** “GPT” captures the three key elements behind how state-of-the-art LLMs like ChatGPT 4.1, Claude Opus 4, Llama 4, and Grok 3 work: 1. **Generative:** The model doesn’t just classify or analyze—it can generate new content. 2. **Pre-trained:** It first learns general language patterns by being trained on massive amounts of text (as described above) and can then be fine-tuned for more specific tasks. 3. **Transformer:** This refers to the breakthrough architecture (described above) that allows the model to understand context, relationships, and meaning across language. The combination of these three ideas—generation, large-scale pre-training, and the transformer architecture—is what made tools like ChatGPT feel intelligent, coherent, and surprisingly useful across a wide range of tasks. ### **Token** A token is the basic unit of text that an AI model understands. For LLMs, this is sometimes a word, but often just a part of a word. There are analogous concepts like “patches” for image models and “frames” for voice models. For example, “ChatGPT is smart.” might be split into the tokens “Chat,” “GPT,” “is,” “smart,” and “.” Even though “ChatGPT” is one word, the model breaks it into smaller pieces to make learning language more scalable, flexible, and efficient. [This explanation of transformers](https://ig.ft.com/generative-ai/) also explains tokens really well, and [here you can see how top models tokenize words](https://tiktokenizer.vercel.app/). Here’s the paragraph above tokenized by GPT-4o: ![Image from An AI glossary](https://substack-post-media.s3.amazonaws.com/public/images/d3088588-455d-4845-9430-7957104ffa7b_1160x550.png) ### **Agent** An agent is an AI system designed to take actions on your behalf to accomplish a goal. Unlike chatbots like Claude or ChatGPT, which take a prompt and quickly respond with an answer, an agent can plan, work step-by-step, and use external tools, often across many apps or services, to achieve some outcome you set for it. To borrow a definition from [this recent guest post, “Make product management fun again with AI agents,”](https://www.lennysnewsletter.com/p/make-product-management-fun-again) it’s best to think of the term “agent” as a spectrum, where AI systems become “agentic” the more of the following behaviors they exhibit: 1. **Acts proactively**,as opposed to waiting to be prompted 2. **Makes its own plan**,as opposed to being given instructions 3. **Takes real-world action**, such a**s** updating a CRM, running code, or commenting on a ticket—as opposed to only sharing recommendations 4. **Draws on live data**, such as a web search or a customer support queue—as opposed to relying on static training or your manually uploading a file 5. **Creates its own feedback loop**,where it watches its own output and iterates without human assistance [Here’s a great guide from Anthropic](https://www.anthropic.com/engineering/building-effective-agents) for how to build effective agents. ### **Vibe coding** The term originated from [this tweet](https://x.com/karpathy/status/1886192184808149383) by Andrej Karpathy: ![Image from An AI glossary](https://substack-post-media.s3.amazonaws.com/public/images/c479abab-b22e-4d16-b666-b28492803ee1_1190x848.png) Vibe coding has come to mean building apps using AI tools like Cursor, Windsurf, Bolt, Lovable, v0, or Replit, by describing what you want in plain English (i.e. prompts) rather than writing code. In many cases, you never look at the code at all. ### **AGI (artificial general intelligence)** AGI refers to AI being “generally” smart—not just good at narrow tasks like coding, math, or data analysis, but capable of performing a wide range of tasks well, as well as learning how to tackle new problems without needing specialized training. When people talk about reaching AGI, they usually mean the point where AI is more intelligent than the average human across most subjects. Some people believe we’ve already reached this point. Artificial “superintelligence” (ASI) refers to the next step beyond AGI—AI that is much more intelligent than the best human minds in virtually every domain. We don’t believe we have reached this point yet, and there are debates about whether AGI to ASI will be [a fast or slow takeoff](https://www.lesswrong.com/w/ai-takeoff). [Watch on YouTube](https://www.youtube.com/watch?v=SEkGLj0bwAU) ### **Hallucination** A hallucination is when an AI model generates a response that sounds confident but is factually incorrect or entirely made up. This happens because the model doesn’t actually “know” facts or look things up in a database. Instead, it generates responses by predicting the most likely next token/word based on patterns in its training data. When it lacks the right information, it may confidently fill in the gaps with something that sounds plausible but isn’t real. The good news is that newer models are getting better at avoiding hallucination, and there are proven strategies—like RAG and prompt engineering—that help mitigate the risk ([here’s a guide from Anthropic](https://docs.anthropic.com/en/docs/test-and-evaluate/strengthen-guardrails/reduce-hallucinations)). ### **Synthetic data** To train ever more intelligent models, you need ever more data. But once the models are trained on the entire internet, every book ever published, every recording, dataset, etc., how do we give them more data? Part of the answer is “synthetic” data. Synthetic data is artificially generated data. It follows the same patterns and structure as human-generated data, and, amazingly, it can be just as effective at helping models learn. It’s valuable when real data is limited, sensitive, or fully exhausted. Synthetic data is generated differently depending on the type of data: 1. **Text:** LLMs are prompted to generate fictional customer support chats, medical notes, or math problems based on real-world examples. 2. **Images:** [Diffusion models](https://huggingface.co/learn/computer-vision-course/en/unit10/datagen-diffusion-models) and [GANs](https://www.impetus.com/resources/blog/synthetic-data-generation-using-gans/) create realistic visuals, from street scenes to x-rays to product photos, without copying actual images. 3. **Audio:** Voice and sound models synthesize speech, background noise, or music that mimic real recordings. To a human, synthetic data can often be indistinguishable from real data, for example, a chatbot transcript that seems authentic but was entirely generated. [Watch on YouTube](https://www.youtube.com/watch?v=ZPPBujNssnU) Any other AI jargon you’d like to see explained? Did I get something wrong? Drop me a comment 👇 [Leave a comment](https://www.lennysnewsletter.com/p/an-ai-glossary/comments) ### Bonus: Some of my additional favorite videos [Watch on YouTube](https://www.youtube.com/watch?v=eMlx5fFNoYc)[Watch on YouTube](https://www.youtube.com/watch?v=LCEmiRjPEtQ) *Thank you to [Dennis Yang](https://www.linkedin.com/in/dennisyang/), [Hilary Gridley](https://www.linkedin.com/in/hilarygridley/), and [Aman Khan](https://www.linkedin.com/in/amanberkeley/) for feedback on this post.* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [25/46] The definitive guide to mastering analytical thinking interviews In my ongoing efforts to help you land your dream job, I’m excited to bring you this truly epic guide by [Ben Erez](https://www.linkedin.com/in/benerez). Below you’ll find everything you need to know to nail your analytical thinking interview—a staple of most product interview loops. I’ve never seen a guide this in-depth, specific, and full of so many real-life examples. I hope this helps them see exactly why you’re the perfect fit for the role. *After a decade in product (at Facebook, as the first PM at three different startups, and as a founder), Ben is now a full-time interview coach, helping PMs land their dream roles. He teaches a [top-rated course on PM interviewing](https://bit.ly/4iQm5xq) (use code “LENNYSLIST” to get $100 off), and on July 9th, he’ll be hosting a free 30-minute lightning lesson where he’ll show you [how to use AI to practice for your analytical thinking interviews](https://maven.com/p/f6c75d/practice-for-analytical-thinking-pm-interviews-with-ai). Ben also co-hosts [Supra Insider](https://suprainsider.substack.com/), a weekly podcast for the product community.* *Bonus: You can now listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/7b4623f5-4826-495f-b145-08183f374d19_2912x2394.jpeg) In part one of this comprehensive guide—**[The definitive guide to mastering product sense interviews](https://www.lennysnewsletter.com/p/the-definitive-guide-to-mastering)***—*I explained why product sense and analytical thinking interviews matter and I walked through my framework for acing the former. Now I’ll walk through my framework for acing analytical thinking interviews. **In this post, I’ll cover:** - **How to structure your answer to demonstrate the exact signals interviewers are looking for** - **Common pitfalls to avoid and how to navigate challenging scenarios** - **Practical preparation techniques to build your analytical muscle** Whether you’re actively preparing for upcoming interviews or you simply want to understand how top tech companies evaluate PMs, this guide will equip you with all the tools and frameworks you need to succeed. We’ll start by exploring exactly what analytical thinking interviews assess and how they’re structured. By understanding the interviewer’s perspective, you’ll be better equipped to provide the signals they’re looking for. ## Understand the analytical thinking interview Analytical thinking (AT) interviews assess a candidate’s ability to understand a product in the context of its broader company and its market, establish metrics to track success, identify team goals, and evaluate tradeoffs in a structured way. Here are some examples of typical AT questions: 1. ***“How would you measure success for Spotify?”*** 2. ***“You’re a PM at Meta. Set a goal for Instagram Reels.”*** 3. ***“What should be the North Star metric for DoorDash?”*** Although an AT interview normally lasts 45 minutes, you’ll have only about 35 minutes of actual working time after accounting for introductions and wrap-up questions. That might seem like plenty of time, but it flies by, so you’ll want to let your inner “time cop” run the show. Since the interviewer won’t be looking for a “correct” answer—rather, they’re evaluating your thought process—I recommend approaching the interview with this linear flow to maximize your chances of providing strong signals for the following dimensions in a structured way: ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/fcb9ef32-98bd-4c94-a409-978b52a87d15_2912x2494.png) *Note: Analytical thinking interviews can sometimes include other question types, such as debugging/root cause analysis questions and estimation questions, but these are becoming less common. I included some thoughts about tackling debugging and estimation questions at the bottom of this post.* ## Step 1: Assumptions and game plan The opening minutes of an analytical thinking interview can make or break your performance. I’ve watched countless candidates struggle because they jumped straight into goals without aligning with the interviewer on the structure. My advice is to approach the interview as a game with clear rules rather than a casual conversation, and spend the first minute making a handful of assumptions at the start of the interview that narrow the scope to a manageable exercise without prematurely limiting the solution space. Then outline your game plan for the interview, showing the interviewer how you want to spend your time together. When transitioning between key sections of the interview structure (e.g. from product rationale to metrics framework), take a pause to think through your content for that section of the interview. Once you’re ready, check in with the interviewer to walk them through your thinking and confirm they’re following before you proceed to the next section. To set yourself up for success, state a few potential assumptions to align on the scope of the exercise. Here are some flavors of assumptions that can be helpful: ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/74fa4fb6-f59c-4667-889a-20222b403b37_2912x2330.png) Let’s now apply our assumptions framework to the first typical AT question I gave above, which we’ll be using as an example throughout this post: *How would you measure success for Spotify?* (Other potential questions and answers will be available in links at the bottom of each section.) The assumptions below focus on the core consumer-facing music streaming service globally across all platforms, preventing getting lost in niche features or regional specifics. This grants you the freedom to explore meaningful metrics that capture Spotify’s fundamental purpose without being overwhelmed: > *I’ll focus on Spotify’s core music streaming service as the primary product.* > > *I’ll emphasize the consumer-facing experience rather than artist/label tools or enterprise solutions.* > > *I’ll assume we’re focusing on global metrics, not specific to any one region.* > > *I’ll assume we’re talking about all platforms where Spotify is available (mobile, desktop, etc.).* **Additional examples:** - ***Question: You’re a PM at Meta. Set a goal for Instagram Reels**.* - Assumptions[here](https://docs.google.com/spreadsheets/d/1IW89pLp9S-pgieiyVmFLHmf5rfc50zoTbuELyC7ishQ/edit?usp=sharing) acknowledge Reels’s worldwide reach, concentrating on mobile because that’s where the primary user engagement happens. - We also choose to focus on the current state of the product to ensure alignment with the interviewer that we’re not measuring success at launch. - **Question:** ***What should be the North Star metric for DoorDash?*** - Assumptions [here](https://docs.google.com/spreadsheets/d/1658kGtofGK8FOTlWAKnhmlbgtYxhpZQxm5LeFHDzPwE/edit?usp=sharing) mention customers, restaurants, and dashers up front, signaling that you think about the entire marketplace. - Focusing on the U.S. market prevents overwhelming complexity, and including all platforms/devices keeps the exercise comprehensive. After stating assumptions, share your game plan for the interview to make sure you’re on the same page as the interviewer about how you’ll spend your time together: > *“I want to start by reviewing the product’s landscape and reason for existing, then identify key stakeholders and ecosystem health metrics, define a North Star metric with guardrails, and finally set specific team goals. We can discuss tradeoffs along the way at any point. Does that sound like a good plan for our time together?”* Interviewers love hearing this because it tells them you have a plan to generate the signals they need. To put you in the interviewer’s shoes, this is kind of like sitting down to play a game with someone who knows how to play vs. someone who doesn’t. It’s just better. Outlining your game plan up front also gives the interviewer a chance to redirect you if they want to spend the time differently, preventing you from going in the wrong direction and wasting everybody’s time. ## Step 2: Product rationale With the interviewer on board with your assumptions and game plan, the next step is articulating the product rationale. This is your opportunity to establish your understanding of the product context and its strategic importance to the business before diving into metrics. Without this contextual grounding, your metrics will feel arbitrary rather than aligned with the product’s purpose and the business needs. While most good candidates describe what a product does, you’ll set yourself apart by building a compelling strategic foundation touching on three key areas: 1. **Product context:** Describe the product, its maturity level, and business model. Explain the core problem it solves for users and why this problem matters to the market and the company. Articulate how the product creates and captures value, being specific about revenue streams. 2. **Market positioning:** Review the competitive landscape and highlight the company’s unique advantages relative to its peers. Let your awareness of relevant trends in the market shine. 3. **Company and product alignment:** Create a concise, powerful mission statement that captures the product’s core purpose. For products within a larger company, establish a clear throughline from company mission to product mission. Connect the product’s purpose to the parent company’s broader vision and strategic objectives. ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/11beb202-1942-48b3-b159-94ccd7018321_2912x1864.png) Top candidates don’t just describe a product’s features but establish a compelling case for why it matters to users, the business, and the market. This foundation makes all subsequent metrics and recommendations feel naturally aligned with the product’s core purpose. *💡 Tip: Spend about a minute organizing your thoughts before sharing product rationale with the interviewer.* ### Examples of articulating product rationale Let’s now apply our metric framework principles to our question: *How would you measure success for Spotify?* The rationale below articulates the core problem Spotify solved (music piracy and limited access), establishing why the product exists beyond its features. Second, it positions Spotify within its competitive landscape, highlighting specific differentiators like recommendations and platform support. Most importantly, it concludes with a powerful mission statement that will serve as our North Star for all subsequent metrics and decisions. > *Spotify is a music and podcast streaming service offering millions of tracks through a freemium model, generating revenue via premium subscriptions and advertising to free users.* > > *Spotify addressed the critical problem of music piracy by providing legal, affordable access to extensive content when consumers previously faced limited, expensive purchasing options or risky illegal downloads, creating a sustainable model benefiting both listeners and creators.* > > *Currently in late growth/early maturity, Spotify has secured significant market share while expanding globally and diversifying beyond music.* > > *Competing with Apple Music, Amazon Music, and YouTube Music, Spotify distinguishes itself through superior recommendations, social features, broader platform support, and dedicated focus on audio content as its core business rather than an ancillary service to sell hardware or other subscriptions.* > > *Mission statement: “Unlock the potential of human creativity by giving artists the opportunity to live off their art and fans the ability to enjoy and be inspired by it.”* **Additional examples:** - **Question:** ***You’re a PM at Meta. Set a goal for Instagram Reels.*** - The framework [here](https://docs.google.com/spreadsheets/d/1IW89pLp9S-pgieiyVmFLHmf5rfc50zoTbuELyC7ishQ/edit?usp=sharing) demonstrates how to craft a product rationale that connects product-level purpose to company-level strategy. - Note how it establishes the market context (shift to short-form video content), competitive positioning against TikTok and YouTube Shorts, and Meta’s unique advantages. - It also explicitly connects Reels to Meta’s broader mission of building human connection, creating a clear throughline from company vision to specific feature. - **Question:** ***What should be the North Star metric for DoorDash?*** - The framework [here](https://docs.google.com/spreadsheets/d/1658kGtofGK8FOTlWAKnhmlbgtYxhpZQxm5LeFHDzPwE/edit?usp=sharing) demonstrates how to address multiple stakeholders in a marketplace. - Notice how it explicitly identifies the problem solved for each ecosystem player: convenience for customers, expanded reach for restaurants, and flexible earnings for drivers. - It also acknowledges the product’s maturity stage and expansion opportunities beyond its core offering. - The mission statement elegantly captures the multi-sided value proposition in a single sentence. ### How to practice product rationale Get reps in with an exercise that develops both written and verbal muscles: - Write down the rationale for 3 of your favorite products (touching on the key elements above), and grab time with a friend to verbally walk them through these in roughly 2 minutes. - Then ask them to read back to you what they remember about why the product exists, who it’s for, how it makes money, and what makes it different from its competitors. If they understood those, you’re on track! ## Step 3: Metric framework Next, you’ll leverage the product rationale to **define concrete metrics that track ecosystem health**. This section is nuanced and often entails a decent amount of back-and-forth with the interviewer to make sure they follow your thinking. So allocate the largest chunk of time in the interview to your metric framework. While good candidates can identify relevant metrics, what will set you apart is a cohesive story about healthy growth: 1. **Ecosystem value:** Start by listing key players who derive value from the product ecosystem rather than jumping straight to metrics. For each ecosystem player, identify their value proposition for participating (“What’s in it for me?”) and the specific actions they must take to realize this value. Leave out nice-to-have actions. 2. **Metric definition:** Track key actions through metrics that a data scientist could implement, including time frames based on real user behavior. Define a North Star metric (NSM) that reflects value creation across ecosystem players and can grow indefinitely as the product succeeds. Make sure you include a time frame (e.g. “weekly”) with the NSM definition. 3. **Critique NSM:** Call out the drawbacks of your NSM, identifying 1-2 ways that NSM growth could unintentionally damage ecosystem health. Define guardrail metrics that specifically address the key drawbacks. If you can’t describe your metric to a data scientist in a way that they could run a query with, it’s not a useful metric. Always define metrics so specifically that someone could implement them tomorrow, and focus on 3-5 primary metrics per ecosystem player rather than trying to capture everything. Metric selection can also become problematic if you use averages or ratios as North Star metrics. Here’s why this backfires: if your NSM increases while your ecosystem actually shrinks, you’re getting a false positive. I’ve watched candidates confidently present metrics that could look great even as their product dies (not exactly what you want to signal to an interviewer). *💡 Tip: Spend about 2 minutes organizing your thoughts before sharing your metric framework with the interviewer.* ### Examples of metric frameworks Let’s now apply our metric framework principles to our question: ***How would you measure success for Spotify?*** The framework below organizes metrics by players (listeners, creators, advertisers, Spotify), tracking value creation across the ecosystem. I like tracking daily, weekly, and monthly (“DWM” in short) in the ecosystem metrics section before narrowing in on one time frame for the NSM. The NSM, “total streaming hours per week,” measures the total volume of our unifying action that benefits all ecosystem players in a way that matches real engagement patterns. The guardrail metric prevents becoming a passive platform. ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/f4140c1e-4dd7-4148-b71d-e0e3a4103f36_2912x4784.png) **Additional examples:** - **Question:** ***You’re a PM at Meta. Set a goal for Instagram Reels.*** - The framework [here](https://docs.google.com/spreadsheets/d/1IW89pLp9S-pgieiyVmFLHmf5rfc50zoTbuELyC7ishQ/edit?usp=sharing) balances adoption indicators (daily Reels viewers) with engagement depth (watch time). - The North Star metric “total Reels watch time per week” captures the core value: engaging short-form video. - Guardrail metrics prevent optimizing watch time at the expense of content quality or broad user adoption. - **Question:** ***What should be the North Star metric for DoorDash?*** - The framework [here](https://docs.google.com/spreadsheets/d/1658kGtofGK8FOTlWAKnhmlbgtYxhpZQxm5LeFHDzPwE/edit?usp=sharing) measures marketplace success by tracking key indicators for all players: customers placing orders, restaurants fulfilling them, and dashers delivering. - The North Star metric, “total completed deliveries per week,” captures the core transaction creating value across all sides. - Guardrail metrics prevent pitfalls of pursuing delivery volume alone: order satisfaction for quality, retention rates for sustainability, and profit margins to avoid unprofitable growth. ### How to practice metrics frameworks With the products you chose earlier, spend about 10 minutes doing the following exercise with each: - Identify the ecosystem players, map out what value they get and what actions they take, and create specific metrics with time frames that actually make sense. - Then define a North Star metric that can grow indefinitely, and pair it with guardrail metrics that address the biggest ways your NSM could mislead you. - To build your verbal communication muscle, walk a friend or two through your thinking. Ask them if they understand who benefits from the product, how success would be measured, and why you chose your specific North Star metric over alternatives. If they can clearly explain back your measurement logic and see the connection between value and metrics, you’re in good shape! ## Step 4: Goal-setting After you’ve nailed down a solid metrics framework, here comes the critical transition that trips up a lot of candidates: making an “altitude shift” from company/product-level metrics to specific team-level goals. This is where you show the interviewer you can bridge strategy and execution—a skill they’re evaluating. I generally recommend spending about 5 minutes on this section. The transition from metrics (your indicators of success) to goals (specific objectives to pursue) is where many candidates stumble, because they’re fundamentally different types of thinking. Tracking a metric is very different from executing against a specific goal. Think about it this way: if your metrics section focuses on the 50,000-foot view of what constitutes a healthy ecosystem, then your goals section shows you can operate at ground level, identifying specific focus areas for a single product team to tackle in the next 3-6 months. **The strongest candidates demonstrate their expertise by first picking one ecosystem player who can drive the most growth given their established framework and then connecting that choice directly back to the North Star metric and product mission. You want to show a clear throughline from company vision to team objectives by explaining why this particular player represents the highest-leverage opportunity right now.** Next, you’ll work backward from the key actions that contribute to your North Star metric, mapping out the user journey for your chosen ecosystem player. Look for friction points or opportunities along that journey, then score potential goals on both their impact on the North Star metric and your team’s ability to influence them within 3-6 months. End with a clear decision rather than hedging between multiple options. The best goals are ones your team can directly influence and execute, so consider cross-team dependencies and the political capital required to make progress against a goal. *💡 Tip: Spend 2 minutes organizing your thoughts before walking the interviewer through your goals.* ### Examples of goal-setting With our metrics framework established, let’s make that critical altitude shift from product-level metrics to specific team-level goals for the question ***How would you measure success for Spotify?*** The framework below focuses on listeners as the ecosystem player with the highest leverage for driving the NSM of total listening time. The response systematically maps the user journey, evaluates potential goals on impact and ability to influence, and makes a clear recommendation: increasing session continuation rate. What makes this analysis particularly strong is the explicit evaluation of multiple options against consistent criteria before selecting the highest-impact, highest-ability-to-influence goal. ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/20c39f36-1465-4222-9ab3-7a8c77a376aa_2912x3680.png) Additional examples: - **Question:** ***You’re a PM at Meta. Set a goal for Instagram Reels.*** - The framework [here](https://docs.google.com/spreadsheets/d/1IW89pLp9S-pgieiyVmFLHmf5rfc50zoTbuELyC7ishQ/edit?usp=sharing) showcases how to identify the critical leverage point in a content platform. - By focusing on new-creator retention, it recognizes that content supply is the foundation for viewer engagement. - The analysis maps the creator journey, evaluates three potential goals using consistent criteria, and selects improving second Reel creation. - This demonstrates strategic understanding of the “cold start” problem—many creators post once but never return. - **Question:** ***What should be the North Star metric for DoorDash?*** - The framework [here](https://docs.google.com/spreadsheets/d/1658kGtofGK8FOTlWAKnhmlbgtYxhpZQxm5LeFHDzPwE/edit?usp=sharing) focuses on cart-to-checkout conversion, a high-leverage point in the ordering funnel with direct impact on the NSM. - The analysis considers both ability to influence (team control over checkout) and potential impact (direct precursor to delivery completion). - What makes this example strong is its specificity—rather than broadly targeting “customer experience,” it identifies a precise friction point where abandoned carts represent existing intent that can be captured. ### How to practice goal-setting To get solid reps translating ecosystem metrics into team goals, take each of the products you chose earlier and spend 10 minutes on the following exercise: - Identify which ecosystem player to focus on, map out that player’s user journey working backward from the NSM event, and identify 3 opportunities along that journey. - Then score each opportunity as low/medium/high on “ability to influence” (considering your agency, cross-team dependencies, etc.) and “impact on NSM.” Pick the highest-impact, highest-influence opportunity and articulate why it would be a strong candidate for your team’s goal. - Next, develop the verbal skills for this section by walking a friend through your chosen ecosystem player, the specific goal you picked, and your rationale. Ask them if they understand how this goal would move your NSM and why your team could realistically achieve it in 3-6 months. If they can clearly follow your logic, you’re in good shape! ## Step 5: Tradeoff evaluation With team goals established, the final step is to demonstrate how you navigate real-world tradeoffs by applying your strategic framework to tactical choices. Reserve 10 minutes for tradeoff evaluation to show how you’d handle real-world decisions under constraints. **Tradeoff questions assess your decision-making abilities under pressure with incomplete information.** Though they may appear at any point in the interview, they’re typically posed after you’ve established metrics and goals, allowing you to refer to your foundation when making decisions. There are several common tradeoff question types you might encounter: ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/fbce3d17-a769-49c6-b9ba-2e128fc3fef0_2912x2479.png) When evaluating tradeoffs, you can set yourself apart by identifying the common benefit of both options and outlining the pros and cons of each option, pinpointing the crux of the decision. Then clearly state your decision rather than hedging or dodging, connecting your rationale back to the company strategy, product maturity, product mission, and any relevant metrics. I love when candidates also specify what would need to be true for them to change their mind, because it solidifies how they think about the fundamental factors that influence their thinking. And remember to be decisive! *💡 Tip: Spend about a minute organizing your thoughts before walking the interviewer through your decision.* ### Examples of tradeoff evaluation Let’s examine a tradeoff scenario for the question ***How would you measure success for Spotify?*** The analysis below frames the core tension between user breadth and engagement depth and then directly connects the decision to Spotify’s mission. By choosing broader reach with moderate engagement, it prioritizes growth potential while reducing concentration risk. What elevates the analysis is its clear conditions for changing course—stating exactly when deeper engagement might become preferable. ![Image from The definitive guide to mastering analytical thinking interviews](https://substack-post-media.s3.amazonaws.com/public/images/1d0da2c0-9069-4909-aa60-bb5b7cf44326_2912x4053.png) **Additional examples:** - **Question:** ***You’re a PM at Meta. Set a goal for Instagram Reels.*** - The tradeoff framework [here](https://docs.google.com/spreadsheets/d/1IW89pLp9S-pgieiyVmFLHmf5rfc50zoTbuELyC7ishQ/edit?usp=sharing) explores a tension between strengthening social connections (Stories) versus prioritizing entertainment and discovery (Reels). - What makes this analysis particularly strong is how it explicitly connects the decision back to Meta’s mission of building human connection, providing a strategic anchor for the recommendation. - **Question:** ***What should be the North Star metric for DoorDash?*** - The tradeoff framework [here](https://docs.google.com/spreadsheets/d/1658kGtofGK8FOTlWAKnhmlbgtYxhpZQxm5LeFHDzPwE/edit?usp=sharing) explores a tension between the North Star metric (completed deliveries) and a key guardrail metric (average order value). - What makes this analysis particularly strong is its consideration of second-order effects across all ecosystem players: customer ordering behavior, restaurant economics, and operational efficiency. - The decision to roll out the change shows alignment with the North Star while providing clear thresholds that would trigger reconsideration. ### How to practice tradeoff evaluation Generate a tradeoff scenario for each of your practice products by drawing from the five types I mentioned above, and do the following exercise: - Identify the fundamental tension, frame pros and cons of each option, and make a decision that connects back to the mission and metrics framework you outlined. - When evaluating tradeoffs, ask yourself: Do both options align with the mission and NSM equally? Would either option trigger your guardrail metrics? - To get comfortable talking through a tradeoff decision, grab time with a friend to walk them through your reasoning. Ask if they understand your options, why you made your specific decision, and what would change your mind. If they can follow your logic and see how you connected the decision to mission and metrics, you’re in good shape! ## Other analytical questions: debugging and estimation While most analytical thinking interviews follow the five-part structure I’ve outlined, some companies assess analytical skills through alternative question formats. These questions evaluate the same fundamental abilities, just through different problem-solving lenses. Understanding these additional formats will prepare you for any analytical thinking interview. To be clear: it’s extremely unlikely you’ll ever be asked a goal-setting question, tradeoff question, and debugging or estimation question all in a single interview. At most, you’ll get two of these in one session. ### Debugging questions Debugging questions (also known as “root cause” analysis) assess your ability to systematically diagnose problems and identify potential causes. These questions typically present a scenario where some metric has moved and you need to determine what happened. Although debugging questions are becoming less common (particularly at Meta), they still occasionally appear to assess analytical skills. Here’s an example: ***“You’re a PM at DoorDash and noticed orders were down 10% last week. What happened?”*** To approach this, you’ll start by clarifying the observed issue: restate the scenario to ensure understanding, including the magnitude of the metric movement and the time frame in which it’s been observed. Then you’ll want to break down key dimensions like geos/markets, platforms, and segments to identify where exactly the anomalous behavior might have originated and for which users. This will allow you to generate hypotheses for potential causes across technical issues, product changes, user behavior shifts, external factors, data issues, etc. Explicitly state each hypothesis before explaining how you’d test it. The most common mistake I’ve seen here is candidates fixating on a single hypothesis and failing to generate multiple potential explanations for the observed issue. Don’t explore complex theories if simpler causes can explain what’s going on. ### Estimation questions Estimation questions, which historically are popular at Google, assess your ability to “ballpark” with limited information. These questions aren’t about arriving at the precise answer but, rather, demonstrating a logical approach to breaking down complex problems and reaching an answer within a reasonable order of magnitude. Here’s an example: ***“How many shuttle buses does Google need to transport employees from San Francisco to Mountain View each day?”*** To approach this, you’ll want to start by restating the question, ensuring that you understand what you’re being asked to estimate. Next, you’ll want to break down the question by creating a logical formula that isolates all of the necessary components for reaching an answer. You’ll then plug reasonable assumptions into that formula, relying on personal experience and common knowledge as reference points (use round numbers in the right order of magnitude for simplicity). Remember to include all critical components; otherwise your estimates will be fundamentally flawed. The last step is calculating and sanity-checking your response before asking the interviewer if the final estimate seems reasonable. They’ll likely ask some follow-up questions to better track your thinking. ## Now you’re ready to take on the analytical thinking interview The five-step framework we’ve explored—assumptions and game plan, product rationale, metrics framework, goal-setting, and tradeoff evaluation—creates a coherent structure to demonstrate your analytical chops and gives the interviewer the signals they need. As you use this framework, make sure you consistently communicate these key elements: - A compelling product mission that guides all subsequent decisions - Ecosystem-first thinking that tracks value creation for all stakeholders - Decisive recommendations with clear rationale rather than hedging - Clear connection between team goals and North Star metric I’ll leave you with one last thought: The analytical skills you develop by practicing this framework will benefit you beyond interviews! My students consistently tell me that their new interviewing muscle also makes them stronger in their current role. ## Resources to prepare for analytical thinking interviews The most effective preparation combines understanding the framework with extensive practice on real interview questions. No one “wings” these interviews successfully! Deliberate practice makes all the difference. To help you master analytical thinking interviews, I’ve created several resources: - I’ll be hosting a free 30-minute lightning lesson on July 9th where I’ll show you [how to use AI to practice for your analytical thinking interviews](https://maven.com/p/f6c75d/practice-for-analytical-thinking-pm-interviews-with-ai). - You can make a copy of my [Analytical Thinking Interview Template](https://docs.google.com/spreadsheets/d/13rY2E48My2ze8NMzdDRPqVF4T7UFfNz61SZT05cKex8/edit?gid=217460033#gid=217460033) to work through as many practice questions as you’d like. - My top-rated Maven course: [PM Interview Bootcamp with AI Copilot: Product Sense & Analytical Thinking](https://bit.ly/4iQm5xq). (Use code “LENNYSLIST” to get $100 off. Enrollment for the July cohort closes July 13th.) - [How to Ace Analytical Thinking PM Interviews](https://maven.com/p/fb644e/how-to-ace-analytical-thinking-pm-interviews): Free lecture I published on Maven - My [AI Practice Copilot for Product Sense & Analytical Thinking Interviews](https://www.benerez.com/copilot) - [Lewis Lin’s PM Question Bank](https://docs.google.com/spreadsheets/d/1rz10oEeLx-eGnilahKczYPhGfCUzIEKL-xRnjoQ-SX4/edit?gid=1024620532#gid=1024620532): The best PM interview question bank, updated daily *Thanks, Ben! Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [26/46] What people are vibe coding (and actually using) Everyone’s talking about [vibe coding](https://www.lennysnewsletter.com/i/153296003/vibe-coding), and how important it is to get your hands dirty by using the latest AI tools. But what are people building? And are these products useful, like for real? The answer is a resounding yes. I asked on [X](https://x.com/lennysan/status/1938252382447845524), [LinkedIn](https://www.linkedin.com/posts/lennyrachitsky_whats-a-producttool-you-vibe-coded-that-activity-7344093906420056064-PG_J?rcm=ACoAAABGvmoB4S920iEQfSFO_P91nw2wPqfPoic), and my subscriber Slack community: **What’s a product or tool you vibe coded that you actually use regularly in your work or life? And what tool/platform did you use to build it?** The response was overwhelming: Within 24 hours, I got over 1,000 enthusiastic replies, ranging from a buzzer app that automatically answers apartment deliveries, to a hyper-personalized greeting card generator, to a workplace accomplishments tracker, to a daily newsletter to help you learn a new language, to the perfect playlist curator for your next festival road trip. Your stories opened up my mind to what’s possible, and even inspired me to vibe code a few new tools for myself, including this sweet [YouTube thumbnail preview tool](https://project-youtube-podcast-title-and-thumbnail-previewer-348.magicpatterns.app), [a tool to help me craft tweets for my podcast clips](https://tweet-punch-perfect-posts.lovable.app/), and the beginnings of a [tracker of the most mentioned books by podcast guests](https://v0-lenny-s-book-mentions.vercel.app/) (data isn’t real yet). To nudge you forward on your own vibe-coding journey, I’ve pulled together over 50 of my favorite examples from everything you shared. As you’re reading through this list and wondering what to do, try opening up one of the AI tools (or a few at a time) and **simply describe what you want in plain English**, as if you were talking to a remote engineer. Then, iterate by describing what you want to change about what you see, as if you’re speaking with a remote engineer. You’ll be surprised by how far you’ll get. #### **Some high-level takeaways from the stories you shared:** - **Cursor**, **Claude Code**, **Replit**, and **Lovable** were your favorite vibe-coding tools, followed by v0, Bolt, and ChatGPT. Honorable mentions to Gemini, n8n, Zapier Agent, Warp, and Windsurf. - **Almost no examples are alike.** Everyone is solving their own hyper-specific problem (e.g. group drafting app for your multi-sports fantasy league), or exploring a random idea they (or their kids) suggested (e.g. a cats and sushi browser-based video game). Welcome to the era of n-of-1 personalized software. - **You’re creating a lot of Chrome extensions.** This makes sense—we spend most of our time in the browser. - **Even though you’re solving your own problem, many of the products end up being used by tens/hundreds/thousands of other people.** Excellent sign! - **Women are vibe-coding like crazy.** The male-female ratio in the responses is more balanced than in most tech conversations. Another excellent sign! Thank you to all 1,000+ of you who shared your stories 🙏 ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/76fa567a-807e-450c-b3ed-3cbd5515fded_2912x1940.png) ## **Health, wellness, and style** ### Carb counter, by [Morgan Brown](https://www.linkedin.com/in/morganb/) using Replit > “I built CarbScanto help manage my son’s diabetes and blood glucose levels with faster carb counting. I used Replit. Has become a daily go-to.” [[Check it out](https://carbscan.ai)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/d6282e81-3b18-46ca-918c-27036e3e3264_948x1242.png) ### Lash tracker, by [Jackie Bavaro](https://x.com/jackiebo/) using Replit > “Every time I put on new eyelashes, I take a picture and record the styles and method used.” [[Check it out](https://lash-map-tracker.replit.app/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/65d03d29-dccb-40bb-a1d2-20f0d6b2a1d7_1170x1554.jpeg) ### How many layers should I wear today?, by [Vijith Quadros](https://x.com/vijithq/) using Lovable > “I vibe coded this with Lovable and use it myself every single day. It’s grown to 85K users in total in nine months. Weather apps had too much clutter for simple daily clothing decisions.” [[Check it out](https://howmanylayersidag.se/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/0b593729-f3df-491d-a1d3-ad1f26f6c403_1218x1206.png) ### A “stupidly specific” workout app, by [Faraz Khan](https://x.com/farazKhan404) using v0 and Claude Code > “I built a stupidly specific workout app for myself. It’s pretty gross and not for anyone else, but I love it. I was chatting with Claude about switching up my program, and I was about to ask it to format it into a table when I realized, ‘Wait, I’m in Claude and it can make Artifacts.’ Finally I ended up dumping the code into v0 to get it deployed.” ![Image](https://substack-post-media.s3.amazonaws.com/public/images/59c68d47-3073-42c2-8dac-58c70d8e3649_1206x2622.jpeg) ### Personalized meal plans based on what’s in your fridge, by [Nick Markman](https://www.linkedin.com/in/nickmarkman/) using Lovable and Cursor > “I built a web app using Lovable, Supabase, and Cursor where you can upload photos of your fridge/pantry/receipts (or use voice) to detect ingredients you have at home, save to ‘inventory,’ and generate recipes using AI. > > It factors in the user’s set dietary preferences and input of how strict/flexible recipes should be with using what you have in inventory vs. things you need to buy. It also auto-generates a shopping list for ingredients you need to buy for selected recipes.” [[Check it out](https://mealmuse.ai/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/379572df-f331-438d-a5c4-65f903468223_1232x817.png) ### Find out what’s blocking your flow, by [Su](https://x.com/su_dreams) using Bolt > “Flowbound is an app that gives me games/exercises to do when I find myself procrastinating on something (which is often).” [[Check it out](https://www.flowbound.app/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/eb71a55a-c353-43db-a7fe-10521f5062a9_2744x1572.png) ### Pickleball games tracker, by [Jacob Jolibois](https://www.linkedin.com/in/jolibois/) using Replit > “I (and users from all across the U.S.) use Paddles.ai to track and analyze my pickleball matches. Vibe coded and deployed on Replit.” [[Check it out](https://www.paddles.ai/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/78e1289e-bbf7-410b-82a2-f31d9fa68ec7_2334x1514.png) ### Nicotine pouch tracker, by [Thatcher](https://x.com/thatchhh)using Cursor > “I have no coding background and vibe coded this as my first app with Cursor/Xcode + SwiftUI. It helps you taper off high nicotine usage. It’s sticky until you’re ready to quit. Took a month to launch. Recently redesigned from scratch in a couple days.” [[Check it out](https://apps.apple.com/us/app/nicotine-pouch-tracker-pouched/id6740014859?itscg=30200&itsct=apps_box_link&mttnsubad=6740014859)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/f991d2c3-d9d0-437c-9064-633590516c80_2018x890.png) ## **Parenting and family** ### Create stories by dragging emoji into a pot, by [Akshan Ish](https://x.com/akshanish) using Replit > “I built Storypot entirely on Replit for my kid. Other kids and parents also seem to like it. We and 60-odd families use it frequently.” [[Check it out](https://app.thestorypot.com)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/ed3382a3-6665-4cb6-9fb1-314ecfafedc6_1438x1542.png) ### Teach your kids budgeting and savings, by [Sanjeev Nair](https://x.com/sanjeevn72) using Claude Code > “I built a simple app using Claude Code to help my kids figure out expense planning and saving when in college. Great value add considering it’s a tough topic for parents to make teenagers focus/lean toward.” [[Check it out](https://claude.ai/public/artifacts/f4a86536-5062-4570-a8ef-5a93e6bf671f)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/05d41378-fb22-40b1-a571-18c82cc1d30a_2452x1496.png) ### Turn family photos into videos, by [Oren Saban](https://www.linkedin.com/in/oren-saban/) using Lovable and Bolt > “I built Timeless Memories with Lovable, which is pretty insane to think such a complex app can be built with that. And all my friends have used it already. > > I also built a logo animator with Bolt that uses the GSAP library to animate our logo in cool ways, for different onboarding experiences.” [[Check it out](https://timelessmemories.me/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/d7d22a0c-6d6f-4d7c-bb66-29021065c43e_2724x1648.png) ### Baby care tracking, by [Javier Evelyn](https://x.com/hah_vee_AIR) using Lovable > “I’m a new dad, and while waiting for the little one to arrive, I started tinkering around with Lovable. I built My Baby Logger over two weekends. Tracks feedings, sleep, diapers, and meds. It’s been helpful for us.” [[Check it out](https://mybabylogger.com)] ![Image](https://substack-post-media.s3.amazonaws.com/public/images/639f4c9b-bed2-45a6-b9b4-e85212f46163_947x2048.jpeg) ### A chore app for kids, by [Ben Ogren](https://x.com/benogren) using v0 and Claude Code > “I built a chores app for my kids. It’s my first iOS app! I started with v0 but quickly moved off it and started using Claude to help me through areas where I got stuck.” [[Check it out](https://www.chores-ai.com/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/99a5a0c6-5f7e-47fd-a724-381c01d30aee_2532x1384.png) ### A bedtime storytelling app, by [Harshitha P.](https://www.linkedin.com/in/harshitha-p-991776149/) using Bolt > “I built this for parents, and now I use it every night. It’s a bedtime storytelling ritual that turns your daily emotions into personalized stories for your child. You reflect, drop a ‘pebble’ or plant a ‘story seed,’ and the app creates a calming, age-appropriate story shaped by what’s on your heart. I built it solo in Bolt + Supabase**,** and what started as a side project became my nightly parenting ritual.” [[Check it out](https://stories-of-life.vercel.app/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/1a234757-3dcf-42ca-ba49-05ef2da2109c_2394x1564.png) ### Personalized greeting card generator, by [Bob Sheth](https://x.com/bobsheth) using Cursor, Windsurf, and Claude Code > “I vibe coded an AI greeting card generator, and send everyone I know these hyper-personalised cards.” [[Check it out](https://jenicards.com/cards)] ![Card Main](https://substack-post-media.s3.amazonaws.com/public/images/4e8b6997-956b-483b-a66c-0b5f2e54200c_1024x1536.png) ## **Work productivity** ### Meeting prep automation, by [Marissa Goldberg](https://x.com/mar15sa) using Zapier Agent > “It checks my calendar each morning, identifies who I’m meeting with, and compiles everything I need to show up prepared.” [[Here’s how to build it](https://ideakitchen.substack.com/p/ai-recipe-create-your-own-meeting)] ![Image](https://substack-post-media.s3.amazonaws.com/public/images/684d2aaf-9d45-4a50-817e-e5276673b17c_1200x734.jpeg) ### A standup order randomizer, by [Rob Balderstone](https://www.linkedin.com/in/robbalderstone/) using Lovable > “I use this daily at work.” [[Check it out](https://standup-buddy.lovable.app/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/decb63ea-31f3-45fa-855a-9a61d0ddbfd8_1318x1032.png) ### Chrome extension to share your availability, by [Olena Vozna](https://www.linkedin.com/in/alionavozna/) using Replit > “A Chrome extension that adds my availability to emails in natural, human-style language. It also analyzes suggested times from others, picks the best one, and books the meeting. İt learns my meeting habits to recommend the most suitable slots. > > Then, a Slack app for team scheduling that uses the same core algorithm. It learns everyone’s preferences and finds the best time when I type something like *Schedule a 45-minute call with Mary and Jane this week.* Or *Schedule a focus time for me today. I need 2 hours.* Both were built with Replit.” ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/e8c97ef4-9643-4aae-8fdf-406e933e82c5_1176x360.png) ### A time tracking app, by [Asif](https://x.com/asifkabeer) using Warp.dev > “I built a time tracking app with Warp.dev. All done in a day.” [[Check it out](https://time.wisdemic.com/)] ![Image](https://substack-post-media.s3.amazonaws.com/public/images/195271f4-ec45-4682-b97e-fdce3140d986_889x646.png) ### Inbox focus tool, by [Ari Klein](https://www.linkedin.com/in/kleinari/) using Gemini > “I built a Gmail add-on that doesn’t let my inbox hijack my day or steal my attention. I use it every day. It holds my inbound email and delivers it to my inbox on a schedule I set (e.g. 10 a.m. and 3 p.m.) but still lets important emails through right away if they match certain VIP keywords (e.g. verification code), VIP domains (e.g. my-kid’s-school.org), or VIP email addresses (e.g. wife@gmail.com). Built with Gemini.” ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/68d7b391-b68e-409d-bee8-371b17eae1a0_1280x1059.jpeg) ### Accomplishments tracker to showcase your wins at work, by [Kevin Kirkpatrick](https://www.linkedin.com/in/kevinkandco/) using Lovable > “Built this for myself because I hate updating my resume, and I think other people hate it too. Built it with Lovable and I use it weekly to track what I’ve accomplished.” [[Check it out](https://www.goaccomplishit.com/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/274b84bb-9721-47ea-90f9-b3e3cceb2c4b_2032x1584.png) ## **Personal productivity** ### Automatically answers your apartment buzzer calls, by [Darren Rulofs](https://www.linkedin.com/in/darrenrulofs/) using Bolt and Cursor > “I built Buzzerbee using Bolt and Cursor. It automatically answers your apartment buzzer calls based on the access rules and schedules you set in the app. I’ve been using it daily, and others have been joining as well. Turns out people don’t like missing deliveries!” [[Check it out](https://buzzerbee.app/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/ced5b879-9a10-4bbe-b164-c52157ad6f05_2368x1346.png) ### Bill splitter, by [Shashikiran Devadiga](https://www.linkedin.com/in/shashikirandevadiga/) using Lovable > “My one-screen, tip-first bill splitter. Whenever friends float dinner plans, I plug in the menu subtotal, head count, pick the state, and nudge the tip slider. In two seconds, I see each person’s final after-tax share—something Splitwise can’t show until the check arrives. It’s become my quick ‘Can I afford this outing?’ filter and saves the awkward math at the table. And I built it using Lovable dev + Unicorn Studio.” [[Check it out](https://my-bill-split-project.lovable.app)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/555d915a-a3ae-475f-9ef8-3762b811ca94_1592x1228.png) ### A personalized restaurant recommendation app for friends, by [Lani Young](https://www.linkedin.com/in/lani-young/) using Replit > “I used Replit to bring to life something I’ve always wanted to build: a restaurant recommendation site for my friends to give them a curated restaurant recommendation based on what they are looking for in a specific meal/experience. It’s a combo of a database, RAG, and natural language chat with an LLM. Right now, it’s only Denver-based, but aiming to fill that gap between Yelp (too much junk/can’t trust the recommendation) and trying to find a new place.” [[Check it out](https://www.curated.now)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/1dd2495d-41dd-4a88-a6ce-8f3964dc12e9_2440x1496.png) ## **Personal development** ### Conversation intelligence, by [Peter Nixey](https://x.com/peternixey/status/1938537295763636456) using Claude Code > “This lets you paste in the transcript from a call and gives you a breakdown of what you did well and what you can improve on. I just used Claude Code. It’s surprisingly powerful.” [[Check it out](https://claude.ai/public/artifacts/cf207ca7-a35a-42e6-8843-3d5edcccf9f8)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/f2375201-609e-42dd-ba31-7118e00cf57a_2302x1378.png) ### Daily newsletter to help you learn a language, by [Dustin Coates](https://www.linkedin.com/in/dustincoates/) using v0 > “I created a daily newsletter (with an audience of 1) that emails me every day at lunchtime: > > - A story in the language I’m learning, geared toward my beginner level. The story is about something that happened on that day in history, generated by Gemini. > - A list of vocabulary from that day’s story > - A short grammar lesson from that day’s story > > I tried Replit, Bolt, Gemini, and v0, and v0 came out with the best newsletter design. Then I had v0 also create the script for running the email (via Resend) and a script for packaging the code to upload to Lambda.” ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/3554a0a7-df44-4412-9a31-6424f290f877_3000x2078.png) ### Talk to an AI about your day, by [Alex Mathew](https://x.com/alxmthew) using Replit > “I built an app that lets you talk to an AI about your day-to-day problems. Used Replit.” [[Check it out](https://talk.berryplush.com/)] ![Image](https://substack-post-media.s3.amazonaws.com/public/images/0ea5d188-f881-4dd8-b89a-5f91fa460ff0_1200x767.jpeg) ## **Music** ### Curate the perfect playlist for your next festival road trip, by [Steven Newstead](https://www.linkedin.com/in/steven-newstead-33547414/) using v0 and Cursor > “I made a website to help with artist discovery for UK festivals. I use it to quickly generate playlists. I wouldn’t say it’s widely used beyond a handful of people. But it’s quite decent. I used Cursor and backed by Supabase for Spotify authentication. Currently all running on free tier infrastructure.” [[Check it out](https://www.fyredrill.dev/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/0d2effae-ed91-4377-9d2b-cf5ed6a03827_1980x1670.png) ### A music production tool to inspire new ideas, by [Scott Korchinski](https://www.linkedin.com/in/scottkorchinski/) using v0 and Cursor > “I made this music production tool the first week we got the Lenny AI bundle, with 90% v0 and 10% Cursor. I think it took three days from idea to shipping. I use it weekly just to get inspired for new musical ideas. Hotjar shows a handful of users per week, too!” [[Check it out](https://splintr.dreamgazeraudio.com/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/5d8659fb-2d0f-4a65-b8a2-a494423ee67d_2270x1676.png) ## **Other/Fun** ### An RPG game, by [Jean Kaddour](https://www.jeankaddour.com/) using Cursor > “I love RPG games and was looking for an online, AI-based RPG game so I can play without friends around. I then vibe coded Rehearsal—a browser game where you have 10 minutes to convince an AI character. For example, to convince a hostage taker to release hostages, or convince a broke flatmate to pay his share of rent. I used Cursor (Sonnet 3.7) + Next.js. I built a first version in a day. Since then, both my g.f. and I are addicted to it and play daily.” [[Check it out](https://rehearsal.so/)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/7fa3101a-b327-4d34-9ea0-39d84284d6e1_2492x1684.png) ### Immigration document manager, by [Shree](https://x.com/sreeenidhi) using v0 > “Immigrants from India and China have a 10- to 20-year journey where you have to file something every couple of years. VisaMonkey keeps all your immigration-related documents in one place so you can find the information you are looking for quickly instead of searching through mountains of documents. It’s also useful to fill out forms.” [[Check it out](https://visamonkey.com)] ![Image from What people are vibe coding (and actually using)](https://substack-post-media.s3.amazonaws.com/public/images/95a40999-7c78-4d19-8302-26b5964154f9_2602x1690.png) ### **Have you vibe coded something that you use regularly? Share a link (and what you used to build it) in the comments to inspire others on their journey. 👇** [Leave a comment](https://www.lennysnewsletter.com/p/what-people-are-vibe-coding-and-actually/comments) *Have a fulfilling and productive week 🙏* ### Bonus: Even more inspiration So many people shared so many great examples, I couldn’t help but include more of them. Skim this whenever you’re looking for inspiration: 1. “My 11-year-old just started her summer vacation and was getting bored, so I made her a Replit account and asked her to build a game. A few hours later, voila! She combined her two favorite things—cats and sushi—to make this game 😀. [Enjoy the game](https://sushi-cat-chef-c3913136.replit.app/).” —[Amit Murumkar](https://www.linkedin.com/in/amitmurumkar/) 2. “I created a tool that alerts me when one of around 360 healthcare payers updates their policies.” —[Steven P. Walsh](https://x.com/StevenPWalsh) 3. “I’m a new mom, and I coded an app that helps me [track how much milk](https://rama-milk-monitor.lovable.app) I’ve fed my son (vs. target quantity for his weight). I had been doing math in my head, but this app makes it so easy, and I made it in five minutes on Lovable using just two prompts!” —[Shilpa Cromer](https://www.linkedin.com/in/shilpa-rajgopal/) 4. “A Chrome extension that lets me right-click to open an article behind a paywall.” —[Vincent Turner](https://www.linkedin.com/in/vhturner/) 5. “A Bible study app that uses AI to generate daily bible study guides, declaration, God-said quotes across different themes. I use it daily. Built with Lovable and connected to OpenAI.” —[Princess (Ebiboye Adesuwa) Edo-Osagie](https://www.linkedin.com/in/princess-edo-osagie/) 6. “I built [Trove Dad](https://www.trove.dad/) in Cursor while on paternity leave with my newborn. I’m helping new dads like myself preserve more core memories with their kids through a prompted journal of fatherhood that converts into a baby book.” —[John Stone](https://x.com/JohnStoneBlog) 7. “I created a [voice AI therapy journal](https://reston-nayeemrahman13.replit.app/signin) for my sessions with my therapist using Replit Agent + the OpenAI Realtime API.” —[Nayeem Rahman](https://www.linkedin.com/in/nayeemrahman13/) 8. “In the past few months, I’ve created some custom Figma plugins to help me with design system work: exporting Figma variables to tokens to work with our frontend component repo, and swapping component styling from one design system to another to migrate legacy Figma files to a newer design system. I’m a designer, and I used Cursor + ChatGPT. The ability to quickly spin up Figma plugins for niche uses is a game changer!” —[Kayt (Campbell) Wilson](https://www.linkedin.com/in/kayt-wilson-4973b350/) 9. “Not a product per se, but I vibe coded our backend systems across Women Defining AI to help us centralize member management—auto sync to Mailchimp and Slack, and help us auto-approve registrations to Luma events using the member list.” —[Helen Lee Kupp](https://www.linkedin.com/in/helenleekupp/) 10. “I built a tool for our agency that reads AI summaries of meetings and creates tasks and puts them on calendars. We love it. Used Lovable/Supabase/Vercel.” —[Alan Edgett](https://x.com/ACEdge) 11. “Made an app for myself called Frens Circle super-quickly with Lovable. Now I find myself using it regularly. Basically let me save my best friends’ birthdays, their favourite artists, and it sends me a reminder if I haven’t messaged them in a week. —[Arun Philips](https://x.com/arunphilips) 12. “I created [this game](https://taupe-cassata-525706.netlify.app/) for my 7-year-old, and he’s been enjoying it. There are plenty of games and apps out there for learning English, but I wanted to build something that would be customized for him. Built with Bolt.” —[Daniel Orsher](https://www.linkedin.com/in/daniel-orsher-60008a4/) 13. “I made a handy tool for conferences and networking events. I built initially with v0 and dabbled with Claude Code and Cursor as well.” —[Kelly Vaughn](https://x.com/kvlly) 14. “Helping my father-in-law write his book. Made a scene inventory tool that’s been incredibly useful. Claude Code CLI.” —[Jason Curry](https://x.com/jasonacurry) 15. “I built a workout motivation app in under 15 minutes using Lovable. It pings me every two days to do my strength training, drops mobile notifications every couple of hours till I get it done, integrated YouTube workout video, barbell weight progress log, workout tracker. Completely personalised for me and my current fitness goals 👌” —[Emma Gordon](https://www.linkedin.com/in/emma-gordon-product/) 16. “Built a voice-based workout tracker to log my lifts so I don’t have to type anymore. A few weeks ago, opened it up for others to try, and we’re seeing some small early traction.” —[Omar Ganai](https://x.com/omarganai) 17. “Something to generate Anki flashcards for my son learning German. From PDF file of words to learn to a series of flashcards with pronunciation of each word.” —[Jeff Foster](https://www.linkedin.com/in/fffej/) 18. “I built a time audit tool using Replit to track how and where I spend my time. I also use this to compare if the actual activity list is as per the planned one. Another tool built with Replit to track my energy levels during the day to optimize for when I should schedule certain types of activities for best outcomes.” —[Venkatesh Varadachari](https://www.linkedin.com/in/venkyvchari/) 19. “Dashboard a monitor at work that shows 5-star App Store ratings (ingests webhooks from Appbot), built on Replit. Also have a script that runs in our Slack channels daily and notifies folks who is on-call/when handoffs happen—Claude + Google App Script.” —[Ian Mabie](https://x.com/callmemabie) 20. “Calories tracker, with voice mode into Kcal estimate and suggestion for the rest of the day.” —[Gpt4thier](https://x.com/gpt4thier) 21. “I asked ChatGPT to code me a very simple PDF viewer with a zoom function. That’s my main PDF tool now.” —[Gary](https://x.com/Eldenofthering) 22. “[A fully local video-to-GIF converter](https://vivekaithal.co/llm-tools/tool/vid2gif.html). I was tired of sketchy, buggy websites. Built by Claude Code.” —[Vivek Aithal](https://x.com/nuwandavek) 23. “Built [a restaurant picker](https://www.decidedforme.com/) to combat those never-ending ‘What do you wanna eat? Anything, you decide. No, you decide. . .’ situations.” —[Adi Komari](https://www.linkedin.com/in/adi-komari/) 24. “I coded my [Mac App ‘Battery Lens’](https://apps.apple.com/us/app/battery-lens/id6466673198) in Swift with Code using ChatGPT and a bit of Claude. I wanted a hands-free indicator of how much battery I had left on my laptop, so on a configurable basis, it will overlay that on the screen without your having to do anything.” —[Thomas J. Quinlan](https://x.com/thomas_quinlan) 25. “Built for fun to [easily track different plate/weight combos for workouts](https://barbell.pages.dev). Using Cursor + Gemini Pro.” —[Ado](https://x.com/adocomplete) 26. “I am making a simple mobile page to act as a handover between me and the nanny for my son. The goal is to pass on information like what medicines to give and when, what food to give, and photos of any moments worth capturing. Not for commercial use. Just for daily use for myself. Used Lovable and Cursor.” —[Vipul Agarwal](https://x.com/vipulawl) 27. “I built a group drafting app for a multi-sports fantasy league that my friend has been running offline for 15 years. We did the MLB draft, fully automated across the pool of GMs (snake draft, not auction). We’ll add the other pro sports leagues soon. NFL drafting is coming up. Using Lovable + Supabase.” 28. “I built a sales leaderboard. Tracks top performers, core metrics, and any offer communication. Used Lovable.” —[Siddhant Jain](https://www.linkedin.com/in/siddhant-jain-881942147/) 29. “I have a friend who homeschools, and they vibe code learning games for their kid.” —[Jeffrey Price](https://www.linkedin.com/in/pricejeffreym/) 30. “I built [a personal date tracker app](https://date-dot-tracker.lovable.app/welcome) for different types of recurring commitments—like prepaid swim classes (need to track remaining sessions), flexible hair appointments, house cleaning schedules, etc. Used Lovable to build it and honestly use it multiple times a week. I’m still iterating based on my own usage patterns.” —[Archana C.](https://www.linkedin.com/in/archanac42/) **Have you vibe coded something that you use regularly? Share a link (and what you used to build it) in the comments to inspire others on their journey. 👇** [Leave a comment](https://www.lennysnewsletter.com/p/what-people-are-vibe-coding-and-actually/comments) **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [27/46] Essential reading for product builders—part 1 *👋 Welcome to a **🔒 paid edition 🔒** of my weekly newsletter. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lennybot](https://www.lennybot.com/) | [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny) | [Swag](https://lennyswag.com/)*** *Annual subscribers now get a year free of [Bolt, Perplexity Pro, Notion, Superhuman, Linear, Granola, and more](https://www.lennysnewsletter.com/p/announcing-the-greatest-product-bundle). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** For something a little different, I’m going to share seven essays that have had the most impact on my product career—that you likely haven’t read. There’s so much content flying at us these days, it’s hard to separate the “this sounds so smart!” from the “this is genuinely correct, helpful, and timeless.” The essays below are ones I find myself quoting from, sharing with people, and coming back to most often, even though most are decades old. “The thing I’ve tried to do the last few years is ‘barbell’ my inputs. I basically read things that are either up to this minute or things that are timeless. I’m trying to not read anything that’s from yesterday through to like 10 years ago.” —Marc Andreessen Note that this isn’t an exhaustive list. And I’m not including books—yet. This post is the beginning of an essential and timeless reading library meant specifically for product leaders. **I’d love your help building out this list. What’s missing? Let me know in the comments. Bonus points for sharing the impact it had on your career/life, and more bonus points for sharing stuff people may not have heard of. 👇** [Leave a comment](https://www.lennysnewsletter.com/p/essential-reading-for-product-builderspart/comments) *Bonus: You can also listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from Essential reading for product builders—part 1](https://substack-post-media.s3.amazonaws.com/public/images/c230bf55-47d9-4798-a01d-1d4fbb48624d_2912x1940.png) ### 1. [Who’s Got the Monkey](https://www.med.unc.edu/uncaims/wp-content/uploads/sites/764/2014/03/Oncken-_-Wass-Who_s-Got-the-Monkey.pdf)?, by William Oncken, Jr., and Donald L. Wass > “Let us imagine that a manager is walking down the hall and that he notices one of his subordinates, Jones, coming his way. When the two meet, Jones greets the manager with, ‘Good morning. By the way, we’ve got a problem. You see. . . .’ As Jones continues, the manager recognizes in this problem the two characteristics common to all the problems his subordinates gratuitously bring to his attention. Namely, the manager knows (a) enough to get involved, but (b) not enough to make the on-the-spot decision expected of him. Eventually, the manager says, ‘So glad you brought this up. I’m in a rush right now. Meanwhile, let me think about it, and I’ll let you know.’ Then he and Jones part company. > > Let us analyze what just happened. Before the two of them met, on whose back was the ‘monkey’? The subordinate’s. After they parted, on whose back was it? The manager’s. Subordinate-imposed time begins the moment a monkey successfully leaps from the back of a subordinate to the back of his or her superior and does not end until the monkey is returned to its proper owner for care and feeding. > > In accepting the monkey, the manager has voluntarily assumed a position subordinate to his subordinate. That is, he has allowed Jones to make him her subordinate by doing two things a subordinate is generally expected to do for a boss—the manager has accepted a responsibility from his subordinate, and the manager has promised her a progress report. > > The subordinate, to make sure the manager does not miss this point, will later stick her head in the manager’s office and cheerily query, ‘How’s it coming?’ (This is called supervision.) Or let us imagine in concluding a conference with Johnson, another subordinate, the manager’s parting words are, ‘Fine. Send me a memo on that.’ > > Let us analyze this one. The monkey is now on the subordinate’s back because the next move is his, but it is poised for a leap. Watch that monkey. Johnson dutifully writes the requested memo and drops it in his out-basket. Shortly thereafter, the manager plucks it from his in-basket and reads it. Whose move is it now? The manager’s. If he does not make that move soon, he will get a follow-up memo from the subordinate. (This is another form of supervision.) The longer the manager delays, the more frustrated the subordinate will become (he’ll be spinning his wheels) and the more guilty the manager will feel (his backlog of subordinate-imposed time will be mounting).” ### 2. [On agency](https://www.henrikkarlsson.xyz/p/agency?publication_id=313411&post_id=167633827&isFreemail=true&r=13nam&triedRedirect=true), by Henrik Karlsson > “Last year, when I talked about learning ‘how to handle being sentenced to freedom,’ a phrase I borrowed from Sartre, I meant roughly what people these days call ‘cultivating high agency.’ But I need to define my words, since some ways the phrase high agency is used feel foreign to me, and depressing. > > Agency, as I see it, is an amalgamation of two skills, or mental dispositions: autonomy and efficacy. > > 1. Agency requires the capacity to formulate autonomous goals in life—the capacity to dig inside and figure out what wants to happen through you, no matter how strange or wrong it seems to others. In other words, it requires *autonomy* (which was what I was getting at when I said ‘authentically, and responsibly’). > 2. Agency also requires the ability and willingness to pursue those goals. It requires the ‘will to know,’ the drive to see reality as it is, so you can manipulate it deftly and solve the problems you want to solve, instead of fooling yourself that certain problems are ‘unsolvable.’ In other words, *efficacy* (‘handle it effectively’). > > Or phrased negatively, the *opposite* of agency can mean one of two things. Either (1) doing what you are ‘supposed to do,’ playing social games that do not align with what, on reflection, seems valuable to you and/or (2) being passive or ineffective in the face of problems (assuming your problems can’t be solved, that someone else should solve them, or working on things that do not in a meaningful way address the problem). > > Agency is often framed as a hard-edged, type-A, aggressive approach. But over the last year, as I’ve been thinking about writing this essay, I’ve talked to a lot of highly agentic people, and I’ve read biographies about and interviews with people whose agency I admire and . . . hard-edged does not fit what I’ve seen. Often, agency is almost gentle—an attunement to the world and the self, a feeling out the details of reality, and a finding of the path of least resistance. There is sometimes considerable force involved, hard work, but it is like the force of a river being pulled toward the sea.” ### 3. [On building good taste](https://www.youtube.com/watch?v=X2wLP0izeJE), by Ira Glass *(Technically this isn’t an essay, but who cares.)* > “All of us who do creative work, we get into it because we have good taste. But it’s like there is this gap. For the first couple years that you’re making stuff, what you’re making isn’t so good. It’s not that great. It’s trying to be good, it has ambition to be good, but it’s not that good. > > But your taste, the thing that got you into the game, is still killer. And your taste is good enough that you can tell that what you’re making is kind of a disappointment to you. A lot of people never get past that phase. They quit. > > Everybody I know who does interesting, creative work went through years where they had really good taste and they could tell that what they were making wasn’t as good as they wanted it to be. They knew it fell short. Everybody goes through that. > > And if you are just starting out or if you are still in this phase, you gotta know it’s normal and the most important thing you can do is do a lot of work. Do a huge volume of work. Put yourself on a deadline so that every week or every month you know you’re going to finish one story. It is only by going through a volume of work that you’re going to catch up and close that gap. And the work you’re making will be as good as your ambitions. > > I took longer to figure out how to do this than anyone I’ve ever met. It takes a while. It’s gonna take you a while. It’s normal to take a while. You just have to fight your way through that.” ### 4. [Strategies for learning](https://andymasley.substack.com/p/strategies-for-learning), by Andy Masley > “Pretending to learn feels good. Actual learning often feels bad. A lot of people are just pretending to learn (memorizing passwords) when they think they’re learning. [. . .] > > A lot of people read whole books just for the feeling of being the type of person who reads them, and retain no actual information about the world. > > Avoiding playing pretend during your valuable learning time is important and psychologically difficult. A lot of people’s learning is limited by their insecurities and need for their social status to be reinforced. It often feels amazing to pretend to learn things, and unpleasant to actually learn. People are drawn to activities that make them feel high status, and repelled from activities that make them feel low status.” ### 5. [If your product is](https://paulbuchheit.blogspot.com/2010/02/if-your-product-is-great-it-doesnt-need.html) *[great](https://paulbuchheit.blogspot.com/2010/02/if-your-product-is-great-it-doesnt-need.html)*[, it doesn’t need to be](https://paulbuchheit.blogspot.com/2010/02/if-your-product-is-great-it-doesnt-need.html) *[good](https://paulbuchheit.blogspot.com/2010/02/if-your-product-is-great-it-doesnt-need.html)*, by Paul Buchheit > “I believe this ‘more features = better’ mindset is at the root of the misjudgment, and is also the reason why so many otherwise smart people are bad at product design (e.g. most open source projects). If a MacBook with OSX and no keyboard were really the right product, then Microsoft would have already succeeded with their tablet computer years ago. Copying the mistakes of a failed product isn’t a great formula for success. > > What’s the right approach to new products? Pick three key attributes or features, get those things very, very right, and then forget about everything else. Those three attributes define the fundamental essence and value of the product—the rest is noise. For example, the original iPod was: (1) small enough to fit in your pocket, (2) had enough storage to hold many hours of music and (3) easy to sync with your Mac (most hardware companies can’t make software, so I bet the others got this wrong). That’s it—no wireless, no ability to edit playlists on the device, no support for Ogg—nothing but the essentials, well executed. > > We took a similar approach when launching Gmail. It was fast, stored all of your email (back when 4MB quotas were the norm), and had an innovative interface based on conversations and search. The secondary and tertiary features were minimal or absent. There was no ‘rich text’ composer. The original address book was implemented in two days and did almost nothing (the engineer doing the work originally wanted to spend five days on it, but I talked him down to two since I never use that feature anyway). Of course those other features can be added or improved later on (and Gmail has certainly improved a lot since launch), but if the basic product isn’t compelling, adding more features won’t save it. > > By focusing on only a few core features in the first version, you are forced to find the true essence and value of the product. If your product needs ‘everything’ in order to be good, then it’s probably not very innovative (though it might be a nice upgrade to an existing product). Put another way, if your product is great, it doesn’t need to be good.” ### 6. [The only thing that matters](https://fictivekin.github.io/pmarchive-jekyll/guide_to_startups_part4), by Marc Andreessen > “If you ask entrepreneurs or VCs which of *team*, *product*, or *market* is most important, many will say *team*. This is the obvious answer, in part because in the beginning of a startup, you know a lot more about the team than you do the product, which hasn’t been built yet, or the market, which hasn’t been explored yet. > > Plus, we’ve all been raised on slogans like ‘people are our most important asset’—at least in the US, pro-people sentiments permeate our culture, ranging from high school self-esteem programs to the Declaration of Independence’s inalienable rights to life, liberty, and the pursuit of happiness—so the answer that team is the most important *feels* right. > > And who wants to take the position that people don’t matter? > > On the other hand, if you ask engineers, many will say *product*. This is a product business, startups invent products, customers buy and use the products. Apple and Google are the best companies in the industry today because they build the best products. Without the product there is no company. Just try having a great team and no product, or a great market and no product. What’s wrong with you? Now let me get back to work on the product. > > Personally, I’ll take the third position—I’ll assert that *market* is the most important factor in a startup’s success or failure. > > Why? > > In a great market—a market with lots of real potential customers—the market *pulls* product out of the startup. > > The market needs to be fulfilled and the market *will* be fulfilled, by the first viable product that comes along. > > The product doesn’t need to be great; it just has to basically work. And, the market doesn’t care how good the team is, as long as the team can produce that viable product.” ### 7. [The Cook and the Chef: Musk’s Secret Sauce](https://waitbutwhy.com/2015/11/the-cook-and-the-chef-musks-secret-sauce.html), by Tim Urban > “I’m fascinated by those rare people in history who manage to dramatically change the world during their short time here, and I’ve always liked to study those people and read their biographies. Those people know something the rest of us don’t, and we can learn something valuable from them. Getting access to Elon Musk gave me what I decided was an unusual chance to get my hands on one of those people and [examine them up close](https://waitbutwhy.com/2015/11/examining-elon-musk-up-close.html). If it were just Musk’s money or intelligence or ambition or good intentions that made him so capable, there would be more Elon Musks out there. No, it’s something else—what TED curator Chris Anderson called Musk’s ‘secret sauce’—and for me, this series became a mission to figure it out. > > The good news is, after a *lot* of time thinking about this, reading about this, and talking to him and his staff, I think I’ve got it. What for a while was a large pile of facts, observations, and sound bites eventually began to congeal into a common theme—a trait in Musk that I believe he shares with many of the most dynamic icons in history and that separates him from almost everybody else. [. . .] > > The thing that tantalized me is that this secret sauce is actually accessible to everyone and right there in front of us—if we can just wrap our heads around it. Mulling this all over has legitimately affected the way I think about my life, my future, and the choices I make—and I’m going to try my best in this post to explain why.” ### Other important reads that you probably already know about, but including anyway for those who haven’t read them: 1. [How to Do Great Work](https://paulgraham.com/greatwork.html), by Paul Graham 2. [Maker’s Schedule, Manager’s Schedule](https://paulgraham.com/makersschedule.html), by Paul Graham 3. [Good Product Manager/Bad Product Manager](https://a16z.com/good-product-manager-bad-product-manager/), by Ben Horowitz 4. [Product vs. Feature Teams](https://www.svpg.com/product-vs-feature-teams/), by Marty Cagan 5. [Aggregation Theory](https://stratechery.com/aggregation-theory/), by Ben Thompson 6. [Status as a Service](https://www.eugenewei.com/blog/2019/2/19/status-as-a-service), by Eugene Wei 7. [1,000 True Fans](https://kk.org/thetechnium/1000-true-fans/), by Kevin Kelly ### What’s missing from this list? Let me know in the comments (along with the impact it had on you). Bonus points for stuff people may not have heard of 👇 [Leave a comment](https://www.lennysnewsletter.com/p/essential-reading-for-product-builderspart/comments) *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [28/46] Build your personal AI copilot *👋 Welcome to a **✨ free edition ✨** of my weekly newsletter. Each week I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lennybot](https://www.lennybot.com/) | [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny) | [Swag](https://lennyswag.com/)*** One of my favorite collaborators, [Tal Raviv](https://www.linkedin.com/in/talsraviv?originalSubdomain=il), is back with another incredible piece that will change how many of you operate at work. If you’re looking for the promised AI productivity gains everyone’s talking about, start here. P.S. You can now also listen to these posts in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads). *For more from Tal, check out his [63 free video tutorials](https://talraviv.co/) on using AI agents at work, his other guest post, [“PM is an unfair role. So work unfairly”](https://www.lennysnewsletter.com/p/product-manager-is-an-unfair-role), and his [episode on Lenny’s Podcast](https://www.lennysnewsletter.com/p/the-super-ic-pm-tal-raviv). You can also book Tal for an [AI build sprint](https://talraviv.co/build-sprints) with your team.* ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/80cae2b6-17fa-4a25-b3c9-f26691a33b56_2913x1876.png) Here’s my embarrassing story: 11 months ago I wasn’t using AI at work—at all. My team and I were building AI products used by tens of thousands of people. But when it came to using AI in my own job, I was a proud Luddite. I didn’t want to sound like “the average of the internet.” I was worried that if I let AI tools do things for me, I’d lose my edge. When I tried using ChatGPT, I found it disappointing for strategic and innovative work—like consulting a chatty Wikipedia. Deep down, I felt discouraged that I didn’t know the exact magic spells of prompting that only influencers seemed to know. Then my engineering team kicked off a new, paperwork-intensive version of Scrum for a large initiative that required me to write dozens of detailed user stories. I couldn’t keep up, and all of a sudden I became my team’s bottleneck. I did what any experienced product manager would do: I whined and complained. Eventually my eng manager, [Oleksii](https://www.linkedin.com/in/oleksii-monchyk-91ba09171/?originalSubdomain=de), took pity on me and showed me how to write detailed, near-perfect user stories with ChatGPT ([which I covered in my discussion on Lenny’s Podcast](https://www.youtube.com/watch?v=wFhurV1l6Jk&t=975s)!). Oleksii instructed me to (1) paste in an example template of a user story and (2) use speech-to-text as I blabbed on and on about how the product experience should work. I also ended up dictating the background story of the initiative to ChatGPT, just like I would at a team kickoff. ChatGPT replied with user stories that were concise, thoughtful, and airtight. The edge cases were extremely relevant, as if it had been a member of the team for months. Not only did the results translate my ramblings to engineering’s structured template, but each user story also included stipulations (mobile responsiveness, accessibility, pricing tiers) that I’d often forget. The results were nearly perfect, and my fingers never even touched the keyboard. ChatGPT was exhibiting superpowers I didn’t know existed. I had a hunch that mere user stories were just the beginning of how I could use it at my job. Could AI help me with higher-level tasks? Could it answer the ultimate question of knowledge work: *What is the most important thing I should do next?* I realized that AI tools had felt like blunt, generic instruments becaus*e* I wasn’t providing enough *context*.LLMs can be highly effective for intelligent knowledge work, but only when we provide them with the same background knowledge that any person would need to do the same work. ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/c10b0bdd-9f55-4579-9395-5f441aa547c0_1388x760.png) Recent advances in LLMs allow us to maintain (and grow) this context over time and across conversations, the same way we might keep a colleague in the loop. And we can even instruct LLMs to use consistent behaviors, values, and guidelines for interacting with us. **When LLMs have this valuable and ongoing context about our goals, our role, our projects, our team, and our wider org, they become our “AI copilot”—a real thinking partner for long-term, complex work.** Remember my fear of losing my edge? When we treat LLMs as partners that inspire rather than replace our thinking, our output only gets sharper. An AI copilot might be the least “[agentic](https://www.lennysnewsletter.com/p/make-product-management-fun-again)” way of using AI, and that’s the point. As AI tools proliferate and become more capable, our AI copilot is there to support us and make us better at our jobs. An AI copilot can work *with* us, not just for us. And that can make all the difference. Over nearly a year now, I’ve taught workshops to more than 20,000 tech employees and founders on how to leverage AI for complex, innovative tasks at work. I can confirm: regardless of company size, vertical, or fluency in AI, many tech workers today are missing out for the same reason I was. They’re not giving LLMs the context needed to experience their full potential. Below, I’ll share how to change the way you think about working with AI, and how to practically construct your own AI copilot. ## What are AI copilots good for? Leveraging a copilot isn’t about AI producing perfect answers—it’s about getting the most out of ourselves. Daniel Kahneman won the Nobel Prize in economics for his collaboration with Amos Tversky, and I love his reflection: “I am not a genius. Neither is Tversky. Together, we are exceptional.” After we work together to build AI copilots, my students start using LLMs to help them make strategic decisions, brainstorm roadmap ideas, develop their soft skills, and even support them emotionally. Their copilots challenge their thinking, analyze data, apply decision frameworks, prioritize, and help rehearse hard conversations. AI copilots also make it infinitely easier to [vibe code an AI prototype](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product), [run meaningful data analysis](https://maven.com/p/eddc27/how-experienced-growth-p-ms-analyze-data-with-ai), [launch an AI automation](https://www.lennysnewsletter.com/p/make-product-management-fun-again), generate AI slide decks, and insert-AI-tool-we-don’t-even-know-about-yet. Because our copilots hold up-to-date context about us and our work, we’re often one chat message away from getting a detailed prompt we can take anywhere else. And yes, AI copilots can also draft PRDs, user research plans, meeting agendas, and Slack updates. (When you regularly loop LLMs into your thinking, generating documents feels like an afterthought.) Months after building a copilot, students tell me they “use it every day” and it’s “always open at work.” Their copilots keep them focused, remind them who to talk to, catch nuances they would have missed, and turn them into superpowered contributors who can analyze data and discuss technical concepts. They move so fast that leadership taps them to teach everyone else how they got so productive. And they do! One student created a [career development copilot](https://www.youtube.com/watch?v=sblODeX-FGk) for women on her team. Another is piloting AI coaching for sales managers doing one-on-ones. A third [built an onboarding copilot](https://darshen.substack.com/p/the-pm-onboarding-copilot) when joining a new company. Two weeks in, when he presented his product strategy, his CEO remarked, “It feels like you’ve been here for months.” ## How do I build an AI copilot? You can build a valuable AI copilot in a number of ways—anything from keeping a long chat thread to creating a [geeky Cursor repository](https://maven.com/p/0a96cb/cursor-isn-t-just-for-coding-how-ai-native-p-ms-work). But my preferred method is using the “Projects” feature in my favorite LLM. Projects are available in [ChatGPT](https://help.openai.com/en/articles/10169521-projects-in-chatgpt), [Claude](https://claude.ai/projects), [M365 Copilot](https://learn.microsoft.com/en-us/training/modules/build-your-first-agent-microsoft-365-copilot-use-copilot-studio/), and [Gemini](https://gemini.google/overview/gems/)’s paid and enterprise plans. (The latter call them “declarative agents” and “Gems,” respectively.) With Projects, building your AI copilot is much like bringing on a new teammate. Each project is made up of three elements: **project knowledge**, **instructions**, and **chat threads**. And each of these elements corresponds with a step in the onboarding process: 1. **Hire your copilot:** Use **instructions** to set your copilot’s role, personality, and behaviors. 2. **Onboard your copilot:** Fill out the **project knowledge** with company and org-level documents that act as shared context across all future conversations with your copilot. 3. **Kick off an initiative:** Begin a **chat thread** for each initiative you’re working on. 4. **Put your copilot to work:** Create simple, conversational prompts. ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/77058e97-c68c-4d2c-8f13-69004a20c5d7_1600x722.png) With this framework in mind, open your favorite LLM (upgrade to paid if you haven’t already), navigate to Project/Gem, and let’s get hands-on. *What if your LLM at work doesn’t have projects? No worries, I’ve included a workaround for you at the end of this post.* ## Step 1: “Hire” your AI copilot ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/158b9226-4b42-4654-83eb-0b561f4218bc_2912x760.png) The first step is to create a new project. I’m naming mine “Tal’s AI Copilot.” Don’t worry about giving it a description. ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/d018f7de-cfcf-41b2-8a21-daeca21fd114_2913x1081.png) Inside our project, we’ll “hire” our thinking partner by deciding what values and behaviors we want it to have. Like when selecting a colleague or a coach, we’ll choose traits that transcend the initiatives we’re working on, company strategy, or what quarter we’re in. Projects allow us to do this by defining **instructions**. Think of these as the “super-prompt” that the LLM applies throughout the conversation, slightly stickier than any prompts you type into a chat conversation. (If you’ve worked on an AI product, you might recognize this as extending the “system prompt.”) ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/43ec5499-416e-4c9c-bd74-8055a809a83b_2913x2141.png) To get you started, copy and paste the following prompt into your project’s instructions. Make sure to fill in the placeholders with what’s relevant to your role: `I am a [role] at [company name], and you are my expert coach and advisor, assisting and proactively coaching me in my role to reach my maximum potential.` `I will provide you with detailed information about our company, such as our strategy, target customer, market insights, products, internal stakeholders and team dynamics, past performance reviews, and retrospective results.` `In each conversation, I will provide you with information about a particular initiative so you can help me navigate it.` `I expect you to: ask me questions when warranted to gain more context, fill in important missing information, and challenge my assumptions. Ask me questions that will let you most effectively coach and assist me in my role.` `Encourage me to: [list the values and behaviors that make you successful in your role]` `I want you to find the balance of: [traits you want in a thought partner and coach that will be both effective and fun to work with]` There’s nothing sacred about this prompt. It’s a starting point, and you should tweak it over time, according to how you want your copilot to behave. Maybe you want it to be sassier, or more skeptical of what you say. Maybe you want it to be more supportive or perhaps more wildly creative (or all of the above!). The key is to make it yours: treat these instructions as a set of knobs you can play with for the lifetime of your copilot. ## Step 2: Onboard your copilot ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/f9d45675-24da-424f-b0c8-bee090bd5e83_2912x760.png) Congrats! You’ve hired a copilot with a stellar personality and aligned values, and now it’s their first day on the job. Time to onboard them to the realities of your organization, team, and role. Think: What would you provide to a new teammate in their first week on the job, before they’re assigned any particular initiative or responsibilities? What’s relevant to and informs all initiatives that you work on? We’ll start by uploading these documents to our **project knowledge**. Here are some examples of valuable context for your AI copilot: - Your company/product’s landing page (use your browser to “save as PDF”) - Your company strategy deck (export as a PDF) - All of your customer segmentation and research insights - Research on your competitive landscape - Company vision, strategy, quarterly planning docs, etc. - Company processes - Your team’s org chart and roles - Your team’s past retrospectives - Your last performance review or informal manager feedback Grab two or three that are readily available; don’t overthink it. ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/48e281ce-6a8d-4766-acda-4aa96034cf44_2927x2147.png) If you don’t have a lot of written documentation available, ask your LLM to interview you to make up for any gaps in its knowledge. Imagine this as having coffee with a sharp colleague a few days after they join the team—they’re probably bursting with questions. `Please review what I’ve shared with you and ask me questions to help complete your knowledge.` `What important information are you missing (company- and product-level) that you need to help me across all my initiatives, present and future?` `Focus more on company, org, and industry level. Focus less on short-term conditions or initiatives (e.g. resources, constraints, or individuals), as those will change over time.` `I want you to ask me these questions sequentially (most crucial questions first), so I can answer one at a time.` When you’re ready to wrap up the conversation, ask it to generate a document that you can upload to your project knowledge: `Please create a document with the new information you’ve learned in our conversation. Only include new information learned in this conversation (i.e. that wasn’t already in project knowledge). Don’t outline any outstanding gaps in the document.` `Optimize the document for adding to the project knowledge, so it can apply in all our future conversations.` `Use exact quotations of my original words for the most salient and important things.` Once the document is ready, download it as a file, and re-upload it to your project knowledge. ## Step 3: Kick off an initiative ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/46468383-2887-4238-a002-158a0fb1cf91_2912x760.png) Now that your new copilot is up to speed on the big picture, it’s time to involve it in a specific initiative. I recommend keeping each initiative in its own **chat thread**. That way, your LLM can track the full context with the least amount of effort from you. You’ll use this dedicated chat thread to kick off, brainstorm, and make decisions with your copilot. It’s where you’ll update your copilot throughout the life of the initiative, and where together you’ll navigate the inevitable twists and turns. (Below we’ll cover what to do if you hit thread limits.) Inside your project, open a new chat thread, and paste in the following prompt: `Now that you have the context on my company and my team, I want to tell you a bit about the initiative that I am working on and give you the specific initiative context.` `This is the starting point of what I know, and I’ll be updating over time as more information and insights come in:` `[Share what you know about this initiative at the start in any order.]` I recommend using the dictation feature (built into ChatGPT, Gemini, and M365 Copilot, or found in Claude’s mobile app) to ramble about what you know to get things started. Let loose about the problem, customer, background, stakeholders, organizational politics—everything that’s on your mind and relevant to this initiative. Don’t worry about structuring your thoughts in a clean, professional way. Just get it out there in a stream of consciousness. I find that when I start to talk, I’m reminded of additional details I may not have included if I’d structured my thoughts prematurely. ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/f605c83d-3a10-427d-bde0-f4861d955655_2913x1019.png) At this point, people usually ask me what model to select. My answer is always: the smartest model available to you at work. We’re not using AI at scale here, so get the best brain you can, even if it’s considered “expensive” or “slow” by engineering terms. (Your only constraints should be where and how your company allows you to upload sensitive information.) To recap: You “hired” your copilot with instructions, you “onboarded” your copilot with project knowledge, and now you’ve kicked off an initiative in a chat thread. With all this context, it’s time for the copilot to dig in. ## Step 4: Put your copilot to work ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/3a0f9f07-29f2-4ff4-97ba-2ff551b4cb48_2912x760.png) With your initiative thread kicked off, paste the following into the chat: `What is the single most important thing I should do next?` Take a look at the results. Can you imagine sending that prompt to a blank LLM chat thread (before you had a copilot)? Since you’ve provided ample context and instructions, your copilot returns thought-provoking reflections, spot-on questions, and concrete next steps. In my case, since I told my copilot about my team members and stakeholders, it specifically calls out my colleagues by name (!). When we approach AI as a thinking partner full of relevant context, it can serve as endless inspiration. Here’s a list of use cases and example prompts to get your imagination going: ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/d7d8cdc0-8f0a-4c6d-b910-59b4715885ed_2912x5873.png) Notice how there’s nothing fancy about these prompts. Since the LLM has so much context, conversational one-liners are often all you need. Expect surprising and delightful moments. In the words of [Karina Nguyen](https://youtu.be/DeskgjrLxxs?si=4boISMhdv6pSkiOc&t=3071), an AI researcher who’s worked at both Anthropic and OpenAI, “Models are really good at connecting the dots.” ## Other ways to leverage your copilot ### Using your copilot to build AI prototypes Using [AI to prototype](https://www.lennysnewsletter.com/p/a-guide-to-ai-prototyping-for-product) shortens the path to making something people want, in three ways: 1. Iterate with yourself: hone your vision 2. Iterate with your team: keep conversations grounded 3. Iterate with customers: validate usability and viability ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/59df72c6-3ba5-41f2-8767-75fedfe19ea5_2913x1525.png) AI tools for building prototypes (i.e. vibe coding) have gotten so good at this that the hardest part is articulating what you want. Since our copilot has been with us on our entire journey, it becomes much easier to come up with a detailed description of our solution. Once you’ve converged on a direction, ask your copilot to create a prototype directly in the chat (e.g. Claude Artifacts or OpenAI Canvas). Continuing in the same conversation thread, fill out the following prompt: `Help me generate an interactive prototype of this idea.` `We are going to build only the interactive, client-side prototype version of this (without deep functionality) to serve as a good internal feedback tool that can also help test the usability with customers.` `Keep the scope extremely focused on what we’ve discussed explicitly. “Happy path” only for now.` `[Optional: I have attached a screenshot of the current interface. Make sure it looks pixel-perfect and as similar as possible to the original.]` `Don’t build it yet; make a plan first. If needed, remind me what other files or images or resources I should provide you to help you do an amazing job.` You can also have it generate a prompt to use in an external text-to-UI tool like v0, Replit, Lovable, Bolt, and so on. Just append the following lines to the prompt above: `Create a separate document with a prompt that I can directly paste into an external AI prototyping tool (all commentary should be outside, so I can just copy and paste the doc as is). This should be a developer-ready specification that can be used to build this interactive prototype.` `Assume the tool has no prior context except what you include in the prompt.` ### Using your copilot to build AI automations AI automations can take repetitive, energy-draining tasks off our plate. Unfortunately, they’re not magic genies, and shine under specific conditions. This makes it hard to get started, because the first challenge is thinking of a use case that is both genuinely valuable and also a good fit for AI automation platforms. Did someone mention *generating ideas based on personalized context*? Sounds like a job for your AI copilot. You can use this prompt in an existing chat thread or start a new one as a throwaway brainstorm. Either way, it’ll base its ideas on your project knowledge: `Based on what you know about me and my organization, please brainstorm five ideas for an AI automation I can build using platforms such as Zapier Agents/Lindy AI/Relay app/Cassidy AI/Gumloop/etc. Limit yourself to what makes sense to build in those platforms.` `These should help me, as a product manager, save time on draining yet essential tasks that take me away from more valuable, strategic, and creative uses of my attention and energy.` `Ask yourself: What ongoing repetitive work requires some judgment and writing abilities but not my full expertise and intuition? Put another way, if my company assigned me a junior intern, what would I have them do?` `Try to phrase each idea in one or two sentences, exactly the way you would in a message asking a junior intern to do something. Additionally, follow up each idea with a brief explanation of why you chose this over other ideas. Pitch me as the product manager in terms of business outcomes and the value of my time.` `# IMPORTANT: These should be event-driven AI automations, not batch tasks.` `Only suggest event-driven automations that process items one at a time as they arrive——with currently available information. Do NOT suggest batch tasks that process multiple items on a schedule (e.g. “every morning, scan all...” or “weekly, compile...”).` `[Why: AI automations shine in one-at-a-time, repetitive tasks. They do best when designed for immediate responses to individual triggers.]` `❌ WRONG (Batch Task): “Every morning, scan all new support tickets and summarize them”` `✅ RIGHT (Event-Driven): “When a new support ticket arrives, analyze it and alert me if it’s urgent”` `Before finalizing your suggestions, verify each one:` `- Does it trigger on a specific event? ✓` `- Does it process one item at a time? ✓` `- Could it run multiple times per day as events occur? ✓` `If any answer is “no,” revise it to be event-driven.` `The only exception to this is if the end-result value is being delivered on a particular timeline, for example to centralize information and reduce noise (e.g. weekly update, daily brief on upcoming events, or weekly digest of changes). If you believe that a use case falls under this exception, please note your reasoning.` `# Examples` `Below are examples of use cases where product managers have gotten a lot of value from AI agents. Remember, these are just examples to inspire you to think of use cases that make sense for you.` `1. Compile fragmented information that would require a lot of clicks:` `“When a new message is posted in the #feature-requests Slack channel, distill the customer request into 2-5 keywords. Search those keywords in recent Slack threads, HubSpot conversations, and Gong snippets, and reply to the thread with links to what you find.”` `“Every morning, scan my calendar for customer calls, and instead of searching the web, DM me with recent interactions from this customer in Salesforce, Gong, and Zendesk.” (Example of an exception where the value is the daily delivery timeline to centralize information.)` `“Every Monday morning, prepare a competitor activity digest by scanning recent blog posts, App Store updates, and X announcements.” (Example of an exception where the value is the weekly delivery timeline to reduce noise.)` `“When a customer churns, post a message in the #churn-lessons channel with recent support interactions, NPS rating and date, and churn survey response.”` `2. Boring, Sisyphean tasks with high upside:` `“Monitor the pricing pages of 5 competitors for changes.”` `“When a bug nears its SLA deadline for the associated customer, ping this dedicated Slack channel and cc each respective customer success representative.”` `3. Scanning exhausting amounts of data:` `“DM me when a support case is resolved as a ‘product confusion’ reason rather than a technical issue.”` `“Ping this channel when you see a feature request that showed up for the first time.”` `“When an NPS survey text response is posted in the Slack channel, decide if it’s clearly a technical issue and, if so, create a support ticket in Zendesk.”` `4. Drafting updates:` `“Every Friday at 10 a.m., write up a summary of progress made across all teams in our project board, across all epics, changes made to scope, and highlighting any timeline changes.” [Example of an exception where the value is the daily weekly timeline to centralize information]` Notice the pattern? Good automations start with “When [specific event happens],” not “Every [time period],” unless there’s specific value in triggering on a daily/weekly/etc. cadence. Scan the resulting use cases and use them as inspiration. Have a conversation! Engage with the results and push back on parts that don’t sit right with you. The goal is finding a use case you’re excited to make happen. With this in hand, turn on deep research mode, and ask your copilot to [generate a personalized, step-by-step tutorial](https://www.lennysnewsletter.com/i/162096097/ai-agent-builder-prompt) on how to construct the automation. ### Gossiping to your AI copilot Shit happens. A stakeholder changes their mind, new data arrives, or things turn out to be way higher-effort than planned. One hallway conversation can upend everything. For your AI copilot to help you navigate, you’ll need to loop it in. I call this habit “gossiping” to AI, because I fill in my copilot the same way I’d lean over and vent to a colleague sitting next to me. I love using speech-to-text here as well: “You won’t *believe* what just happened in my conversation with so-and-so. . .” That’s it. No structured format, no formal update process. Just the exasperated sharing that comes naturally. Even if I don’t have any action items, my AI copilot remembers and references these updates later. Unlike with most coworkers, you can bluntly add: `I don’t want solutions right now; I want you to listen. Confirm with just a “yes.”` Gossip keeps the context fresh and effective. Pro tip: I personally use my LLM’s native mobile app for dictation on the go. This makes gossiping feel as lightweight as sending a WhatsApp voice message to a friend or coworker. ## Habits for getting the most out of your AI copilot Getting the most out of our copilot depends on us, not the AI. The difference between an okay AI copilot and an exceptional one isn’t the tech—it’s the habits we build around it. ### Cultivating an “AI mindset” Your AI copilot will undoubtedly be delightful, but it most certainly won’t always be *right*. That brings me to the most important component of using an AI copilot: our mindset. Working successfully with a thinking partner—human or machine—is about taking the good and disregarding the bad. If our car’s driver assist beeps loudly on a wide-open highway, it’s no harm done. We still keep it activated in case it alerts us to something on the road we didn’t see. Or think of your favorite mentor or confidant: we don’t constantly judge if their advice could “replace our thinking.” Rather, we focus on the parts of what they say that are valuable and inspiring. Even when they’re wrong, LLMs can be weirdly inspiring and motivating. Often when I disagree with what my AI copilot suggests, it still gets me thinking. As with any outside advice, it spurs me to crystallize what *I* thinkI should do instead. It’s up to us to apply taste and judgment based on hard-earned product intuition. To quote David Lieb (YC partner and founder of Google Photos), “Your gut is the world’s most sophisticated machine learning model ever created.” When working with generative AI, apply your craft and taste to identify what’s valuable. If you’re frequently getting outputs you’re not happy with, ask yourself, “What context does the AI need to succeed here?” or “Does it have enough guidance to deliver results?” Since we’re using LLMs for their intelligence (rather than for their knowledge), it’s on us to think about what we should provide to set them up for success. To tweak your inputs, hover your mouse over your last chat message and click the “edit” button (or ✏️ icon) found in all major LLMs. Provide a tiny bit more guidance or background information, and submit again. If you find the same gap happening over and over again, consider adding that information to the project knowledge. ![Image from Build your personal AI copilot](https://substack-post-media.s3.amazonaws.com/public/images/cf84c2c6-c800-4458-860e-6bc3bd25328d_2913x419.png) ### Growing your copilot’s knowledge over time Our investment in our AI copilot also earns compound interest. Beyond regularly providing context in each initiative thread, we can also make sure our project knowledge continues to grow, instantly benefiting current initiatives as well as those to come. The end of an initiative is a great opportunity to update the project knowledge. Start by sharing the outcome, good or bad. Include any retrospectives your team ran. Turn on speech-to-text and add your personal reflections. Finally, ask it to output a “new lessons learned” document. You can guess what’s coming next: download the document, and upload it to the project knowledge. When we add real-world lessons to our project knowledge, we accelerate the opportunities for our copilot to “connect the dots” and surprise us with delightfully wise counsel. You can think of this as memory, but a transparent, curated memory that’s completely under your control for a professional context. ### Note: What if you hit chat size limits? Different LLMs have different context window limits. If you run into a limit, you can use a prompt to have the LLM summarize the conversation so you can start a new thread with the key context carried over. To use this prompt, hover your mouse over your last chat message and click the “edit” button (or ✏️ icon). Replace with the following prompt, and hit enter. Finally, start a new thread using the output. `This conversation has reached its context limit. Create a document that can serve as the initial context for a fresh blank LLM thread. Your goal is to preserve approximately 90% of the conversation’s value and context while reducing its length by ~90%.` `Act as an expert handing off to another expert who will help me, and set them up for maximum success, as close as possible to having been there all along. Tell the new expert what instructions or behaviors to exhibit that I’ve implicitly or explicitly requested of you. Tell a chronological story.` `Use your best judgment when it would be particularly valuable to use the original exact words of either the user or AI.` `Skip any context that is already found in the project knowledge or in your system instructions, since this will be available automatically to the new thread.` `Create a second, separate document with what you chose not to include, and why.` ## Where might copilots evolve next? It feels like we’re only at the beginning. Every day I use my AI copilot, I bump into moments where I think, “I wish it could just. . .” From what I hear from my workshop participants, I’m not alone. Today’s friction points reveal where the real opportunities lie for both founders and incumbents alike, and many startups today are gunning for these opportunities from different angles. Here’s my wish list for what comes next: ### Make it easier to update project knowledge Today, I’m the human API between my copilot and everything else. I manually recount what happened in conversations, export strategy decks as PDFs, and update my project knowledge when I remember to. It works, with a lot of work. Imagine if your copilot could pull directly from new department templates, project management, or team messaging, dynamically updating as your world changes. I wish it could have all this information without getting confused, overwhelmed, or over-indexing to irrelevant details. I want information to flow the other way as well: what if thinking, planning, and [scrapbooking](https://www.lennysnewsletter.com/i/151079492/get-a-head-start-on-discovery-with-product-scrapbooking) could happen in the same place? But what really excites me is transparent, editable knowledge that feels like a living document. Imagine today’s consumer “memory” features, but with professional-grade control. I’m inspired by how Cursor works for coding: imagine *selective* context, where you choose which documents and initiatives your copilot considers for each conversation. Sometimes you want one feature initiative to take another into account; sometimes you don’t. ### I want to benefit from being on a team What if every teammate joining you had a copilot waiting for them that was 80% ready to rock? Instead of starting from scratch, picture shared knowledge layers curated by leadership. You’d still have your personal context, but now it sits on top of your team’s collective intelligence. New PMs would start with company-specific templates, relevant frameworks, and real lessons learned. Over time, our copilots become smarter not just from individual experiences but from your entire team’s wins and failures. ### Push me Even with the best-constructed copilot, there’s that moment of writer’s block. What should I update it on? What question do I even ask? Building the habit is harder than building the tool. I would love for AI to push me like a sharp chief of staff that doesn’t wait to be asked. Imagine checking in based on your calendar: “Hey, you have that stakeholder review tomorrow. Want to role-play scenarios?” Or pushing me to focus: “This would be a really good time to schedule user research and get input from the sales team.” The opportunity isn’t in making copilots smarter, but rather in making them more connected, more collaborative, and more proactive. The future of AI at work may be more about asking better questions of us than offering better answers. ## Context changes everything Using AI at work shouldn’t mean sifting through generic responses or conjuring perfect prompts. The most fruitful way to think about using LLMs is like collaborating with an inspirational colleague who knows your company strategy, remembers that conversation with your manager from last Tuesday, and can challenge your assumptions when everyone else has gone home. When we give AI the context it needs—and treat it like a partner sitting next to us rather than a magic oracle—we don’t lose our edge; we sharpen it. *Thank you to [Alex Furmansky](https://www.linkedin.com/in/alexfurmansky/) for being an amazing (human) thinking partner for these ideas. And thank you to [Colin Matthews](https://maven.com/tech-for-product/ai-prototyping-for-product-managers?promoCode=LENNYSLIST) for lending his deep expertise on AI prototyping.* ## Appendix: What if my LLM doesn’t have “projects”? While projects are supported by the major LLM providers on all paid plans, many tech workers (especially in Big Tech) don’t have access to projects as part of the internal tools they’re required to use to access LLMs. If this is you, don’t worry. Projects are not a fundamental LLM capability; they are a UX convenience. The simplest solution is to do things manually. Assemble your instructions and project knowledge in one long text file. Each time you start a new chat thread, start by pasting in the entire file. This is a decent approximation of how projects work behind the scenes (and way better than working without context). It’s not elegant, but hopefully you won’t have to do this very frequently (more on chat threads later). For those who want a more adventurous solution: A few Big Tech employees have realized that while they can’t access projects, they *do* have access to AI-powered software development environments, like Cursor. They’ve realized that those contain all the same ingredients as the projects feature (knowledge, instructions, and chat threads), and they’ve repurposed these tools as thinking partners (using them with plain English instead of code). (If the geeky approach sounds appealing, I’m hosting a [free lightning lesson featuring live hands-on demos](https://maven.com/p/0a96cb/cursor-isn-t-just-for-coding-how-ai-native-p-ms-work) of how people are doing this.) *Thanks, Tal!* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [29/46] Lenny's Product Pass: 20+ free premium products, available exclusively for paid annual subscribers ![Image from Lenny's Product Pass: 20+ free premium products, available exclusively for paid](https://substack-post-media.s3.amazonaws.com/public/images/b3576ec2-7a10-4ad4-b57c-2446802ffe4f_1200x630.png) #### **I’m excited to announce the newest, most unbelievable collection of premium products now available exclusively to paid annual subscribers.** **Paid subscribers get a free year of (while supplies last):** 1. **Research:** [Manus](https://www.manus.im/campaign/lenny), [Perplexity](https://www.perplexity.ai/pro) 2. **Design:** [Canva Business](https://www.canva.com/canva-business/)\*, [Framer](https://framer.link/RmH3UwE), [Gamma](https://gamma.app/partners/lenny)\*, [Mobbin](https://mobbin.com/lenny) 3. **Build:** [Lovable](https://lovable.dev/lenny)\*, [Replit](https://replit.com/)\*, [Bolt](https://bolt.new/)\*, [n8n](https://n8n.io/), [Amp](https://ampcode.com/), [Factory](https://factory.ai/product/ide), [Devin](https://devin.ai/)\*, [Warp](https://go.warp.dev/lenny), [Magic Patterns](https://magicpatterns.link/lenny), [ElevenLabs](https://elevenlabs.io/)\* 4. **Scale:** [Railway](https://railway.com/) 5. **Track:** [PostHog](https://posthog.com/) 6. **Collaborate:** [Linear](https://linear.app/), [Wispr Flow](https://wisprflow.ai/), [Granola](https://www.granola.ai/), [ChatPRD](https://www.chatprd.ai/lenny) 7. **Incorporating your startup:** [Stripe Atlas](https://stripe.com/atlas)\* *\* These products are available for [Insider-tier](https://www.lennysnewsletter.com/subscribe?plan=founding) subscribers only* **→ Already a paid subscriber?** [Grab your free products here](https://lennysproductpass.com/). **→ Not yet a subscriber?** [Subscribe](https://www.lennysnewsletter.com/subscribe) and then [grab your free products here](https://lennysproductpass.com/). This curated collection includes the most important, innovative, and beautifully crafted products out there—some that you already love, some that are under the radar, and all of which I recommend. **The collection is worth over $25,000, for just $200-350/year (depending on your tier).** Yes, this is for real. It’s arguably the most ambitious product bundle ever attempted, and none of these companies have ever offered this deal on their products before. *Offer codes are limited, and granted on a first-come, first-served basis. So act quickly. We’ll be adding new products quarterly, and products will rotate in and out of the collection.* ### **Why am I doing this?** Things are moving fast. Jobs are disappearing, new skills are emerging, and every week, a new tool launches that we need to become experts in to stay relevant. What we need most to navigate this moment is practical advice from people in the trenches, community to support us through all the change, and access to the latest tools and technologies to truly understand where things are heading. That’s precisely what I’m trying to provide here. We’ve already built the world’s largest newsletter and podcast for product builders, and a thriving global Slack community of tens of thousands (with local meetups, AMAs, mentorship programs, and more). Now, I want to do more to increase access. **Think of your Lenny’s Newsletter subscription as your product builder’s starter pack: all the advice, community, and tools you need to build and grow a world-class product (and career). For just $200-350/year.** In the coming months, I’ll share guidance on how to take full advantage of these products to help you work better, deeper, and faster. ### **Important offer details (read before subscribing):** 1. **You must be a new customer of the products** to take advantage of the free year. If you’ve already paid for one of the products before (monthly or yearly), or redeemed a code from our previous bundle, you won’t be able to get a free additional year of that specific product. 2. **You must get an** **Annual or Insider subscription** **to Lenny’s Newsletter** to be eligible for this offer. Monthly subscribers aren’t eligible. 3. **Both existing and new subscribers** of Lenny’s Newsletter are eligible for this deal. Existing subscribers can [redeem the offer here](https://lennysproductpass.com/). 4. **We may run out of codes for a product at any time**. Your code is only secured once you [claim it](https://lennysproductpass.com/). You need to claim each code individually, so hop on the products you are most excited about. 5. **Your access to the Product Pass is contingent on your subscription remaining active.** If you cancel or don’t renew your paid subscription at any time, your codes may be deactivated. 6. **Your one free year of the product begins when you redeem the deal on the partner’s website**,not when you purchase this newsletter subscription or claim your code. 7. **All sales are final. We won’t issue refunds if a couple of products are out by the time you claim them.** The deal is already so good, even if a couple of products are missing. If you request a chargeback for your subscription, all of your codes will be deactivated. ### **Introducing the Insider tier: priority access to every product** Since our offer codes are limited (i.e. our partners aren’t giving us unlimited free accounts), we’re expecting to run out of some of the more popular products. To make sure you get access to everything you want (without worrying about missing out), **we’re introducing the Insider tier—for $350/year—which includes two benefits:** 1. **Priority access:** While Annual subscribers get access on a first-come, first-served basis, Insider members get front-of-the-line access to every tool we have in the collection. Which means you don’t have to feel stressed about missing out on a product. 2. **Exclusive premium partner offers:** We will be adding a limited number of select premium partner products on an ongoing basis, which will be available *exclusively* to Insider members. **Over $25,000 in value for just $350/year—without the stress.** Many readers expense this subscription out of their workplace’s learning and development budget. [Here’s an email to send to your manager](https://docs.google.com/document/d/1o_Szso0-TQAEEBCxKwvLHTm1fzE4BMiSjeYIKOfUYl0/edit). If you’re already an annual subscriber, you can upgrade [here](https://www.lennysnewsletter.com/subscribe?plan=founding). ### **💥 How to redeem the offer 💥** - **Step 1:** [Get an Annual or Insider subscription](https://www.lennysnewsletter.com/subscribe?https%3A%2F%2Fwww.lennysnewsletter.com%2Fsubscribe%3F=) to Lenny’s Newsletter. - **Step 2:** [Redeem your codes here](https://lennysproductpass.com/). - **Step 3:** Profit. **To summarize, your paid subscription now includes:** 1. A deeply researched newsletter and podcast with a 5+ year back catalog of evergreen content 2. An invite to a 30,000+ member-only global Slack community for product builders, with local meetups, AMAs, mentorship programs, book clubs, and more 3. This growing collection of free world-class tools It’s everything you need to build a world-class product—and career. *If you’ve already got a subscription, consider [gifting a subscription](https://www.lennysnewsletter.com/subscribe?gift=true) to a colleague or [purchasing a group subscription](https://www.lennysnewsletter.com/subscribe?group=true) for your team.* Sincerely, Lenny --- ## [30/46] 25 proven tactics to accelerate AI adoption at your company What does it take to drive employee AI adoption at your company? Longtime product operator and creator [Peter Yang](https://www.linkedin.com/in/petergyang/) has been investigating, sharing, and applying AI best practices for years now, and he’s been getting this question more than any other. So have I! We decided to collaborate to bring you the best, most in-depth answer to this increasingly important topic. Peter interviewed founders and product leaders at six fast-growing AI-forward companies and collected all of their best tactics for driving employee AI adoption. This post includes their best 25 battle-tested AI tips that you can implement right away at your company. ![Image from 25 proven tactics to accelerate AI adoption at your company](https://substack-post-media.s3.amazonaws.com/public/images/390597ea-9c63-44f2-953e-9031e8ed1abd_2912x1047.png) Peter shares more practical AI tips via his [newsletter](https://creatoreconomy.so/) and [YouTube channel](https://www.youtube.com/@peteryangyt). P.S. You can now also listen to these posts in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads). ![Image from 25 proven tactics to accelerate AI adoption at your company](https://substack-post-media.s3.amazonaws.com/public/images/c8508040-e73d-4398-b6c8-1ebf56b2b7b5_2912x2393.png) I never feel alone anymore when I’m stuck on a problem. From summarizing user feedback to crafting product strategy, having AI as a patient, knowledgeable, and always-available thought partner has completely transformed how I do product work. But here’s the reality: According to a recent [Gallup poll](https://www.gallup.com/workplace/691643/work-nearly-doubled-two-years.aspx), only 8% of U.S. workers use AI daily. Meanwhile, teams that crack the AI adoption code are seeing incredible results. For example, Zapier’s sales reps save 10 hours per week on lead research, Ramp built AI user personas that give PMs instant feedback on any spec, and Duolingo went from 100 courses in 12 years to 150 courses in just 12 months with AI’s help. From what I’ve seen and from talking to leaders at AI-forward companies, I’ve learned that: **The biggest barrier to AI adoption isn’t technology; it’s organizational change.** At many companies, employees are struggling with vague AI mandates, procurement bottlenecks, and lack of guidance on the highest-impact use cases to focus on first. After interviewing leaders at the 6 AI-forward companies below, I’ve identified 5 recurring steps that companies can take to supercharge AI adoption: 1. Explain the *how* 2. Track and reward adoption 3. Cut the red tape 4. Turn enthusiasts into teachers 5. Prioritize the high-impact tasks ![Image from 25 proven tactics to accelerate AI adoption at your company](https://substack-post-media.s3.amazonaws.com/public/images/a57910f7-d2b5-49a4-bea2-cc2f120e4986_2912x2912.png) Let’s break down exactly how these companies made it work. ## **1. Explain the** ***how*** Saying “we are AI-first” means nothing if employees don’t know what that actually means for their day-to-day work. The companies that succeed provide **specific tactics** that employees and teams can adopt to meet those expectations. Here’s what this can look like: 1. **Include specific tactics in your memo:** Tobi Lütke, CEO of Shopify, didn’t just say that “using AI is now a baseline expectation” in his now-famous [memo](https://x.com/tobi/status/1909251946235437514?lang=en). Instead, he shared concrete tactics he expects to see, like making AI prototyping part of the company’s GSD (get shit done) process. 2. **Declare a “code red” moment:** Wade Foster, CEO of Zapier, called an all-hands-on-deck moment in March 2023 after ChatGPT’s launch. He then shared a [playbook](https://docs.google.com/document/d/1zbkjL0d-ev87PO0yKJBBMlliByVaUIREKNQfvtVwXbg/edit?tab=t.0#heading=h.x6u8m91ksli4) and gave all employees a week off to put it into practice. 3. **Define what “embracing AI” means:** Luis von Ahn, Duolingo’s CEO, defined AI adoption as both “making our products better” and “empowering employees to do their best work.” Teams were encouraged to use AI for everything from speeding up lesson creation to prototyping. 4. **Embed with individual teams:** Darragh Curran, Intercom’s CTO, set a [goal](https://fin.ai/ideas/2x/) to “2x productivity with AI” and then spent a week every month embedded with individual teams to identify and execute on the 2x opportunities. 5. **Lead by example in real time:** When a PM brings a problem to Hilary Gridley, Whoop’s Head of Product, she’ll say, “Want me to show you how I solve this with AI?” Then she shares her workflows live. You don’t need to be an executive to drive this change. At Roblox, I constantly share my favorite AI workflows in Slack and one-on-ones. I’ve even started screen sharing AI’s output during meetings to solve problems with my teammates in real time. Seeing a colleague save time with AI is an incredibly strong incentive to try it yourself. ## **2. Track and reward adoption** Like any good PM, you should track AI adoption as inputs (who’s using AI) and outputs (what business value it’s creating). You should also reward employees who are leading the charge to keep the momentum going. Here’s how top companies are tracking and rewarding adoption: 1. **Make AI adoption part of performance reviews:** Shopify asks employees to rate colleagues on a 1-to-5 scale for how well they “reflexively use AI tools for improving and amplifying work outputs.” 2. **Publish AI usage by team:** At Ramp, leadership shares the number of AI power users (5+ actions a week) for tools like Cursor, Claude Code, and ChatGPT. This transparency creates natural accountability across teams. 3. **Track team-specific impact:** Zapier tracks the impact of AI adoption by function. In sales, for example, when targeted leads engage with marketing content, AI [auto-packages](https://zapier.com/templates/details/target-account-engagement-alert-rep-outreach-kit) that information for the account rep—leading to 10 hours saved per week per rep. 4. **Use proxy metrics for productivity:** Intercom tracks merged pull requests as a proxy for productivity gains. They’re already seeing a “durable improvement (about 20% year-over-year)” from AI-assisted development. 5. **Make it a daily habit:** Whoop gave employees a [30-day challenge](https://docs.google.com/spreadsheets/d/1zJ4rbi9YcQuGqGxc6-AQD0-44oT9l4Eyono0AdpgJbA/edit?gid=0#gid=0) with bite-size 2-minute tasks to complete and rewarded those who kept the longest streak. The point is that: **People will change their behavior with the right incentives.** Tracking both the leading indicators (AI usage) and lagging indicators (business results) creates a link between AI adoption and the real impact on teams, businesses, and people’s careers. Connect the dots on impact for your employees and you’ll see adoption skyrocket. ![Image from 25 proven tactics to accelerate AI adoption at your company](https://substack-post-media.s3.amazonaws.com/public/images/89cd76c0-6121-4d52-abf3-53a6fc7ce7f9_1600x1062.png) ## **3. Cut the red tape** Most companies have long approval processes for AI tools. But what they don’t realize is that their employees are already using AI. They’re just using it from their personal accounts. Cut the red tape if you don’t want employees to use AI tools that aren’t approved: 1. **Create an AI learning budget:** Duolingo gave every employee $300 to try AI tools, courses, and subscriptions. This incentivizes constant experimentation. 2. **Assign a lead to expedite approvals:** Zapier assigned a lead PM to own working with procurement, legal, and engineering to fast-track AI tool approvals and eliminate bottlenecks. 3. **Give employees time to tinker:** “No time” was the main reason employees cited for not trying new AI tools, so Intercom CTO Darragh encouraged managers to give employees dedicated time to skill up. 4. **Provide multiple tool options:** Shopify provides access to a wide range of tools, including Claude, Perplexity, Cohere, Gemini, Cursor, Copilot, and Claude Code. They also encourage employees to contribute to a growing library of AI prompts and agents. 5. **Embrace internal enthusiasm:** Whoop lets employees nominate tools they’re excited to trial, like Fireflies for note-taking and Zapier for automation. If you’re wondering which tools to approve, here are my personal favorites: ![Image from 25 proven tactics to accelerate AI adoption at your company](https://substack-post-media.s3.amazonaws.com/public/images/6799a3f4-629a-44f0-8916-50de2e9fb6f3_2912x2122.png) As an employee, don’t be afraid to champion your favorite AI tools even if you’re not responsible for procurement. ## **4. Turn AI enthusiasts into teachers** Every company has AI power users who are just itching to share what they’ve learned. Set up the right channels and resources for them to train everyone else. Here’s how to turn your AI enthusiasts into teachers: 1. **Run AI learning villages:** Shopify hosts dedicated AI training at company events, with hands-on challenges and subject-matter experts. They have a “build in the open” culture where teams constantly share AI experiments. 2. **Host FriAIdays and show-and-tells:** Duolingo protects two-hour blocks every Friday for AI experimentation and demos. They also do “AI Show and Tell” at all-hands meetings. 3. **Create an AI SWAT team:** Ramp’s cross-functional team identifies each group’s top AI automation need and builds solutions. They also run PM AI offsites for vibe coding. 4. **Host regular demos and hackathons:** Zapier hosts live demos that attract over 60 employees every week and quarterly hackathons where the best prototypes are presented to leadership. 5. **Make “upleveling others” part of promotion:** Helping others get better at using AI is a great signal of leadership. Hilary at Whoop looks for examples of how employees improve others’ AI fluency when reviewing promotions. An AI-forward culture is one where employees feel comfortable sharing not just what they’ve built but how they did it. At Roblox, we have an “#AI-tips-and-workflows” Slack channel where employees share everything from basic prompting tips to weekend AI projects. ## **5. Prioritize the high-impact tasks** Most companies fail at AI adoption because they try to do everything at once. Instead, find the high-volume and repetitive tasks that are taking up the most employee time and streamline those first with AI. Here are the tasks that AI-forward companies are focusing on: 1. **Automate support triage workflows:** Zapier built a workflow that checks Zendesk tickets and identifies customers ready for sales conversations, turning customer support into a revenue driver. 2. **Streamline performance review prep:** Shopify built an AI tool to help employees summarize feedback and identify strengths and working areas. This cuts review time and improves discussion quality for managers and their reports. 3. **Create content with AI:** Duolingo rebuilt their course content creation process with AI, going from 100 courses in 12 years to 150 courses in just 12 months. From CPO Cem Kansu: “We vibe coded the first Duolingo Chess lesson in hours instead of weeks.” 4. **Build AI personas:** Ramp built AI personas loaded with all their user research context. PMs can now give any product spec to these personas and get instant feedback on what they’re missing or haven’t thought through. 5. **Empower employees to build internal AI tools:** A PM on Hilary’s team at Whoop built an evaluation tool that tests whether prompts perform better across dozens of persona types—without going through engineering. ![Image from 25 proven tactics to accelerate AI adoption at your company](https://substack-post-media.s3.amazonaws.com/public/images/0b3cce24-1481-4267-832c-aa5603f00f11_1600x1040.png) When it comes to building internal (and external) AI tools, here’s what separates real adoption from theater: **Instead of focusing on shiny demos, look for rigorous data and evaluations.** I’ve seen too many PMs present flashy demos to executives only to stumble when asked about data and evaluations (including some lessons I learned the hard way!). When you share your AI product, you should clearly answer four questions: What customer problem are you solving? Can AI solve it better than non-AI solutions? What ground-truth dataset and evaluations do you have? How have you prepared for the model to fail? True AI adoption happens when teams become skeptical of flashy demos and demand to see the rigor behind AI products. When employees start asking about accuracy metrics, evaluation frameworks, and failure modes—that’s when you know AI adoption is real. ## **Where to go from here** The companies succeeding at AI adoption understand that it’s not just about making vague AI mandates. Instead, they explain the “how” with detailed tactics, and track both usage and impact. They eliminate procurement roadblocks, turn enthusiasts into teachers, and build AI products with rigorous evaluations to streamline high-impact tasks. The goal is to empower every employee to work with AI to solve their problems. Someone in sales, product, or operations might create an AI tool that transforms your business. What tactics are you implementing at your company? I’d love to hear about your wins (and failures) in the comments. *Thanks, Peter! For more from Peter, check out his [newsletter](https://creatoreconomy.so/) and [YouTube channel](https://www.youtube.com/@peteryangyt).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [31/46] Essential reading for product builders—part 2 *👋 Welcome to a **🔒 paid edition 🔒** of my weekly newsletter. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lennybot](https://www.lennybot.com/) | [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny)*** *P.S. Annual subscribers get a free year of 15+ premium products: **[Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass)** (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** Less than a month ago I published [part 1 of this series](https://www.lennysnewsletter.com/p/essential-reading-for-product-builderspart), and it’s already my [9th most popular post of all time](https://www.lennysnewsletter.com/archive?sort=top). This tells me there’s a growing need for curated, thoughtful content as an antidote to the endless slop filling our feeds and inboxes. To continue building an essential reading library for product leaders, I’ve picked 10 more timeless reads that you probably haven’t read but should. This includes a handful that popped into my mind after I published part 1, suggestions you sent me, and a few honorary mentions for extra credit. The pieces below cover a wide spectrum of advice around growth, leadership, communication, entrepreneurship, and more. I’m not including books here—that list is yet to come. If you have suggestions for essays I’m still sleeping on, please share them in the comments. 🙏 [Leave a comment](https://www.lennysnewsletter.com/p/essential-reading-for-product-builderspart-1ac/comments) *You can listen to this post in convenient podcast form here: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads)* ![Image from Essential reading for product builders—part 2](https://substack-post-media.s3.amazonaws.com/public/images/77f6a1f9-9234-41fc-b5ee-d4e1c0e3d07b_2912x1940.png) ### **1. [Building Products](https://medium.com/the-year-of-the-looking-glass/building-products-91aa93bea4bb), by Julie Zhuo** > 1. “A product succeeds because it solves a problem for people. This sounds very basic, but it is the single most important thing to understand about building good products. > 2. The first step in building something new is understanding what problem you want to solve, and for whom. This should be crystal clear before you start thinking about any solutions. > 3. The second question you should ask yourself is ‘Why is this particular problem worth solving?’ > 4. If the audience you are building for is narrowly defined (and one that you are a part of), then you may be able to rely on your intuition to guide your product decision-making. If not, then you should rely on research and data to inform your decisions. > 5. If you are a start-up founder, your path will be easier if you go after a problem for a narrowly defined audience, and then expand to broader audiences after you have some initial traction. > 6. The problem you’re trying to solve should be easy to communicate in a sentence or two and resonate with someone from your target audience. If not, consider that a big red flag.” ### **2. [Communication Is the Job](https://boz.com/articles/communication-is-the-job), by Andrew Bosworth** > “When I’m doing a poor job of communicating it can feel like I’m pushing with a rope. I have some clear vision in my head and people just aren’t doing what I expect. It can be a frustrating experience and it is tempting to blame the audience for not understanding. But make no mistake, when this happens, it is your fault. You have to sit down and ask questions from a place of humility to hear what they took away from what you said. Take full responsibility for any discrepancy from what you intended and make corrections with your entire audience.” ### **3. [Executive Communication](https://www.heavybit.com/library/video/executive-communication/), by Barbara Minto via Michael Dearing** > “The best executive communication starts with ‘Situation.’ The nickname for this is SCQA, or the Pyramid Principle. > > The best executive communication starts with the state of affairs. It’s fact-based, unambiguous. It’s totally not controversial. No matter what side of an issue or a hard choice you’re on, you should be able to read the situation and go, ‘That pretty much sums it up.’ Whether you’re on one side or another of an issue, it doesn’t matter. You should read the situation and agree to it. Agree that’s fair. > > The next thing that comes out is the complication. A crisp, short statement about what has changed or what’s making things harder. What’s changed, what’s making things harder? The question, the ‘Q,’ falls automatically out of ‘S’ and ‘C’ and it’s almost always ‘What should we do?’ You can save yourself a lot of anxiety if you’re trying to practice this. If you get hung up on ‘Q,’ it’s almost always ‘What should we do?’ > > ‘Answer,’ ‘A,’ answer first at the top of a pyramid. Pyramid-shape your evidence underneath it, and it has to resolve the complication 100%. […] Every conversation in [your] life is an opportunity to SCQA. Friends, loved ones, pets, conversations at the deli counter, everything.” ![Image from Essential reading for product builders—part 2](https://substack-post-media.s3.amazonaws.com/public/images/b4b4289c-bfc8-4853-a95a-f0d7a41aa076_676x398.png) ### **4. [Distribution](https://a16z.com/distribution/), by Ben Horowitz** > “When I ask new entrepreneurs what their distribution model will be, I often get answers like: ‘I don’t want to hire any of those Rolex-wearing, BMW-driving, overly aggressive enterprise sales slimeballs, so we are going to distribute our product like Dropbox did.’ In addition to taking stereotyping to a whole new level, this answer demonstrates a deep misunderstanding of how sales channels should be designed. > > What is a sales channel? It’s a route to market for a product or set of products. It can range from your website to a sophisticated sales force. The sale itself must be supported with the right marketing, process, and optimization strategy. Selecting the right channel is critical for any business—and products often fail because the company chose the wrong route to market.” ### **5. [The Market Curve](https://medium.com/sequoia-capital/the-market-curve-44097b626f6d), by Mike Vernal** > “The market you choose to serve is one of the most important factors for an early-stage startup. And for most technologists, it’s a blind spot. To help, we wanted to share one way of thinking about market. […] > > The simplest and most important way to think about market size is (a) how many potential customers are there and (b) what might each customer be worth to you. […] > > Thinking about the number of customers and the revenue per customer is a tremendously clarifying way to think about a single company.” ![Image from Essential reading for product builders—part 2](https://substack-post-media.s3.amazonaws.com/public/images/39b591b0-8d1d-4cad-b3bf-0b2ef74fc513_1210x582.png) ### **6. [The Four Fits](https://brianbalfour.com/four-fits-growth-framework), by Brian Balfour** > “How do you grow? The ‘go-to’ answer for almost every question in startups is ‘build a great product’ or you need ‘product market fit.’ Every time I hear that answer, it feels completely [unsatisfying]. Building a great product is a piece of the puzzle, but it’s far from the full picture. There are terrible products that have reached $1B+ and amazing products that never make it anywhere. Why is that?” ![Image from Essential reading for product builders—part 2](https://substack-post-media.s3.amazonaws.com/public/images/fa9fae11-cf31-482b-9eae-8b7e8beddb68_1782x1230.png) ### **7. [Startup = Growth](https://paulgraham.com/growth.html), by Paul Graham** > “A startup is a company designed to grow fast. Being newly founded does not in itself make a company a startup. Nor is it necessary for a startup to work on technology, or take venture funding, or have some sort of ‘exit.’ The only essential thing is growth. Everything else we associate with startups follows from growth.” ### **8. [Give Away Your Legos](https://review.firstround.com/give-away-your-legos-and-other-commandments-for-scaling-startups/), by Molly Graham** > “The best metaphor I have for scaling is building one of those huge, complex towers out of Legos. At first, everyone’s excited. Scaling a team is a privilege. Being inside a company that’s a rocket ship is really cool. There are so many Legos! You could build anything. At the beginning, as you start to scale, everyone has so many Legos to choose from—they’re doing 10 jobs—and they’re all part of building something important. > > You have so many choices and things to build during this early phase that it’s easy to get overwhelmed. There’s too much work—too many Legos. You’re not sure you can do it all yourself. Soon, you decide you need help. So you start to add people. That’s when something funny happens on a personal level and to teams: People get nervous. […] > > Almost everything about scaling is counterintuitive. And one of the foremost examples is that reacting to the emotions you’re having as your team adds more people is usually a bad idea**.** Everyone’s first instinct is to grab back the Legos that the new kid took—to fight them for that part of the tower or to micromanage the way they’re building the tower. But the best way to manage scaling (and one of the secrets to succeeding in a rapidly growing company) is to ignore those instincts, and go find a bigger and better Lego tower to build. Chances are if you pick your head up and look around, there’s a brand-new exciting pile of Legos sitting right next to you.” ### **9. [We Don’t Sell Saddles Here](https://medium.com/@stewart/we-dont-sell-saddles-here-4c59524d650d), by Stewart Butterfield** > “Consider the hypothetical Acme Saddle Company. They could just sell saddles, and if so, they’d probably be selling on the basis of things like the quality of the leather they use or the fancy adornments their saddles include; they could be selling on the range of styles and sizes available, or on durability, or on price. > > Or, they could sell *horseback riding*. Being successful at selling horseback riding means they grow the market for their product while giving the perfect context for talking about their saddles. It lets them position themselves as the leader and affords them different kinds of marketing and promotion opportunities (e.g., sponsoring school programs to promote riding to kids, working on land conservation or trail maps). It lets them think big and potentially be big.” ### **10. [Schlep Blindness](https://www.paulgraham.com/schlep.html), by Paul Graham** > “One of the many things we do at Y Combinator is teach hackers about the inevitability of schleps. No, you can’t start a startup by just writing code. I remember going through this realization myself. There was a point in 1995 when I was still trying to convince myself I could start a company by just writing code. But I soon learned from experience that schleps are not merely inevitable, but pretty much what business consists of. A company is defined by the schleps it will undertake. […] > > The most striking example I know of schlep blindness is Stripe, or rather Stripe’s idea. For over a decade, every hacker who’d ever had to process payments online knew how painful the experience was. Thousands of people must have known about this problem. And yet when they started startups, they decided to build recipe sites, or aggregators for local events. Why? Why work on problems few care much about and no one will pay for, when you could fix one of the most important components of the world’s infrastructure? Because schlep blindness prevented people from even considering the idea of fixing payments. > > Probably no one who applied to Y Combinator to work on a recipe site began by asking ‘Should we fix payments, or build a recipe site?’ and chose the recipe site. Though the idea of fixing payments was right there in plain sight, they never saw it, because their unconscious mind shrank from the complications involved. You’d have to make deals with banks. How do you do that? Plus you’re moving money, so you’re going to have to deal with fraud, and people trying to break into your servers. Plus there are probably all sorts of regulations to comply with. It’s a lot more intimidating to start a startup like this than a recipe site. > > That scariness makes ambitious ideas doubly valuable. In addition to their intrinsic value, they’re like undervalued stocks in the sense that there’s less demand for them among founders. If you pick an ambitious idea, you’ll have less competition, because everyone else will have been frightened off by the challenges involved. (This is also true of starting a startup generally.)” ### Honorary mentions: - [Eigenquestions: The Art of Framing Problems](https://coda.io/@shishir/eigenquestions-the-art-of-framing-problems)**,** by Shishir Mehrotra and Matt Hudson - [What Makes a Strong Product Culture?](https://www.bringthedonuts.com/essays/what-makes-a-strong-product-culture/), by Ken Norton - [How to Work with Designers](https://medium.com/the-year-of-the-looking-glass/how-to-work-with-designers-6c975dede146), by Julie Zhuo - [The Next Feature Fallacy](https://andrewchen.com/the-next-feature-fallacy-the-fallacy-that-the-next-new-feature-will-suddenly-make-people-use-your-product/), by Andrew Chen - [Product Management Mental Models for Everyone](https://blackboxofpm.com/product-management-mental-models-for-everyone-31e7828cb50b), by Brandon Chu #### **What’s still missing from this list? Share it in the comments (along with the impact it had on you). Bonus points for stuff people may not have heard of 👇** [Leave a comment](https://www.lennysnewsletter.com/p/essential-reading-for-product-builderspart-1ac/comments) *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [32/46] Why your AI product needs a different development lifecycle *👋 Welcome to a **✨ free edition ✨** of my weekly newsletter. Each week I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lennybot](https://www.lennybot.com/) | [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny)*** *P.S. Annual subscribers get a **free year** of 15+ premium products: **[Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Warp, Granola, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass)** (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** In this AI era, tech leaders need to re-evaluate every single industry best practice for building great products. AI products are just built differently. The teams that realize that and adjust the most quickly will have a huge advantage. Based on their experience leading over 50 AI implementations at companies including OpenAI, Google, Amazon, Databricks, and Kumo, [Aishwarya Reganti](https://www.linkedin.com/in/areganti/) and [Kiriti Badam](https://www.linkedin.com/in/sai-kiriti-badam/) have developed the Continuous Calibration/Continuous Development (CC/CD) framework to specifically address the unique challenges of shipping great AI-powered products. In this post, they’re sharing it for the first time with you. For more from Aish and Kiriti, check out [their popular Maven course](https://bit.ly/aish-lenny) and [their upcoming free lightning talk](https://maven.com/p/88a325/don-t-build-ai-products-like-traditional-software?promoCode=) that explores this topic in depth. *You can also listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads)* ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/0581856c-0cd1-438d-98f6-c31c86857b9f_1456x1190.png) If you’re a product manager or builder shipping AI features or products, you’ve probably felt this: Your company is under pressure to launch something with AI. A promising idea takes shape. The team nails the demo, the early reviews look good, and stakeholders are excited. You push hard to ship it to production. Then things start to break. You’re deep in the weeds, trying to figure out what went wrong. But the issues are tangled and hard to trace, and nothing points to a single fix. Suddenly your entire product approach feels shaky. We’ve seen this play out again and again. Over the past few years, we’ve helped over 50 companies design, ship, and scale AI-powered autonomous systems with thousands of customers. **Across all of these experiences, we’ve seen a common pitfall: people overlook the fact that AI systems fundamentally break the assumptions of traditional software products.** You can’t build AI products like other products, for two reasons: 1. AI products are inherently non-deterministic 2. Every AI product must negotiate a tradeoff between *agency* and *control* When companies don’t recognize these differences, their AI products face ripple effects like unexpected failures and poor decision-making. We’ve seen so many teams experience the painful shift from an impressive demo to a system that can’t scale or sustain. And along the way, user trust in the product quietly erodes. After seeing this pattern play out many times, we developed a new framework for the AI product development lifecycle, based on what we’ve seen in successful deployments. It’s designed to recognize the uniqueness of AI systems and help you build more intentional, stable, and trustworthy products. By the end of this post, you should be able to map your own product to this framework and have a better sense of how to start, where to focus, and how to scale safely. Let’s walk through the ways that building AI products is different from traditional software. ## 1. AI products are inherently non-deterministic Traditional software behaves more or less predictably. Users interact in known ways: clicking buttons, submitting forms, triggering API calls. You write logic that maps those inputs to outcomes. If something breaks, it’s usually a code issue, and you can trace it back. AI systems behave differently. **They introduce non-determinism on both ends: in other words, there’s unpredictability in how users engage and how the system responds.** First, the user interaction surface is far less deterministic. Instead of structured triggers like button clicks, users interact through open-ended prompts, voice commands, or other natural inputs. These are harder to validate, easier to misinterpret, and vary widely in how users express intent. Second, the system’s behavior is inherently non-deterministic. AI models are trained to generate plausible responses based on patterns, not to follow fixed rules. The same request can produce different results depending on phrasing, context, or even a different model. This fundamentally changes how you build and ship. You’re no longer designing for a predictable user flow. You’re designing for *likely* behavior—both from the user and the product—not guaranteed behavior. Your development process needs to account for that uncertainty from the start, continuously calibrating between what you expect and what shows up in the real world. ## 2. Every AI product negotiates a tradeoff between *agency* and *control* There’s another layer that makes AI systems different, and it’s one we rarely had to think about before with traditional software products: **agency**. Agency, in this context, is the AI system’s ability to take actions, make decisions, or carry out tasks on behalf of the user (which is where the term “AI agent” comes from). Think: - Booking a flight - Executing code - Handling a support ticket from start to finish Unlike traditional tools, AI systems are built to act with varying levels of autonomy. But here’s the part people often overlook: **Every time you give an AI system more agency, you give up some control.** So there’s always an **agency-control tradeoff** at play. And that tradeoff matters (a lot!). If your system suggests a response, you can still override it. If it sends the response automatically, you’d better be sure it’s right. The mistake most teams make is jumping to full agency before they’ve tested what happens when the system gets it wrong. If you haven’t tested how the system behaves under high control, you’re not ready to give it high agency. And if you hand over too much agency without the system earning it first, you may lose visibility into the system, and the trust of your users. What’s more, you’re stuck debugging a large, complicated system that has taken actions you can’t trace, for reasons you’ve lost insight into, so you don’t even know what to change. Which brings us to the core framework we’ve developed to help teams navigate these distinctions. We call it **CC/CD:** **Continuous Calibration/Continuous Development**. The name is a reference to Continuous Integration/Continuous Deployment (CI/CD), but, unlike its namesake, it’s meant for systems where behavior is non-deterministic and agency needs to be earned. # **The Continuous Calibration/Continuous Development framework** ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/5eaeeebe-b854-46ca-baa9-0bdc364f19cd_1456x997.png) Just like in traditional software, AI products move through phases toward an end goal. But building AI requires you to account for two things we mentioned earlier: **non-determinism** and the **agency-control tradeoff**. The CC/CD framework is designed to work around these two realities by: - Reducing a system’s non-determinism through design or by monitoring it closely - Ensuring that agency is earned over time, not granted all at once, because every new capability shifts control further away from humans In our framework, product builders work in a continuous loop of **development** (CD) and **calibration** (CC). During development, you scope the problem, design the architecture, and set up evaluations to keep non-determinism in check. You start with features that are low-agency and high-control, then gradually move up as the system proves it can handle more. Then you deploy, not as a finish line but as a transition into the next phase. Once you’ve deployed, you enter the calibration loop, where you observe real behavior, figure out what broke, and make targeted improvements. With every cycle, the system earns a bit more agency. Over time, this loop turns into a flywheel, tightening feedback, building trust, and making the product stronger with each version. ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/6d1bb912-1bb5-4e9c-9787-68f3b47af2a8_1456x1566.png) Let’s go deeper into each step of the CC/CD loop, what it looks like, why it matters, and how to do it well. The first three steps make up the **Continuous Development** side of the loop: scoping the capability, setting up the application, and designing evals. ### CD 1. Scope capability and curate data ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/5af72369-b94f-468d-a743-6864afd82833_1456x547.png) Let’s say you have a big product idea and you’ve already done your research. It’s clear that AI is the right approach. In traditional software development, you’d typically plan for v1, v2, v3 of the new product based on feature depth or user needs. With AI systems, the versioning still applies, but the lens shifts. ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/780d0364-a534-417a-bb8f-a3f811745809_1456x1085.png) Here, each version is defined by how much agency the system has and how much control you’re willing to give up. So instead of thinking in terms of feature sets, you scope capabilities. Start by identifying a set of features that are **high control** and **low agency** (version 1 in the image above). These should be small, testable, and easy to observe. From there, think about how those capabilities can evolve over time by gradually increasing agency, one version at a time. The goal is to break down a lofty end state into early behaviors that you can evaluate, iterate on, and build upward from. For instance, if your end goal is to automate customer support in your company, a high-control way to start would be to scope v1 (version 1) as simply routing tickets to the right department, then move to v2 where the system suggests possible resolutions, and only in v3 allow it to auto-resolve with human fallback. Remember, this is just one approach. What it looks like in practice will depend on your product, but the process tends to be consistent: Start with simple decisions that are easy to verify and easy for humans to override. Then, as you progress through the CC/CD loop, gradually layer in more autonomy with each version. How long you stay in each version depends entirely on how much behavioral signal you’re seeing. You’re optimizing for understanding how your AI behaves under real-world noise and variation. Here are a couple more examples: **Marketing assistant** - **v1:** Draft email, ad, or social copy from prompts - **v2:** Build multi-step campaigns and run them - **v3:** Launch, A/B test, and auto-optimize campaigns across channels **Coding assistant** - **v1:** Suggest inline completions and boilerplate snippets - **v2:** Generate larger blocks (like tests or refactors) for human review - **v3:** Apply scoped changes and open pull requests (PRs) autonomously If you’ve followed how tools like GitHub Copilot or Cursor evolved, this is exactly the playbook they used. Most users only see the current version, but the underlying system climbed that ladder gradually. First completions, then blocks, then PRs, with each step earned through usage, feedback, and iteration. Now, because user behavior is non-deterministic, you’ll need to build a reference for what expected behavior looks like and how your AI system should respond. That’s where data comes in. Data helps break the cold start and gives you something concrete to evaluate against. We call this the **reference dataset**. In the customer support automation example, for the routing version (v1), your reference dataset might include: - The user query - The department it should be routed to - Metadata used to make the decision, such as product type, user tier, or channel You can pull this from past logs if available, or generate examples based on how your product is expected to work. This dataset helps you evaluate system performance and also tells you what context your assistant needs in order to perform reliably. Since most products start cold, aim to gather at least 20 to 100 examples up front. We’ll continue using the customer support example to walk through the next steps in the CC/CD loop. Imagine you’re building toward a fully autonomous support system for a company. Below are the versions we’ll reference, along with their corresponding agency and control levels. We’ll refer to v1, v2, and v3 throughout the rest of the post. ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/0ed2ba7f-15a2-42f6-ab1c-4547a3417c8d_1456x842.png) ### CD 2. Set up application ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/e1327ae9-7773-42dc-a084-9f926587afb2_1456x457.png) Most people skip step 1 and jump into the setup phase too early, getting lost in implementation choices and overthinking which components are needed. But if you’ve scoped your capability properly in step 1, looked at enough examples, and curated a solid **reference dataset**, setting up the application should be fairly straightforward. You already know what the system needs to do, have a sense of what users are likely to throw at it, and understand what a good response looks like. Now it’s just about wiring together the simplest version that gives you a useful signal. There’s a famous saying in software, for a reason: “[Premature optimization is the root of all evil.](https://stackify.com/premature-optimization-evil/)” It applies here too. Don’t overengineer. Don’t over-optimize. Not at this stage. Just don’t. Build only what’s needed for your current version. Make the system measurable and iterable by setting up logs to capture what the system sees from the user, what it returns, and how people interact with it. This will form the basis of your **live interaction dataset** and help you improve the system over time. We won’t go deep into implementation here, but if you’re exposing this to end users, make sure the basics like guardrails and compliance are in place. One more important point: When setting up the application, make sure control can be handed back to humans seamlessly when needed. We’ll refer to these as **control handoffs**. For example, in the customer support v1, if a ticket is misrouted, the receiving agent (the point of contact for that department) should be able to reroute it easily. Since that correction is logged, it not only helps improve the system over time but also preserves the user experience. Thinking about control handoffs from the start is key to building trust and keeping things recoverable. ### CD 3. Design evals ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/68fa43fa-7ace-4e53-812f-8dad7ce52008_1456x550.png) This is the part that usually takes a bit of thought. Before shipping anything, you need to define how you’ll measure whether the system is doing what you expect and whether it’s ready for the next step. You do this using **evaluation metrics** (**evals**, for short). So, what are evals? Evals are scoring mechanisms that help you assess whether your AI system is working, where it’s falling short, and what needs improvement. Think of them as the equivalent of tests in traditional software. But unlike typical unit or integration tests, where inputs and outputs are fixed and correctness is binary, AI evals deal with ambiguity. They’re entirely application-specific and tied to the task you scoped in step 1. You’re not just checking if the system runs—you’re checking how *well* it does something inherently non-deterministic, like summarizing a document or answering a question. That’s why evals aren’t one-size-fits-all. They serve as the signals to guide your iteration loop, helping you tune and refine behavior over time. For example, in the routing v1 of the customer support system, a simple but strong eval would be **routing accuracy**—how often the system routes to the correct department. That alone can tell you if the model is learning the right distinctions and whether or not your setup is solid. ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/2ee0a765-98b6-4309-b992-0b8fb760e3e1_1456x938.png) In v2 of the customer support system, where you’re retrieving internal standard operating procedures (SOPs) or past resolutions to assist an agent, your evals shift to **retrieval quality**. Are the suggestions actually relevant to the ticket? Are they what a human agent would have looked at? One of the best practices at this stage is to run evals against the reference dataset from step 1. This helps you gauge performance, validate whether your evals are well-designed, and make early tweaks to your product setup from step 2. Some teams wait until after deployment to refine, relying on real user interactions. That approach can work too, depending on the system’s risk profile and how much reference data you can collect up front. You don’t need to over-optimize evals on the reference dataset. The goal is broad coverage across key use cases, not perfection. Production behavior will differ, but a strong eval setup gives you a dependable starting point. To dive deeper into designing and refining evals, a great starting point is [Aman’s piece](https://www.lennysnewsletter.com/p/beyond-vibe-checks-a-pms-complete). — Once you’ve implemented your system and verified that it’s properly scoped and instrumented, it’s time to deploy. Deployment is the transition phase between the **Continuous Development** and **Continuous Calibration** loops. ### Transition: Deploy ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/8bec194d-3eab-4395-bc5c-9ad489963738_1456x459.png) Deployment is the fun part. But you’re not just vibe deploying (for lack of a better term 🙂) and hoping for the best. You’ve set up a pipeline that lets you learn and improve. You’re logging interactions, you’ve built in a way for humans to take back control, and you’ve put evaluation metrics in place to flag when something is off. Now’s your chance to see the system in the wild. Boom. You deploy to a small cohort, and the system starts running. From here, you move into the **Continuous Calibration** phase. This is where real-world behavior starts to show up and where you begin to learn what’s working, what’s breaking, and what needs fixing next. ### CC 4. Run evals ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/ca583531-567d-4f75-8c9d-af3bcb8832f7_1456x544.png) You designed eval metrics in the CD loop. Now that you’ve deployed and have user behavior logs, it’s time to assess how well the system actually performed. Once you’ve collected a meaningful amount of **live interaction data**, you can start running your evaluation. You can run them on a subset of the data or across the full dataset, depending on cost and compute constraints. If you need to evaluate on a subset, use the unique properties of your system to decide which data points to sample. For example, in the v1 for the customer support system, you might use logs showing whether a human agent rerouted the query to a different department as a proxy for routing accuracy. In more complex systems, you could look at things like the number of conversation turns, whether users gave thumbs-up or thumbs-down feedback, or other in-session signals. **Control handoff logs** can also provide valuable eval signals, especially when they capture how a human would have stepped in or adjusted the outcome. **Example evaluation for customer support system v1** ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/6d40db6a-4367-40d8-8537-6656e123891c_1456x972.png) Depending on your use case, choose the most representative sample of **live user interactions** to run your eval metrics on. And if your interaction data is small enough (2,000 to 3,000 logs), go ahead and run them on the full dataset. ### CC 5. Analyze behavior and spot error patterns ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/93a4f781-cda3-4f0a-b744-3ae7d2f49d79_1456x459.png) When you look at your evals, maybe they look solid. Maybe they don’t. If you’ve done steps 1 through 3 of the Continuous Development phase well, your metrics are probably somewhere in the middle, which means there’s something to optimize. Now it’s time to start manually reviewing your data. It’s one of the most underrated but necessary parts of building AI systems. A simple strategy is to begin where your eval metrics are weakest. That’s usually where the most valuable signal is. For example, in the customer support system routing v1, you might: - Pull 20 to 50 low-accuracy tickets per department - Focus more on departments where scores are lagging (maybe Refunds is underperforming, while Billing looks fine) Look at what the user said, what the system did, and what the outcome was. Depending on your application, this might be a single interaction or a multi-turn session. Your eval metrics should help you identify where things went wrong and guide you to the point of failure. Once you’ve manually reviewed enough examples, you’ll start to notice repeat error patterns. This is where you begin documenting them. A simple table format works well at this stage: ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/32bbe46b-44f1-46fe-a452-a08a83e0663b_1456x914.png) These patterns show you what needs to change in your system’s logic, prompts, or inputs. They also help you scope the next version more intentionally. ### CC 6. Apply fixes ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/fd16aa7b-05f6-4def-9ba6-d8d606c40f23_1456x550.png) Once you have actionable error patterns, you can start outlining ways to fix them. This could be anything from a simple prompt tweak to switching to a better model, improving retrieval quality, or adding new components to break down the task. What you change depends entirely on what broke. Remember how in step 2 of the Continuous Development phase we said not to overengineer? That’s because this step is when you should engineer more. Evolve the architecture, but do it intentionally—backed by data, not guesswork. This step is often iterative in itself. You apply a fix, run your eval metrics again, and either keep refining the current version or cycle through steps 2 to 5 until the system performs well enough. And since you already have data, you don’t need to redeploy each time. Also, it’s not uncommon for the eval design itself to fall short. That’s because evals are often designed using the reference dataset, which is based on what you *expect* users to do. Real user behavior, however, can be very different. That non-determinism can throw off your evals too. As you go through steps 4 and 5 again, you may find consistent places where the evals missed issues or gave high scores to flawed outputs. So step 6 may also involve revisiting and rebuilding your evals—and that’s completely normal. You’ll likely end up running multiple rounds of evaluation as you continue to shape the product. Sigh . . . it’s all part of the process. — Once you get to the end of step 6 for the first time, you’re probably getting the hang of it. You’ll go through this loop multiple times, gradually reducing control, allowing the system to become more autonomous, and letting product features guide your design choices. Remember that deployment is just one part of the bigger picture. Most of the work is in designing things well. Which brings us to a small rant: Too many people focus almost entirely on implementation, chasing the latest tools and frameworks, but end up making costly mistakes. We started with a large goal, broke it down, and used more complex solutions only when they actually solved a real problem. Never lead with the tech. Let the problem, evals, and data guide what gets added next. That’s how you keep non-determinism in check with AI products. ## Putting it all together: A CC/CD loop for customer support Here’s a potential table that breaks down the problem into multiple versions, each adding more agency over time. It also outlines the flywheel you could build at each stage, using the customer support example we’ve been discussing. Each iteration sets up the next. This is just one way to approach it. Use this as inspiration to think about how you could break down your own product to build and scale intentionally. ![Image from Why your AI product needs a different development lifecycle](https://substack-post-media.s3.amazonaws.com/public/images/0c9d1a03-5988-4ec0-afd7-3bb181d29bcf_1457x1300.png) If you shipped a fully autonomous support system (v3) right away, a lot could go wrong fast. Take one simple example: A refund request gets incorrectly tagged as a billing issue, the system pulls the wrong SOP and generates a plausible but incorrect resolution, and the user ends up confused and loses trust in the product. And while you may have evals in place to flag that something went wrong, you’re now stuck untangling a chain of failures. The final mistake ended as a generation error, but it started at routing, was compounded by missing context, and led to a poor outcome. This is just one simple case, but you can already see how quickly things can get complicated. The CC/CD approach helps you avoid that spiral. In the customer support **v1**, the system handles only ticket routing, giving you signals on how users phrase issues, which departments are often confused, and what metadata actually matters. You then use that to improve routing logic and refine prompts before moving on. In **v2**, the system drafts responses based on SOPs, but a human still reviews them. This helps you understand where retrieval breaks down and which documents need updates. By **v3**, the system is ready to take on more agency by resolving scoped tickets on its own. But now you know which queries are safe to automate and where fallback is needed. ## Where to go from here AI systems have the incredible potential to operate with some level of agency, but getting there is rarely about stacking complex tools or scaling things with brute force. It’s not about writing the perfect prompt either. Creating an AI system that saves time, money, and energy through effective automation is about solving the actual problem, step by step, by understanding its nuances. We often compare working with AI to onboarding a new teammate. The teammate might be brilliant, but they don’t yet know how your team works. You don’t hand them your highest-stakes projects on day one. You start small, observe, build trust, and, as they show what they can handle, you gradually expand their scope. AI systems need that same path. That’s what the CC/CD framework is designed to support. At the heart of CC/CD is judgment: the kind that helps you decide what to ship, how to protect users when things go wrong, when to hand control back, and how to define what “good enough” looks like. There’s no one-size-fits-all for how much capability to give each version, how long to gather data before moving forward, or how tightly to scope. It depends on your product, your users, and your timelines. Some products need weeks of iteration. Others can move faster. That’s where your judgment comes in. Most of that valuable product thinking already lives in successful product leaders. CC/CD simply gives it structure. The framework offers a loop, a rhythm, and a shared language to apply that already-great product instinct to AI systems. *Thanks, Aish and Kiriti! You can [follow](https://www.linkedin.com/in/areganti/) [them](https://www.linkedin.com/in/sai-kiriti-badam/) on LinkedIn, and check out [their popular Maven course](https://bit.ly/aish-lenny) and [their upcoming free lightning talk](https://maven.com/p/88a325/don-t-build-ai-products-like-traditional-software?promoCode=) that explores this topic in depth.* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [33/46] Taking the week off + a newsletter cadence update I’m taking this week off to recharge ✨ My family from L.A. has been in town to spend time with our son (so special), and we’re heading up north this week for our first (and only) little summer vacation. ![Image from Taking the week off + a newsletter cadence update](https://substack-post-media.s3.amazonaws.com/public/images/94e349b8-aab9-4be3-af84-8d853039d7fc_5712x4284.jpeg) I also want to take this moment to announce that when I return, I’ll be experimenting with some tweaks to my newsletter and podcast cadence. Here are the details: - **What’s changing?** The newsletter will shift from 3 paid posts + 1 free post monthly to 2-4 paid posts per month and no free posts. The podcast (still free) will be shifting from 1-2 episodes weekly to 1 episode weekly starting in 2026. - **Why the change?** The most common feedback I get from readers is that there’s too much content coming your way. I feel that. It’s also hard for me to sustain the current pace forever—and I really want to be doing this forever. Cutting back on a few posts per year will allow me to create deeper, higher-value content while maintaining a pace I can sustain long-term. And with everything that now comes with your newsletter subscription (a thriving Slack community, 6+ years of evergreen content, hand-crafted podcast takeaways, 15+ free amazing products worth over $10k—and more coming soon), the value you get from it will only increase. If you have any questions, just reply to this email. Thanks as always for your support 🙏❤️ --- ## [34/46] How to find the perfect name *👋 Welcome to Lenny’s Newsletter. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny)*** *P.S. Annual subscribers get a **free year** of 15+ premium products: **[Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass)** (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** Vercel, Azure, Pentium, PowerBook, Sonos, Swiffer, BlackBerry, Impossible Foods—if you recognize these names, you know [David Placek](https://www.linkedin.com/in/david-placek-05a82/)’s work. Over 40 years, David and his team at [Lexicon Branding](https://www.lexiconbranding.com/) have named nearly 4,000 brands and companies, pioneering the field of brand naming. Building on [his popular recent podcast appearance](https://www.youtube.com/watch?v=adyIaTopO6g), David breaks down his tried-and-tested step-by-step naming process, including a four-part framework you can use with your team to find the perfect name for any product or company. *You can also listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from How to find the perfect name](https://substack-post-media.s3.amazonaws.com/public/images/d4fd77f7-22bf-43b8-832b-384cbbdb6714_1456x970.png) If you’re launching a new product or company, the name you choose may be the single most important decision you make. A great name creates a foundation of trust with consumers, gives meaning and voice to a new idea, and builds cumulative advantage over time. Names that are too descriptive and too comfortable won’t generate the interest or memorability that a new brand needs to succeed. In my more than 40 years leading Lexicon Branding, we’ve created names like Swiffer, Sonos, BlackBerry, Azure, Impossible Foods, Vercel, Windsurf, and many more. In that time, I’ve seen too many teams treat naming as a low priority with marginal results. It’s one of the highest-stakes parts of building a brand, yet it’s too often delegated to junior staff or tackled in group brainstorming sessions. The difference between a good-enough name and the right name is the difference between solid performance and breakthrough success. The right name is a competitive advantage that no one can take away from you. Nobody knew Codeium until it became **Windsurf**. People didn’t care about computer processors until Intel introduced **Pentium**. No one loved mopping until Procter & Gamble created the **Swiffer**. Nobody wanted to eat a Maraxi. But an **Impossible Burger**? Even meat lovers flocked to it. ### **Why names must work harder than ever** AI and technologies like VR are reshaping how people discover, evaluate, and experience brands. We’ll see more innovation and new brands in the next five years than in the previous 20. This creates six key challenges: 1. **Unprecedented competition:** More players, more noise, more choice. 2. **Trademark clutter:** Congestion across all 45 trademark classes. In the U.S., there are over 5,500,000 registered trademarks. 3. **Attention scarcity:** Audiences are distracted and overwhelmed. 4. **Global digital commerce:** Names must work across languages, cultures, and platforms simultaneously. 5. **Institutional skepticism:** People have learned not to believe hype. 6. **Information overload:** A signal must be stronger to stand out. For startups, the challenge is even greater. Your name must attract both investors and early adopters. ### **The three pillars of effective naming** To address these challenges, focusing on three dimensions will increase your likelihood of creating a truly effective name. While we have been using these three pillars for many years at Lexicon, the first pillar, “Build for trust,” has increased in importance. The marketplace is moving so fast now that a new brand must gain momentum early or it will fail. Trust is a key factor in momentum. #### **1. Build for trust.** The right name inspires both trust and imagination. Consider Microsoft’s Azure cloud services brand. The team at Microsoft considered calling it “Microsoft Cloud Services,” but there’s no news in such a straightforward name. When we presented this name, one senior Microsoft executive called it “a dumb idea.” Turns out, it wasn’t so dumb after all. It offered familiarity as a word while generating significant interest based on customer interviews. The sounds in Azure provide additional benefits: the “z” sound creates a strong signal, while the “zure” (like “sure”) creates a smooth experience. #### **2. Communicate an original idea. Never describe.** Anyone can describe what a company does. For example, Infoseek is descriptive and one-dimensional. A name like that is boring and likely to have a thousand similar names swirling around in the competitive landscape. On the other hand, Google—an invented, surprising word—signals a new idea. In our research, we know we have a winner when customers tell us, “They’re not like the other guys.” TripActions is another example of a very descriptive name. As the company grew and developed an expense platform as well, TripActions no longer worked even descriptively. We created Navan, a palindrome that links to travel (“navigation”) but offers the originality that the company wanted to help them move into financial services—and the flexibility to grow into new industries. Since the rebrand in 2023, Navan revenues have grown from $150 million to over $500 million in 2025. #### **3. Be accessible.** The human brain doesn’t like complexity—or remember it very well. Make sure your name is easy to process and contains familiar components. We used the simple and familiar geographic term “Outback” to create Subaru’s most successful brand to date. ### **Case study: Intel’s Pentium** The story of Intel’s Pentium is a perfect example of how to structure a naming process to get an effective result. Intel needed to differentiate its next-generation processor. Up to that point, chip manufacturers only used model numbers. While Intel’s 486 processor outperformed AMD’s 486, consumers perceived the two as identical. Can you blame them? Our first criterion was to develop a name that would differentiate Intel’s processor from all others. Based on interviews with Intel executives, we agreed that the new name needed to capitalize on the success of the “Intel Inside” campaign. “Prochip” was an internal favorite at the time. But our research gave us clear creative direction: **develop a name that sounds and looks like an element inside the computer that is fundamental to performance.** “Pentium” looked and sounded like familiar, science-based words such as “sodium” and “uranium,” using the “-ium” suffix for fluency. But while it was a familiar-seeming word, it was unexpected for the category—a strategic advantage that made it stick out and stick in people’s memory. Here’s a key insight: Invented names like Pentium actually take less money to build into brands than existing words—despite a common myth to the contrary. Invented names signal “new and innovative” better than most other name types. Further, because they are unexpected rather than descriptive or suggestive names, they generate more attention. Humans are more interested in the new. Research conducted by Lexicon in six markets provided data showing that “Pentium” delivered the magic Intel needed. Within months, consumers were asking for Pentium computers—to the chagrin of Dell, HP, and others. ### **The Lexicon process: beyond brainstorming** Creating the right name requires more than one or two brainstorming sessions. Those brainstorming sessions—especially with four or more participants—can get bogged down by everyone’s need to be right, and the peer pressure to agree on a popular solution. As a result, they usually deliver marginal results. When we work with clients, we stick to key process principles: **Set high standards, not narrow objectives.** Aim for fresh, energetic ideas, not just basic objectives like “support the product’s positioning” or “keep it under seven letters.” We use a framework called the Diamond Framework to force clarity on business, marketing, and naming strategy. At the start of any project, we ask: - What does winning look like? - What do we have to win? - What do we need to win? - What do we need to say? ![Image from How to find the perfect name](https://substack-post-media.s3.amazonaws.com/public/images/c6ada04e-250c-47d8-8987-b2e0ef7612d7_1456x1070.png) This creates alignment on the role of the name versus other brand elements like design and messaging. Here are the results of the exercise we did while working with P&G to develop the Swiffer brand name. ![Image from How to find the perfect name](https://substack-post-media.s3.amazonaws.com/public/images/d1603058-087c-41b2-9e6f-f6c6a908adaf_1456x1070.png) **Quantity breeds quality.** Developing a brand name is a treasure hunt. You need to explore, and then explore again. And you need to explore in places where you think there may be no chance of finding the name. Naming is hard work. We recommend generating at least 1,000 creative ideas before selecting a shortlist of 50 to 100 candidates. (We usually generate thousands.) **Don’t underestimate the power of sound.** Imagine the sound of your brand before you start naming. Should your name sound fast? Full? Reliable? Coca-Cola’s Dasani water brand exemplifies this. An invented name with the Latin word for health (“san”) in the middle, a “da” sound that delivers a crisp, clean experience, and an “i” sound that feels light and slim at the end. ### **Case study: Swiffer and the invention of a new category** P&G had invented what they called a new mop, but it wasn’t really a mop. We started by interviewing 25 people about cleaning routines. Some enjoyed washing windows, others polishing furniture. No one liked mopping. One participant said, “It’s just moving dirt and grease from one place to another.” Another said, “Mopping is just not fun.” We reframed the problem around fun, eliminating names like “Promop” or “EffiClean.” One small team explored toys, other exotic vacations, and other easy-to-use tools. Our linguists focused on the actions and sounds of cleaning—sweeping, wiping, washing. When we merged the work, “Swiffer” sat in the middle. The best names don’t describe reality; they change it. Swiffer made mopping feel like a quick, joyful action. In research with our linguistic network and consumers, Swiffer was selected over 80% of the time as the new brand they would be most interested to try. ### **Case study: The story of Vercel** When we started working with a company called Zeit, the team and technology were impressive. Using our Diamond Framework, we defined winning as being the most innovative and “fastest-moving company in the category.” Based on that, we agreed that a coined, inventive name was the right direction. We built matrices of naming components, including prefixes and suffixes that communicated accuracy, speed, and flow; word parts that spoke to the company’s talent and recruiting; and a set of Greek, Latin, and Sanskrit words that meant reliability and power. And using our proprietary sound-symbolism software, we identified fast-sounding letters. Drawing from these components, we combined 55 prefixes and 102 suffixes and got about 60 interesting ideas—not solutions; not yet. We repeated the steps three times over two weeks. “Vercel” rose to the top with a sound that is alive and daring, six letters, familiar parts assembled in a unique way. ### **What not to do** Here are a few things that we have learned not to do over four decades of naming: - **Don’t think you’ll “know it when you see it.”** You won’t. - **Don’t listen to anyone who dismisses bold ideas as “off-brand.”** Bold often becomes the brand. - **Don’t rely on large brainstorming sessions** (they rarely work) or distant freelancers. Use small teams with real esprit de corps and grit. - **Don’t play it safe.** Most approaches favor comfort over competitive edge, producing forgettable names that disappear. There is no power in comfort. ### **The bottom line** The right name doesn’t just happen. It comes from a disciplined, creative process that pushes beyond comfort and challenges assumptions. It requires understanding your competitive landscape, your audience’s psychology, and the power of sound and meaning working together. Do that work, and your name won’t just label what you built—it will change how people experience it. In today’s noisy, AI-driven landscape, that difference will make or break your success. *Thanks, David!* *For more, check out [Lexicon Branding](https://www.lexiconbranding.com/). And Lexicon is hiring! Reach out to [careers@lexiconbranding.com](mailto:careers@lexiconbranding.com).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [35/46] Building eval systems that improve your AI product *👋 Each week, I tackle reader questions about building product, driving growth, and accelerating your career. Annual subscribers get a free year of 15+ premium products: **[Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass)** (while supplies last).* *For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny)*** [Hamel Husain](https://www.linkedin.com/in/hamelhusain/) and [Shreya Shankar](https://www.linkedin.com/in/shrshnk/)’s online course, [AI Evals for Engineers & PMs](https://bit.ly/4myp27m), is the #1 highest-grossing course on Maven, and consistently brings in sizable student groups from all of the major AI labs. This is because they teach something crucial: how to build evaluations that actually *improve* your product, not just generate vanity dashboards. Over the past two years, Hamel and Shreya have played a major role in shifting evals from being an obscure, confusing subject to one of the most necessary skills for AI product builders. After training more than 2,000 PMs and engineers, and leaders at over 500 companies, they’re now sharing their complete playbook—the same methodology taught at OpenAI, Anthropic, and other leading labs. You’ll learn how to leverage error analysis to understand where your AI product breaks, build robust evals you can trust, and create a continuous improvement flywheel that catches regressions before they ship. In honor of this post, Hamel and Shreya are also offering an exclusive discount on their course: 35% off. This is the largest discount they’ve ever offered. [Use this link](https://bit.ly/4myp27m) to register (or use code LENNYSLIST when checking out). You’ve got three days left to enroll. Thank you for sharing this gold with us, Hamel and Shreya 🙏 *P.S. You can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/9aada4fd-935d-4a7b-a704-241eaa5d3618_1457x1130.png) Aman Khan’s [post](https://www.lennysnewsletter.com/p/beyond-vibe-checks-a-pms-complete) on evals perfectly captured why evaluation is becoming a core, make-or-break skill for product managers. This article provides the next step: a playbook for building an evaluation system to drive real product improvements. Many teams build eval dashboards that look useful but are ultimately ignored and don’t lead to better products, because the metrics these evals report are disconnected from real user problems. This guide provides a process to bridge that trust gap. We will cover three phases: discovering what to measure through rigorous error analysis, building a reliable evaluation suite, and operationalizing that suite to create a flywheel of continuous improvement. ## Phase 1: Ground your evals in reality, with error analysis Before you can improve your AI product, you must understand how it fails. The surface area for what you *could* evaluate is infinite. The most common mistake is to start by measuring [ready-made](https://hamel.dev/blog/posts/evals-faq/should-i-use-ready-to-use-evaluation-metrics.html), fashionable metrics like “hallucination” or “toxicity.” This approach often leads to tracking scores that don’t correlate with the actual problems your users face with your product. You cannot know what to measure until you systematically find out how your product fails in specific contexts. The process that tells you where to focus is referred to as “error analysis” and should result in a clean and prioritized list of your product’s most common failure modes. The process begins not with metrics but with data and a single human expert. For most small to medium-size companies, the most effective approach is to designate a single principal domain expert as the arbiter of quality. This person—a psychologist for a mental health chatbot, a lawyer for legal document analysis—becomes the definitive voice on quality. Appointing a single expert, sometimes called a “benevolent dictator,” provides a consistent and deeply informed signal, which eliminates annotation conflicts and prevents the paralysis that can come from having too many cooks in the kitchen. In many situations, the product manager is the principal domain expert. Larger organizations, or products that span multiple complex domains with different cultural contexts, may require multiple annotators. In those cases, you must implement a more structured process to ensure that judgments are consistent, which involves measuring their agreement. Your next step is to arm this expert with a representative set of around 100 user interactions. As you get more sophisticated, you can sample interactions that are more likely to yield insights based on data analysis. Examples include traces that have negative user feedback, outliers in conversation length, number of tools, and high latency. However, start with random sampling to develop your intuition in the beginning. With a dataset ready, the analysis begins with **open coding**. This is essentially like journaling, but with a bit of structuring. The domain expert reviews each user interaction and writes a free-form critique on anything that seems wrong or undesirable, as well as giving a pass/fail judgment on the AI performance. - For **passes**, we explain why the AI succeeded in meeting the user’s primary need, even if there were critical aspects that could be improved. We highlight these areas for enhancement while justifying the overall passing judgment. - For **fails**, we identify the critical elements that led to the failure, explaining why the AI did not meet the user’s main objective or compromised important factors, like user experience or security. Here is a screenshot of open coding in action in response to an apartment leasing assistant. In the interface, we can see that the AI has hallucinated a virtual tour when that isn’t something that is offered: ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/eff70365-e3fb-43be-9889-f51973469875_1600x1402.png) **As a heuristic, the critique should be detailed enough** for a brand-new employee at your company to understand it. Or, if this is more helpful, **so that you can use it in a few-shot prompt for an LLM judge**.Being too terse is a common mistake. Here are some good examples of open coding in action: ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/58c0d68c-8896-40fa-b8d9-0888f9209c50_1456x1610.png) Note that the example user interactions with the AI are simplified for brevity—but you might need to give the domain expert more context to make a judgment. More on that later. This lightly constrained process is crucial for discovering problems you didn’t know you had. It’s also where teams often discover what they *truly* want from their AI system. [Research shows](https://arxiv.org/abs/2404.12272) that people are not very good at specifying their complete requirements for an AI up front. It is through the process of reviewing outputs and articulating what feels “wrong” that the true criteria for success emerge. After collecting notes on dozens of traces, the next step is **axial coding**, or pattern-finding. The expert reads through all the open-ended critiques and starts grouping them (examples below). This process transforms a chaotic list of observations into a clean, prioritized taxonomy of concrete failure modes. It is part art and part science: group errors in a way that is manageable and sensible for your domain. Here is how you might apply axial coding to the failures above: ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/c9ec8f51-5d10-42be-9de3-8dfb117bd43e_1456x954.png) This grouping process often happens in a spreadsheet or a dedicated annotation tool where you can tag or label each critique. When I was working on this apartment leasing assistant in a [real-life scenario](https://hamel.dev/blog/posts/field-guide/), here are the categories that emerged: - Conversation flow issues (missing context, awkward responses) - Handoff failures (not recognizing when to transfer to humans) - Rescheduling problems (struggling with date handling) You can accelerate this process with an LLM. You can use an LLM to perform a first-pass categorization on the critiques. However, a common trap is to over-automate. Always have the human expert review and validate the LLM’s suggestions. An LLM might miss the nuance that distinguishes a conversation flow issue from a handoff failure. Another trap is creating too many categories; aim for a manageable set of under 10 primary failure modes that capture the most significant problems. The goal is to create a useful taxonomy that you can analyze, not an exhaustive list. The final product of this phase is to simply count the categories so you can get a sense of how to invest your time. Here is how this count looks for the apartment leasing assistant, which I calculated with a pivot table in a spreadsheet: ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/d7a37d90-bcf5-406e-a484-86d4ad48d1c9_658x354.png) As you can see, the most frequently occurring errors were conversation flow, handoff (to a human), and rescheduling appointments. This data gives us concrete problems specific to our product to focus on as we build evals. ### A warning about off-the-shelf metrics While off-the-shelf metrics like hallucination and toxicity are not worth paying attention to directly, they can be used in creative ways. Instead of reporting a hallucination or toxicity score on a dashboard, calculate the scores on your traces and sort them by high/low score. Reviewing the highest *and* lowest scoring examples can reveal surprising failure modes or unexpected successes, which in turn helps you build custom evaluators for the patterns you discover. This is one of the only appropriate uses for off-the-shelf metrics. Note that this is an advanced technique and should be done only after you master the basic approach. ## Phase 2: Build out your evaluation suite After error analysis, you will have a prioritized list of your product’s most common failures. The next step is to build a suite of automated evaluators to track them. The goal is to create a system that is reliable, cost-effective, and trusted by your team. This requires choosing the right tool for each failure mode. Your choice of tools comes down to one question for each prioritized failure mode on your list: Is this failure **objective and rule-based** (for example: “Does the output contain a user ID?”), or is it **subjective and requiring judgment** (example: “Was the tone appropriate for the persona?”)? For objective failures, use **code-based evaluators**. These are simple checks written as code, like assertions in a unit test. They are fast, cheap, and deterministic, making them perfect for verifying things like whether an output is valid JSON, contains a required keyword, or executes without error. Use them whenever a clear rule can define success or failure. For subjective failures, you’ll need to build an **LLM-as-a-judge** to reliably assess qualities that code cannot easily handle, like tone, relevance, or reasoning quality. This can be a rigorous process—as is training any LLM—but it is the only way to scale nuanced and subjective evaluations and ultimately improve your product. The good news is that there is a scientific approach to making sure the judge is sufficiently aligned with your product vision and success criteria. ### The LLM-as-a-judge playbook This is not about writing a clever prompt. It is about a systematic process of grounding an LLM’s judgment in your specific quality bar. The output is an LLM that gives you a binary pass/fail metric for specific error(s). More importantly, you need to trust the metric. The way to establish trust is by measuring the judge against a human-labeled dataset you create. There are two steps involved. The first is to create a dataset that establishes ground truth: #### 1. Establish ground truth Your evaluation system is only as good as its source of truth. For most teams, the most effective approach is to leverage the **principal domain expert** we mentioned earlier. While larger organizations operating across multiple domains may require multiple annotators and processes to measure inter-annotator agreement, starting with a single expert accelerates the process. The expert’s task is to provide two things for every user interaction with your AI, grouped by session: a binary **pass/fail** judgment and a detailed **critique**. Many teams are tempted to use a 1-to-5 Likert scale, believing it captures more nuance. This is a trap. The distinction between a “3” and a “4” is subjective and inconsistent. Binary decisions force clarity. An output either meets the quality bar or it does not. The nuance is not lost; it is captured in the critique, which explains *why* a judgment was made. These critiques are the secret ingredient for building a high-fidelity judge. For example, consider this example from earlier: ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/b8baec27-cfcb-4d1b-9358-b7c98c0cadf1_1456x556.png) Reasonable people may disagree on whether or not this is “good enough.” However, it is important that you strive toward making a judgment call on what is good and bad for your product. In this case, we decided that this interaction was a failure. #### 2. Build and validate the judge After you have collected the ground-truth data labeled by your domain expert, you are ready to build and validate the judge. Do not use your entire dataset to build and test your judge. This leads to overfitting, where you iterate toward performing well on examples you observe but fail on new, unseen data. Instead, split the ground-truth data into three distinct sets: - **Train set (10%–20%):** A small set of clear examples, including the expert’s critiques, to use in the judge’s prompt - **Dev set (40%–45%):** A larger set used to iteratively test and refine the judge’s prompt - **Test set (40%–45%):** A held-out set, untouched during development, for a final, unbiased measurement of the judge’s performance This process of refining your judge’s prompt on the dev set is a meta-evaluation task. You are evaluating your evaluator. It is also where you will discover the nuances of your own quality bar. As [research](https://arxiv.org/abs/2404.12272) on “criteria drift” has shown, the process of reviewing LLM outputs and aligning a judge helps you articulate and refine your own standards. Below is a visualization of this LLM-as-a-judge alignment process at a high level. ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/0aae88ca-fa18-437b-aaec-a4ea61f2ff28_1456x943.png) #### 3. Measure what matters: TPR/TNR over accuracy A common impulse is to measure a judge’s performance with a single accuracy score, but this can be dangerously misleading. Imagine an AI system that succeeds 99% of the time. A judge that always predicts “pass” will be 99% accurate, but it will never catch a single failure. This is a common issue with **imbalanced datasets**, where one outcome is far more frequent than the other. Instead of accuracy, the **true positive rate (TPR)** and **true negative rate (TNR)** measuredtogetherwill tell you precisely how your judge is likely to make mistakes. In plain language: - **TPR:** Of all the examples that *should* pass, what percentage did the judge correctly label as “pass”? - **TNR:** Of all the examples that *should* fail, what percentage did the judge correctly label as “fail”? A judge with a high TPR but low TNR is good at recognizing success but lets failures slip through. The acceptable tradeoff depends on your product. For an AI providing medical advice, a false negative (failing to catch a harmful suggestion) is far more costly than a false positive. For a creative writing assistant, a false positive (flagging a good response as bad) might be worse, as it could stifle creativity. Once you know your judge’s TPR and TNR, you can even statistically correct its raw scores to get a more accurate estimate of your system’s true failure rate. For example, if your judge reports a 95% pass rate on 1,000 new examples but you know it has a 10% chance of mislabeling a failure as a pass, you can adjust that 95% score to reflect the judge’s known error rate. (The mathematical details for this correction will be in an appendix). This rigorous, human-led validation process is the only way to build an evaluation system your team can rely on. When you present a dashboard showing a 5% failure rate for a critical feature, your stakeholders need to believe that number reflects reality. This process is how you build that trust. #### **Considerations for specific architectures** There are special considerations and strategies to keep in mind when designing evals for multi-turn conversations, RAG pipelines, and agentic workflows. We address each of them below. **Multi-turn conversations** Many AI products are conversational, which introduces the challenge of maintaining context over time. When evaluating conversations, start at the highest level: Did the entire session achieve the user’s goal? This session-level pass/fail judgment is the most important measure of success. When a conversation fails, the next step is to isolate the root cause. A common mistake is to assume the failure is due to the complexities of dialogue. Before you dive into multi-turn analysis, try to reproduce the failure in a single turn. For example, if a shopping bot gives the wrong return policy on the fourth turn, first ask it directly: “What is the return policy for product X1000?” If it still fails, the problem is likely a simple knowledge or retrieval issue. If it succeeds, you have confirmed the failure is conversational—the bot is losing context or misinterpreting information from earlier in the dialogue. This diagnostic step saves significant time by distinguishing simple knowledge gaps from true conversational memory failures. **Retrieval-augmented generation (RAG)** A RAG system is a two-part machine: a retriever finds information, and a generator writes an answer using that information. These two parts can fail independently, and an end-to-end correctness score will not tell you which one is broken. You must evaluate them separately. First, evaluate the **retriever**. Treat it as a search problem. To do this, you need a dataset of queries paired with their known correct documents. The most critical metric for RAG is often **recall@k**. This measures what percentage of all the truly relevant documents are captured within the top *k* results your system retrieves. Recall is paramount because if the correct information is not retrieved, the generator has no chance of producing a correct answer. Modern LLMs are surprisingly adept at ignoring irrelevant noise in their context, but they cannot invent facts from missing information. The value of *k* is a critical tuning parameter that depends on your task. For a simple query requiring a single fact, like “What are the property taxes for 123 Main St.?,” a small *k* (e.g. 3 to 5) is often sufficient. The main goal is to ensure that the one correct document is retrieved. However, for a complex query that requires synthesizing information from multiple sources, such as “Summarize recent market trends for 3-bedroom houses in downtown,” you'll need a larger *k* (e.g. 10 to 20) to provide the generator with enough context to build a comprehensive answer. While recall is the priority for the initial retrieval stage, **precision@k** (the fraction of retrieved documents that are relevant) becomes important in systems with a second, re-ranking stage designed to select the best few documents to pass to the LLM. Once your retriever is performing well on a diverse set of queries, you can evaluate the **generator**. Here you are primarily measuring two things. First is **faithfulness**: does the generated answer stick to the facts provided in the retrieved context, or is it hallucinating? Second is **answer relevance**: does the answer directly address the user’s original question? An answer can be perfectly faithful to the source documents but still fail to be relevant to the user’s intent. Fix your retriever first. Only when you are confident that the right information is consistently being fed to your generator should you focus heavily on improving the generation step. It should be noted that RAG is a very nascent topic, and there is still much to be explored in terms of evaluating and optimizing it. [See this series](https://hamel.dev/notes/llm/rag/not_dead.html) for an exploration of advanced RAG topics. **Agentic workflows** Agents—which can execute a sequence of actions, like tool calls, to accomplish a goal—are the most complex systems to evaluate. A single pass/fail judgment on the final outcome is a good start, but it is not diagnostic. When an agent fails, you need to know *which* step in the chain of reasoning broke. For this, a **transition failure matrix** is an invaluable tool. Think of an agent’s workflow as a series of states or steps, like an assembly line. The agent moves from one state (e.g. *generating\_sql*) to the next (e.g. *executing\_sql*). A transition failure matrix is a chart that shows you exactly where the assembly line breaks down most often. ![Image from Building eval systems that improve your AI product](https://substack-post-media.s3.amazonaws.com/public/images/ce7b525b-3a4b-456d-b8ee-673e3995398e_1457x1003.png) The rows of the matrix represent the last successful step, and the columns represent the step where the failure occurred. By analyzing traces of failed agent interactions and mapping them onto this matrix, you can quickly spot hotspots. Instead of guessing, you can see with data that, for example, your agent fails most often when trying to execute the SQL it just generated, or when it misinterprets the output from a tool call. This transforms the overwhelming task of debugging a complex agent into a focused, data-driven investigation. With these targeted evaluation strategies for complex systems, you are now ready to operationalize your full evaluation suite. ## Phase 3: Operationalizing evals for continuous improvement Having a suite of trusted evaluators is a major milestone, but it is not the end goal. An evaluator is a tool; a system of evaluation is a process. The final phase is to build a process that creates a flywheel, using your evals to drive continuous product improvement. This system has two distinct components: a safety net to prevent backsliding and a discovery engine to find new problems. ### The safety net: Code-based evals in continuous integration (CI) Your first goal is to prevent old bugs from reappearing. The best way to do this is by integrating your evaluators into a CI pipeline. The foundation of CI for AI is a **golden dataset**, a curated collection of examples that represent the most important scenarios for your product. This is not a random sample of production data; it is a purpose-built stress test. It should include examples covering your core features, challenging edge cases you discovered during error analysis, and, most importantly, a regression test for every significant bug you have fixed. On every code or prompt change, your CI pipeline runs the system against this golden dataset and checks the outputs with your fast, deterministic, code-based evaluators. If any check fails, the build breaks, and the regressive change is blocked from going to production. It is crucial to understand what this safety net does and does not guarantee. A passing CI build means you have not reintroduced known failures. It is a signal of *stability*, not overall production quality. ### The discovery engine: LLM-as-a-judge and guardrail evals in production While CI protects you from “known unknowns,” production is where you find the “unknown unknowns.” Your production monitoring system is a discovery engine for new failure modes. The process begins with comprehensive logging. You should log the entire trace of the interaction: the user input, all intermediate steps and tool calls, and the final output. You then run your validated evaluators on a sample of these traces. Since many of your most important evaluators—especially the LLM judges—may be too slow or expensive to run in real time, this is typically done **asynchronously**. The results feed into a monitoring dashboard, where you track your key quality metrics over time. By using the TPR and TNR you calculated for your judges, you can even statistically correct their raw outputs to get a more accurate estimate of your system’s true success rate. For critical, high-impact failures, you can use a special type of synchronous evaluator called a **guardrail**. These run in the request path and can block, redact, or regenerate a response before the user ever sees it. Most guardrails are fast, deterministic checks—like regexes, keyword blocklists, or schema validators—because they must add minimal latency. Crucially, they must have a very low false-positive rate; blocking a valid response is a production bug. While less common, an LLM-as-a-judge *can* be used as a guardrail, but only if the application’s latency budget allows for it. The decision involves a direct tradeoff: in high-stakes domains like medical advice, the cost of a false negative (letting harmful advice through) may justify using a slower, more powerful judge inline; in creative applications, the cost of a false positive (blocking a valid response) might be too high. ### Closing the loop: The improvement flywheel Your CI safety net and your production discovery engine work together to create a powerful improvement loop. When your production monitoring flags a new or drifting failure mode, it triggers a new round of error analysis. For example, you might notice a spike in location ambiguity failures after launching a new feature. A manual review of the flagged traces reveals the model is confusing “West Berkeley” with the (non-existent) “Berkeley West.” This discovery feeds back into the development process in two ways. First, you improve the product itself, perhaps by refining a prompt to better handle directional qualifiers. Second, and just as important, you improve your evaluation artifacts. You add the West Berkeley example to your CI golden dataset. Now any future change that reintroduces this specific confusion will be automatically caught. This flywheel—monitor, analyze, improve, deploy—is the engine of AI product quality. It ensures that every failure, once discovered, makes your system permanently smarter and more robust. ## Effective evaluation systems are a process, not a dashboard Building an evaluation system that actually improves your AI product requires a shift in mindset: away from a focus on tools and generic scores, and toward a rigorous, human-centered process that identifies and addresses issues that actually matter. This process begins with error analysis to find the real problems in your product, not by adopting generic metrics. It relies on a single domain expert and clear, binary labels to establish a consistent quality bar. The process demands the right tool for each job—a cheap, fast code-based check for objective rules and a carefully validated LLM-as-a-judge for subjective nuance. Finally, it operationalizes these evaluators within a CI/CD flywheel, creating a system that can help you spot bugs faster. Prompting is just one small part of building a great AI product. The real competitive advantage comes from building a deep understanding of your system’s failures through a robust evaluation lifecycle. This is what gives your team the confidence to iterate quickly and ship an AI product that consistently delivers value. ## Questions It is impossible to explain everything about evals in a single post. Hamel and Shreya have documented the most frequent questions that people have about evals [here](https://hamel.dev/blog/posts/evals-faq/) ([PDF version here](https://hamel.dev/blog/posts/evals-faq/evals-faq.pdf)). *Thanks,* *[Hamel](https://www.linkedin.com/in/hamelhusain/) and [Shreya](https://www.linkedin.com/in/shrshnk/)! For more, check out [their course](https://bit.ly/4myp27m)—and get 35% off using code LENNYSLIST (the largest discount they’ve ever offered).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [36/46] How to get the most out of your product pass, part 1 *👋 Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny)*** ![Image from How to get the most out of your product pass, part 1](https://substack-post-media.s3.amazonaws.com/public/images/911fe309-9d9c-4be2-a2fb-c5cd855686b6_1456x970.png) After [introducing the product pass](https://www.lennysnewsletter.com/p/productpass) a couple of months ago, the most common question I get—other than “Wait, is this for real??”—is how I and others use these tools. So I’ve put together a short 2-part series about how I use each tool, what I love about it, and what it can help you with. This week I’ll cover [Replit](https://replit.com/), [Warp](https://go.warp.dev/lenny), [Linear](https://linear.app/), [Wispr Flow](https://wisprflow.ai/), [Gamma](https://gamma.app/partners/lenny), [Magic Patterns](https://magicpatterns.link/lpp1), [Descript](https://descript.com/lenny), and [Mobbin](https://mobbin.com/lenny). Next month I’ll cover [Lovable](https://lovable.dev/lenny), [Bolt](https://bolt.new/), [n8n](https://n8n.io/), [Granola](https://www.granola.ai/), [Superhuman](https://superhuman.com/), [Raycast](https://www.raycast.com/), [Perplexity](https://www.perplexity.ai/pro), and [ChatPRD](https://www.chatprd.ai/lenny). My goal here isn’t to sell you the products ([you already get them for free](https://www.lennysnewsletter.com/p/productpass)!) but to help you get the most possible value out of your paid subscription. If you already use any of these products heavily and have a pro tip to offer, please share in the comments! [Leave a comment](https://www.lennysnewsletter.com/p/how-to-get-the-most-out-of-your-product/comments) # 1. **[Replit](https://replit.com/)** #### **What it is** A leading vibe-coding platform that takes a plain English description of what you want to create and builds a production-ready app. It includes a native database, user authentication, security scan, and hosting. It also supports most programming languages. They launched [Agent 3](https://replit.com/agent3), which can run for more than 200 minutes autonomously, building, testing, and fixing your app on the fly. #### **Why I love it** It’s a powerful full-stack vibe-coding platform that lets you build highly complex web and mobile apps. Unlike other vibe-coding platforms, you can single-shot fairly sophisticated applications. It includes a fully featured (web-based) IDE and automatically tests the apps it builds using its internal browser (i.e. checking buttons, forms, APIs) and then auto-fixes the issues it finds. #### **What I use it for** The entire (surprisingly sophisticated) [product pass workflow](https://lennysproductpass.com/) is hosted on Replit. #### **How other people are using it** 1. [Coming up with startup ideas](https://x.com/daniel_mac8/status/1966260595143094534) 2. [Vibe-coding automations](https://x.com/omarsar0/status/1966949907149058551) 3. [Building a social networking platform (in 193 minutes)](https://x.com/anthropiast/status/1966109218479714593) 4. [Building a fully featured website for Saastr](https://x.com/jasonlk/status/1967408276825575753) 5. [Taking on Upwork projects](https://x.com/niconley/status/1966505076849074618) #### **Learn more** Check out their [website](https://replit.com/) and their [PM-specific use cases](https://replit.com/usecases/product-managers). # 2. **[Warp](https://go.warp.dev/lenny)** #### **What it is** An agentic development environment that’s essentially a drop-in replacement for your Mac/Windows/Linux terminal. It was originally built for engineers but is increasingly used by PMs, designers, data scientists, and even recruiting teams. #### **Why I love it** It’s truly a magical experience, and unlike anything else I’ve used. It makes you feel like a superhero. Anytime I run into a problem working in my Mac terminal, AI solves it for me. How did we ever work with computers without something like this? #### **What I use it for** I’ve used it to download YouTube videos to convert them to MP3s (for my son’s [Yoto](https://us.yotoplay.com/)), analyze all of my locally-saved podcast transcripts to find the most-mentioned books, and install gnarly packages with a bunch of weird dependencies. I just tell it what I want to do, and it figures it out. Here’s me asking Warp to download the Linear YouTube video below: ![Image from How to get the most out of your product pass, part 1](https://substack-post-media.s3.amazonaws.com/public/images/6760b117-cc4f-4c98-88ae-82f19e328340_2008x614.png) #### **How other people are using it** 1. [Building a Strava clone](https://youtu.be/cmHheGcDWTQ) 2. [Converting MRI imaging files to jpegs](https://x.com/aaronwhite/status/1958281839333794057?s=46) 3. [Dad teaching kids how to build websites](https://x.com/some_user123/status/1950443062016041377) #### **Learn more** Visit their [website](https://go.warp.dev/lenny) or check out their many tutorials at [warp.dev/university](https://warp.dev/university). # **3. [Linear](https://linear.app/)** #### **What it is** A beautifully designed tool for tracking your team’s tasks, organizing projects, and building your roadmap. It even has deep agent integrations, so you can delegate your tickets to an agent to take a first pass at a task. The future is now. #### **Why I love it** In my mind, there’s no question that every startup and modern product team should be using Linear. And at this point, most are. #### **What I use it for** OK, so I don’t actually use Linear, because it’s just me. But if I had a team, I wouldn’t even consider another tool. #### **How other people are using it** 1. [Linear using Linear to manage their own customer feedback](https://linear.app/now/how-we-think-about-customer-experience-at-linear) 2. [Claire at ChatPRD building an agent in Linear](https://linear.app/integrations/chatprd) 3. [Commure building dashboards to monitor their team’s progress](https://linear.app/now/commure-dashboards) #### **Learn more** Check out their [website](https://linear.app/) and this [library](https://www.youtube.com/playlist?list=PLP9v0Y4zla9vG7k8e279bSz5hUl0oXlMH) of how-to videos, and here’s [my podcast conversation with Nan Yu](https://www.youtube.com/watch?v=nTr21kgCFF4) (their head of product). # **4. [Wispr Flow](https://wisprflow.ai/)** #### **What it is** Magical voice-dictation that actually works the way you’ve always expected this to work. Press the function key on your laptop, talk, and it instantly (and accurately) transcribes everything you’ve said. Works across Mac, iPhone, and Windows, and with every vibe-coding tool and IDE. #### **Why I love it** It’s so fast, and it’s wildly accurate. It understands the context of the conversation, filters out filler words, and learns your unique acronyms, terms, and phrases. If you’ve tried the native mic transcription on your phone, this is a whole different ballgame. #### **What I use it for** I’ve always preferred typing over talking, so I’ve never gotten into transcription apps, but Wispr Flow changed how I work. Since it’s so good, I’ve found myself using it constantly, especially on my [phone](https://apps.apple.com/us/app/wispr-flow-ai-voice-keyboard/id6497229487). For example, my wife is a big texter, and instead of having to type out long messages all day, I press a button, say it, and send it. #### **How other people are using it** 1. [Talking to chatbots](https://x.com/MatthewGattozzi/status/1967952481419403609) 2. [Vibe coding](https://x.com/tankots/status/1891138380404060298) 3. [Getting “voicepilled”](https://www.linkedin.com/posts/reidhoffman_i-am-voicepilled-a-major-step-forward-in-activity-7373713536096780289-lUb7/?rcm=ACoAAAQVbqEB-UADeCu9ymhPGPUQCNHXfB6TcEU) 4. [Doctors transcribing their notes](https://x.com/RutaB_/status/1956099269544567236) 5. [Slow typers using it to speed up their workflows](https://x.com/rywiggs/status/1937991165217017879) #### **Learn more** Check out their [website](https://wisprflow.ai/). # **5. [Gamma](https://gamma.app/partners/lenny)** #### **What it is** An AI-powered document creation tool that generates polished presentations, landing pages, and documents from a simple prompt. Describe what you want, upload the raw data, and get back a compelling deck/doc/website. And just last week, [they launched Gamma 3.0](https://x.com/thisisgrantlee/status/1967943621782413388), which adds a bunch of really cool new powers. #### **Why I love it** If Cursor is making engineers faster, Gamma is doing the same for PMs. Creating decks and docs is one of the most time-consuming and annoying parts of most builders’ jobs. I also just love [the story of Gamma](https://www.nytimes.com/2025/02/20/technology/ai-silicon-valley-start-ups.html)—small team, little funding, profitable, and AI-native from day one. #### **What I use it for** Most recently, I’ve been experimenting with using Gamma to create shareable summaries of my podcast episodes. Here’s [Ethan Smith’s episode on AEO](https://gamma.app/docs/Mastering-Answer-Engine-Optimization-n4gxgak9z9tpzhc), [Ben Horowitz’s episode on leadership](https://gamma.app/docs/10-Hard-Truths-That-Built-Billion-Dollar-Companies-onh55jsg26xrzp5), [Brendan Foody’s story of Mercor](https://ai-revolutions-hidden-bo-3k6n3li.gamma.site/), and [a peek at an upcoming episode on evals](https://the-rise-of-ai-evals-q62i0s7.gamma.site/). I copied and pasted the transcripts, clicked a few buttons, and . . . magic. #### **How other people are using it** 1. [Generating custom decks for each sales outreach](https://www.linkedin.com/posts/yonathan-cohen_gamma-30-just-dropped-i-built-a-29-reply-activity-7373718410439942144-sKqM/) 2. [Creating landing pages](https://www.linkedin.com/posts/charlie-hills_how-to-make-a-landing-page-in-5-minutes-activity-7321473049843621888-BTLe/) 3. [Turning meeting notes into a deck and follow-up email](https://x.com/riyazmd774/status/1967978262015160802) ([here’s how to do this with Zapier](https://zapier.com/templates/details/meeting-notes-to-presentations-granola-gamma)) 4. [Creating a unique resume](https://suchita-kaundin-bsb2sbo.gamma.site/) 5. [Turning PRDs into decks](https://x.com/clairevo/status/1968002491724988923) #### **Learn more** Check out [their website](https://gamma.app/partners/lenny). # **6. [Magic Patterns](https://magicpatterns.link/lpp1)** #### **What it is** Magic Patterns is an AI prototyping tool built specifically for PMs and designers. It’s not trying to get you to build production-ready apps or launch startups. It’s designed instead to be the best tool in the world at prototyping your ideas using your existing products’ styles, getting feedback from customers, and then helping you build a real product in your regular development environment. #### **Why I love it** I’ve been playing with all of the popular prototyping tools, and I find that Magic Patterns most often produces the highest-quality, most useful, and most correct results. It’s especially great at quick iterations and ideation because it only produces frontend code, so you don’t have to worry about random database/infra architecture limitations. I encourage you to do your own side-by-side comparison to see for yourself. I also love that it’s not trying to be everything to everyone—it’s designed specifically to solve pain points for product teams at large companies, with features like [using your own component library](https://magicpatterns.link/llp1co). #### **What I use it for** I’ve used it to create [this thumbnail preview tool](https://project-youtube-podcast-title-and-thumbnail-previewer-348.magicpatterns.app), and other internal tools I’m tinkering with that I’m not ready to show the world 👀 #### **How other people are using it** 1. [Copy-and-pasting PRDs, creating a Magic Patterns prototype, and then](https://magicpatterns.link/llp1x) sharing with engineering 2. [Creating personalized customer demos](https://magicpatterns.link/lpp1lu) 3. [Skipping Figma entirely](https://magicpatterns.link/llp1jh) when prototyping 4. [More examples here](https://www.magicpatterns.com/catalog?dub_id=0vANFES1HwRsFVsd) and [tutorial videos here](https://www.magicpatterns.com/docs/documentation/ai-prototyping-course/introduction) #### **Learn more** Check out their [website](https://magicpatterns.link/lpp1), and build something! # **7. [Descript](https://descript.com/lenny)** #### **What it is** An AI-powered editor that gives everyone (including product managers!) the power to create and edit professional videos. Use it to make launch videos, design walk-throughs, for LinkedIn posts, etc. You edit videos like you edit Google Docs (wild!), and it now includes an agentic video editor, called Underlord, that makes it even easier to make tricky changes ([see example](https://www.linkedin.com/posts/fatemeh-alavizadeh_ever-wonder-how-ai-actually-thinks-about-activity-7350929550232444928-fc6s?rcm=ACoAAABAwBcBhqY8pjVXMxROG-p1CJMLG1fMBWUa)). #### **Why I love it** I have zero video editing skills and I’m constantly making and editing video and audio. I wouldn’t be able to do this without Descript. It’s been my tool of choice for over five years now, and I’ve never thought about using anything else. #### **What I use it for** I use Descript basically every day. I use it to record my podcast ads and intros to each episode, and my production team uses it exclusively to take my raw podcast recordings and turn them into the videos you see on [YouTube](https://www.youtube.com/@LennysPodcast). #### **How other people are using it** 1. [Creating YouTube shorts](https://www.youtube.com/watch?v=8rXZkt6YkZw) 2. [Adding big juicy text to videos](https://youtube.com/shorts/phEBSCJP8xw?feature=shared) 3. [Creating rough cuts with AI](https://www.youtube.com/shorts/RtC7_wQ1Arg) 4. [Creating avatar-based demos](https://www.linkedin.com/posts/burkhauser_guys-now-on-descript-you-can-instantly-create-activity-7320888554761453568-Eex_?rcm=ACoAAABAwBcBhqY8pjVXMxROG-p1CJMLG1fMBWU) 5. [Creating ads](https://www.youtube.com/watch?v=SvBGR5cSeNw) #### **Learn more** Check out their [website](https://descript.com/lenny). # **8. [Mobbin](https://mobbin.com/lenny)** #### **What it is** The world’s largest mobile and web design reference library. Use it to unblock your design thinking, benchmark your product flows against best-in-class apps, and stay up to date on the latest UX trends. #### **Why I love it** I wish this existed when I was a full-time PM. I would have saved my team and myself hundreds of hours of screenshotting competitors’ flows looking for inspiration and a lay of the land. #### **What I use it for** I point people to it when they are looking for ideas for optimizing a flow in their product. #### **How other people are using it** 1. [Getting design inspiration](https://www.instagram.com/reel/C3px5ESAwxA/) 2. [Copy-and-pasting best-in-class flows into your Figma](https://youtu.be/E4idM0hC_AY?si=tJA95mUdzSmya0Wa&t=1537) 3. [Designing a viral app](https://youtu.be/T64X2GUoDwc?si=8f98NKkKPygr9QHS&t=360) 4. [Elevating your design taste](https://youtu.be/8mMH6Pq8qnE?si=xCt-KLIZsOhxrV-v&t=533) 5. [Studying the best](https://x.com/rexan_wong/status/1967789508717650388) #### **Learn more** Check out their [website](https://mobbin.com/lenny). — Part 2 is coming next month. [Subscribe](https://www.lennysnewsletter.com/subscribe) if you haven’t yet to get access to all of these tools! Have a fulfilling and productive week 🙏 **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [37/46] Introducing the GAIN framework for feedback: an evidence-based approach to giving feedback that people love, appreciate, and act on *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Courses](https://maven.com/lenny)*** *Annual subscribers get a **free year** of 15+ premium products: **[Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass)** (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** If feedback is a gift, why is it so hard to receive? And to give. It’s because no one taught us how to give feedback. I was introduced to [Jack Cohen](https://www.linkedin.com/in/jackadamcohen/) by (frequent collaborator) [Tal Raviv](https://www.lennysnewsletter.com/p/product-manager-is-an-unfair-role), who told me, “Jack is the single biggest influence on my personal productivity and ‘inner game.’ ” Over the past 11 years, Jack has coached hundreds of founders, executives, and managers to help them grow themselves and their colleagues as fast as their startups. Growth hinges on getting honest and ongoing feedback. Through thousands of feedback conversations and hundreds of hours of research, Jack’s developed an evidence-based, immediately applicable framework for giving feedback that he calls **GAIN**. As Jack puts it, “GAIN is both a useful acronym and the core of the philosophy.” When you tell people what theystand to gain from the feedback, it suddenly starts feeling like a gift again. This guide is essentially a how-to for Radical Candor, and the most specific (and easy-to-remember) framework for giving great feedback that you’ll find anywhere. It’s going to blow your mind. Enjoy. *For more, Jack teaches a highly rated course, [ManagerGPT: The AI Tools and Human Systems to Scale Yourself and Your Team Fast](https://bit.ly/3VK5fGt), focused on helping high-growth leaders build their individual operating systems and interpersonal skills to stop fighting fires (or each other) and start creating what matters most: great products and great teams. For a taste of the course and a demo of both GAIN feedback and the AI tool Jack built to support it, check out his free 30-minute lightning lesson, “[FeedbackGPT: Give Feedback People Thank You For](https://bit.ly/3VIywRW).”* *You can also listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/7b25c4ee-fe49-4add-99c2-a5ed8d37079f_1456x970.png) Over my past 11 years coaching several hundred people from dozens of organizations and teaching thousands more, there’s one challenge I’ve seen people struggle with across every level, industry, company stage, personality type, and culture: giving feedback. Many people come to coaching and workshops already describing themselves as “conflict-averse,” dreading having difficult conversations and giving direct feedback. A good number arrive fed up with failed attempts to get people to change their behavior, with many missing the irony as they bemoan, “I keep giving them the same feedback, and *they* never change. Maybe they just don’t want to.” What these folks miss is their own power to catalyze change by shifting the framing and approach of the feedback they give. I’ve seen and felt firsthand the difference between the “feedback masters” and “feedback disasters.” After every feedback conversation from one colleague who I had initially respected, I felt deflated. As a result, everything suffered: our product, our relationship, and my overall well-being. Later, I worked with another colleague whose feedback left me elated every time (actually!). I couldn’t wait for meetings with her because I enjoyed them so much and I knew that her feedback would help even my best work reach another level. On reflection, I realized that the content of the feedback from these two colleagues was strikingly similar, but the way my second colleague framed and delivered hers lifted me up instead of shutting me down. That difference answered an age-old question: How on earth do you get someone to change or grow? And how can you do that in ways that strengthen your connection instead of breaking it? By mastering the art (and science) of giving great feedback. The good news is that this is a learnable skill—and one that is core to my work as an executive coach and organizational behavior consultant. On a daily basis, I give feedback to clients and help them give feedback to their colleagues, as well as establish effective feedback systems across their teams and organizations. At my company, we also practice this in almost every meeting, and the growth it elicits is remarkable. One colleague said that after regularly getting and giving feedback using the GAIN framework, he felt “much more driven by purpose, rather than by fear.” So what is the difference between the feedback masters and disasters? How can you shift your framing and approach to ensure that the feedback you offer inspires people to take action and keep growing? My experience, informed by a wealth of research from multiple fields, shows that effective feedback shares several key qualities: ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/45cece91-fb95-4310-bf40-8d03869336f4_1476x971.png) I call this feedback approach the **GAIN framework**. GAIN is both a useful acronym for remembering these essential ingredients in effective feedback and the core of the philosophy. **It is profoundly more effective and inspiring to frame feedback based on the experiences or results we want to move toward (the** ***gain*****) instead of focusing only on what we want to move away from (the** ***pain*****).** We can often do this in just a few short sentences. Say an engineering lead is frustrated with a PM’s sloppy handoffs and the resulting delays. The engineering lead’s feedback to the PM might sound like: “I want us to reduce handoff friction so we can ship faster. The last handoff meeting we had, you didn’t share any background customer research, which I failed to flag at the time. Lacking that context led to multiple rounds of engineers pinging you with questions instead of making empowered decisions on their own. Can you include that next time?” ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/79b9d2c5-5959-433d-914a-8827e9c1ace6_1456x1072.png) The challenge is that feedback doesn’t typically show up in our brains in such a crisply articulated, ready-for-prime-time state. It might start as “Product just threw this over the wall; they don’t care about our time.” Or “They’re lazy; they always do this!” It takes some translating to get our raw reactions into a form that’s more likely to have the desired impact (the “gain”). That form must include: - What both parties ultimately care about (the **G**oal) - What needs to change to get there (the **A**ctions) - Why (the **I**mpacts) - *Who* will do *what* by *when* to start making progress (**N**ext actions) Thinking through the GAIN elements is a powerful way to understand and channel those raw reactions constructively. This often starts by using your feelings as a clarifying signal pointing to what you really care about, which should inform your Goal for the conversation. In cases where your feelings are particularly strong or unwieldy, preparation that’s even more focused on your “inner game” (a topic for a different post) becomes an essential step in leveraging the GAIN framework. Especially with complex or sensitive topics, including each of these GAIN elements in the conversation—often, but not always, in order—leads to greater receptivity and, consequently, more behavior change and more satisfying results. Once my clients understand how GAIN works, they see the ways they themselves have been unintentionally eliciting defensiveness and resistance and how, very practically, they can flip that on its head. One founder said, “GAIN helped me have structured conversations that have proven to be successful especially with ‘fragile’ team members. I’ve done this successfully with more than five people in just the past two weeks.” Another client, startup veteran [Brian Martin](https://www.linkedin.com/in/brian-e-martin-50575a2b/), described his experience this way: “After I gave feedback once or twice in this new way and saw people actually listen and change and drive the business results or the behavior that I was trying to help them toward, I was like, ‘This is great. I can do this.’ It created a positive feedback loop, and now it’s something I just do that I think people really appreciate about me.” As Brian notes, people start to realize that the function of feedback isn’t just for behavior change. When shared with skill and engaged with vulnerability, a substantive feedback conversation builds immense trust, meaningful connection, and deep gratitude. (For example, [many](https://www.lennysnewsletter.com/p/he-saved-openai-bret-taylor) [guests](https://www.lennysnewsletter.com/p/making-meta-andrew-boz-bosworth-cto) on Lenny’s Podcast and beyond rave about that exact impact that Sheryl Sandberg’s feedback had on them and their relationship with her.) This post will help you build confidence, fluidity, and effectiveness in giving feedback that people end up thanking you for. I’ll walk through a real scenario from a client who applied the framework within a challenging interpersonal context (with a micromanaging founder), explaining each “move” she makes. Then I’ll review phrases and principles you can use, and wrap up with three ways AI can help you prepare and master feedback. But first, where did GAIN come from? ## Why GAIN? The evidence behind the framework GAIN represents the convergence of principles that I encountered in research across over a dozen diverse domains, ranging from couples therapy to decision-making to workplace collaboration to teaching, all focused on how to inspire enduring behavior change. They all orient to the same thesis: Communicating what we all want more of (the gain) is more effective than communicating what we want less of (the pain). They don’t say that speaking about pain is bad, just that it’s helpful to contextualize it as the gap between where we are and where we want to be. First, let’s look at communication in the most sensitive domain: marriage. Watching just three minutes of a conversation, marriage researcher [John Gottman](https://psycnet.apa.org/record/1999-01159-003) was able to predict with 93% accuracy which couples would be happy, distressed, or divorced six years down the road; most people, even couples therapists, averaged [50% accuracy](https://onlinelibrary.wiley.com/doi/10.1111/j.1741-3737.2003.00130.x). How does Gottman distinguish the relationship masters from the disasters? His meticulous research, filming couples’ conflicts and tracking the couples over years, shows that couples with long-term happy, stable relationships open conflict conversations with more positive and less negative emotions than doomed ones, especially **focusing on sharing the “dream within the conflict.” Instead of just complaining, they articulate what it is they want more of and why (the** ***Goal*** **in GAIN)**. Hearing what you both want more of makes it easier to feel motivated and excited about the future rather than frustrated about what’s not working in the present. When [65% of startup failures](https://www.amazon.com/Founders-Dilemmas-Anticipating-Foundation-Entrepreneurship/dp/0691158304) are attributed to problems *within* management team dynamics, Gottman’s relationship research is also worth paying attention to at work. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/80a5a5aa-56aa-42d5-9a06-988560ea3636_1457x1283.png) The GAIN framing not only leads to behavior change, it can also elicit better decisions. [One study](https://pubmed.ncbi.nlm.nih.gov/7070445/) by Amos Tversky and colleagues showed that doctors given data with an *avoid* framing (“The mortality rate is 10%”) made the most effective treatment decision only 50% of the time, whereas those with the same inputs but using an *approach* framing (“The survival rate is 90%”) made the most effective decision 84% of the time. The meaning of the framing is identical (10% rate of mortality is the same as 90% rate of survival), but the *avoid* framing and its threat mindset distort our ability to think and process information clearly. This is precisely the ability we want people to be engaging when receiving feedback and considering a change in behavior. This dynamic plays out elsewhere in the workplace as well (and in a way I keep thinking about as leaders fret over AI adoption strategies). Amy Edmonson’s [research](https://med.stanford.edu/content/dam/sm/CME/documents/Edmondson-20Amy-20C.-20-E2-80-93-202003-20Framing-20for-20Learning-20Lessons-20in-20Successful-20Technology-20Implementation.pdf) compared teams in 16 different organizations implementing the same new technology. She showed that teams whose leaders pitched it using an “aspirational” framing (e.g. “This will help patients recuperate faster”) were far more effective than those using a “defensive” framing (e.g. “The reason for doing this . . . is to avoid being blindsided by competitive pressures in the future”). The successful leaders not only emphasized what to move toward, they also framed getting there as a team effort. Our brains are engaged in a constant calculus: approach or avoid, friend or foe? When we frame a conversation with the potential benefits of change and acknowledge what we need to change ourselves, it sends a strong approach signal—a powerful “we’re on the same team” message—to the other party. They are then more likely to hear it as “Hey, this information can be really useful to me, and I want to help this person too.” To explore what all this theory looks like in a real feedback conversation, see the case of one of my clients below. ## GAIN in the real world: Giving feedback to a micromanaging founder Laura is a VP of Business Development at a high-growth startup, where she reports to a founder who practiced micromanagement like it was her religion. Unsurprisingly, this was quite frustrating for Laura. It would have been tempting to approach this challenge by telling the founder how deflating this experience was and asking her to change (or by just avoiding the conversation altogether, growing resentful over time, and eventually looking for another job). Instead, Laura processed her own raw reaction first, using her emotions to uncover what she actually wanted to get out of a feedback conversation: the founder to give her more space. Laura understood Gottman’s “dream within a conflict” principle, so when the founder was complaining to her yet again about other teams—this time the sales team—Laura seized the moment. When she began the conversation, she named the founder’s **actions** and the **impacts** that mattered *to the founder*, and empathized with her on those points. Then, instead of dwelling on the founder’s pain or going into her own, Laura turned toward the possible **goal** *for the founder*, emphasizing the benefits for the founder in a way that aligned with her own goal. (GAIN conversations don’t have to present the goal first to be effective, as long as they are in the spirit of GAIN and the goal is clarified soon after.) Finally, she guided the conversation toward **next actions** that would accomplish *both* of their goals. Here’s how she masterfully led the conversation, as she reported it to me: **Actions and Impacts:** *“Right now, I can see you are on every sales call, email, in all of the Slack channels across every team, and I can only imagine how exhausting that must be.* *I imagine that you feel like you have to do it because you feel the job won’t get done unless you do.”* **Goal:** *“I think there might be another approach here that could create more time for you to focus on the areas where your impact is greatest and, at the same time, help the team step up, do better work, and feel more motivated. It could even reduce some of the stress and burnout you were talking about.”* The founder responded honestly, *“I know, I know, and I want to fix it. But I just don’t see any other way of doing things, and I don’t see the sales team getting the results.”* A feedback opening line is an invitation, and Laura’s was so inviting, the founder accepted and immediately engaged with it. The response, though brief, revealed lots of useful information: the founder’s desire to change, the obstacle to that change, and what matters most to her—getting results. Laura can leverage this to move the dialogue forward toward her desired outcomes. Next, Laura took her founder’s concern seriously and engaged her further, asking more questions that invite the founder to reflect more deeply on the obstacle she’s raising while emphasizing the importance of trust: *“If that’s the case, do you have the right people? If you can’t trust people to do the work, are they the right people?”* *“Yeah,”* the founder said, *“I think that’s the bigger question—the thing I need to think about.”* Having built a sense of “I’m on your team” in the conversation, Laura continued and challenged her: *“The interesting thing is, though, I know you trust* me *a lot—you’ve mentioned it, and I’ve even observed it—but you still sometimes do it with me.”* Here, Laura doesn’t judge or criticize; she just references her observation of the founder’s pattern of actions (participating in all calls, emails, Slack channels), as well as the exceptions to that pattern (“You’ve mentioned [your trust in me], and I’ve even observed it”). The founder acknowledged, *“I know. It’s just my way; I’m so used to doing it. I get everything done myself. I push, push, push.”* So Laura asked, *“What would be more helpful for you, so you don’t have to do that?”* Instead of solely focusing on what the founder can do differently, Laura’s question broadens the possible levers for change to include herself and others. Some people are wary that this approach might mean less accountability for the feedback recipient, letting them off the hook. Rather, more levers means more leverage to get what we want. **Next actions:** The founder replied, *“I really appreciate when you proactively send me updates—which I know you already do. But also, you can just tell me when I’m doing it.”* Laura confirmed: *“Sounds great. So moving forward, I’ll continue proactively sending you updates, you’ll work on stepping back—i.e. not jumping in or pinging me with status checks—and I’ll name it in the moment if the old pattern rears its head.”* These are clear, concrete actions that they can both take to change the feedback recipient’s pattern of checking in unnecessarily. The founder conveyed her intent to stop while also acknowledging her imperfection and inviting Laura to help her make this change—which was particularly empowering given the power dynamic. The improvement took time to sink in, though Laura felt full agency to help move it forward. A month later, she shared that the founder had reverted to micromanagement mode three times that week, but then stopped after Laura told her when she would send an update. Over time, the micromanaging pattern ceased almost completely, and not only that; Laura told me that she felt newly confident in her ability to initiate challenging conversations. Laura navigated a tough dynamic with someone holding greater organizational power and emerged with both the founder’s commitment to change and an invitation to support that change—while building her own confidence and skills. ## GAIN feedback principles and phrases This example models both the practical steps of the GAIN framework and the spirit of it: care *and* challenge, offering empathy *and* expressing what the feedback giver wants—in that order. Here’s a simple breakdown of Laura’s conversation through the lens of GAIN, though her goal comes in the third sentence, not the first: ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/25f99121-c4fb-4045-a637-d7a1d9c6c88c_1456x1567.png) You can download the GAIN template [here](https://maven.com/p/ad08cc/the-gain-feedback-template), with prompts to guide you. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/1204b96c-a179-4f1e-9711-e935c8c28fe6_1456x1583.png) Let’s dive into the nuances and some helpful phrases and principles you can consider, drawing on Laura’s example and others. ### G: The Goal The essence of the Goal section is naming the topic and pointing to the possible benefits of making a change in such a way that they feel you’re on their team. These benefits can be for you too, but they *must* be appealing to the feedback recipient. This is not what you want them to *do* differently; it’s the impact created by them doing it. What will they gain from making a change? If they understand that motivation up front, they’re hooked, and everything else flows from there. What struck me most in the way Laura shared her feedback is that it didn’t sound like a complaint or criticism at all. Instead, she shared it in a way that genuinely came across as a desirable opportunity for both the founder and her. She mentioned the current pain but oriented to the possible gain, “the dream within the conflict” that Gottman’s research showed to be so effective. Here are a few ways to approach this. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/c8a0f66f-a2b2-436e-be95-2f2099bde854_1457x1173.png) If you can find even one instance of someone already using the desired behavior, it will help them recognize what that looks and feels like. Further, they will feel more confident in their ability to act in that way. The impact of that behavior serves as the Goal for the feedback. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/257ec72b-4f53-4d1d-8044-31a37eedc02e_1456x1177.png)![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/ddb2ce8d-b4be-4d9d-8723-a361dd5439a2_1456x795.png) “Even more” conveys that they both have some foundations to build on and that they’re not there yet. Sometimes it’s important to emphasize the “not yet,” the gap between where performance is now and where it needs to be. In those cases, we must offer both challenge and support. These qualities are usually understood as opposites (“I’m either extremely honest or extremely caring”), but that is a misconception. One randomized, double-blind controlled [study](https://psycnet.apa.org/record/2013-28213-001) from Stanford tested the impact of integrating these two qualities using the sentence “I’m giving you these comments because I have very high expectations [challenge] and I know that you can reach them [support].” Compared with a placebo—“I’m giving you these comments so that you’ll have feedback on your paper”—the first sentence more than *quadrupled* the number of revisions to the original work that one group of students did, and the revisions themselves were on average 26% better. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/d5e14e86-d0df-459a-a3aa-ac29efbdf401_1457x1164.png)![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/8ae28986-c36d-4a44-831f-76a08ed5887e_1456x1495.png) Each of the phrases above can engage people’s interest in the conversation, but the key is that it be an actual conversation—a dialogue, not a monologue. One of the most common mistakes I see people make is *delivering* feedback instead of exploring it together. The point is not you giving the feedback; it’s them getting it—and ideally, you getting some insights too. In this way, communication is like a game of catch: the other person needs to actually receive and hear the feedback in order for the game to be successful. GAIN largely focuses on throwing better so it’s easier for them to catch whatever you’re throwing, but ultimately, effective communication is an alternating rhythm of throwing and catching. You can initiate this rhythm by pausing routinely and asking a question that elicits their reflections, reactions, and their own feedback for you. This creates two benefits: (1) it can reveal what message they are actually getting, giving you the opportunity to clarify immediately if needed, and (2), equally important, it can expand your understanding of root causes, revealing blind spots you may have (and keeping you from putting your foot in your mouth). ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/a3e32712-7780-4381-a2a8-9045cee75633_1456x1588.png) Especially if you’re not clear yet on why they would care or what *their* goal is, listen before you talk. ### A & I: Actions & Impacts It’s not just the goal that determines receptivity or defensiveness. The most common move I see that leads people to put up their shields is hearing critical judgments. They’re an almost guaranteed way to trigger people and derail the conversation. Feedback is effective when it’s actionable. To make it actionable instead of objectionable, feedback masters share *observations* of people’s actions and their impacts, not judgments. “Response times to client emails averaged 4.2 days last month” lands very differently than “You’re being unprofessional.” Even if this seems obvious, it becomes challenging when we’re annoyed and feeling justified in our annoyance (“Product just threw this over the wall; they don’t care about our time!”). [Our judgments feed our emotions](https://speakerdeck.com/jackcohen/cycle-of-conflict-or-cycle-of-connection?slide=13), which then amplify our judgments. And what seems straightforward in theory can be deceptively difficult in the face of real-world situations. In over 50 feedback workshops across dozens of tech companies, I’ve asked people to share real-world feedback examples, and there hasn’t been a single workshop in which everyone agreed on whether every example we talked through was a judgment or observation. Fortunately, recognizing and translating our judgments is a skill easily developed with practice, one even more easily practiced now with AI. To do so, you can use this [custom GPT](https://chatgpt.com/g/g-68821735dc108191a1829abc763ad284-distinguishing-observations-from-judgments) or just input this prompt: - *Give me examples to practice identifying whether something is an observation or a judgment. Explain why I’m right or wrong after I answer, then give another example, making them increasingly ambiguous each time.* There’s a strong correlation here: the more clearly we see what is impacting progress, the more quickly we can identify new strategies to move toward our Goal. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/8eafbb85-16e4-4557-ae79-b69c19ee2dcf_1456x1404.png) Laura never called it “micromanagement” to the founder. She didn’t even use the word “feedback”; she just described the actions and their effects as she observed them. **This gets to the essence of feedback as I define it: creating visibility into people’s actions and their impacts**. Simple cause and effect is a line of sight that the actor often lacks. Often, not only do we lack visibility into how we are impacting others; we lack the ability to step back from and see our own behavior. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/9c4ca3c3-0d48-410c-b614-7ed6b3400736_1456x1303.png) **Avoid flattering judgments too** Even for well-practiced feedback masters, it can be tempting to deploy critical judgment’s close cousin, flattering judgment. Well-intentioned people might offer praise and compliments, like “You’re a rockstar PM” or “You’re so brilliant. You have such a gift,” to motivate or thank colleagues. But just as critical judgments can trigger defensiveness, flattering ones can destroy growth mindset, autonomy, and productivity, and foster people-pleasing. The “rockstar PM” worries about maintaining that image if she fails, turning challenges into threats rather than opportunities to learn and grow. Carol Dweck’s [mindset research](https://pubmed.ncbi.nlm.nih.gov/9686450/) explains why: **If the cause of my success is my brilliance, what does that imply about the cause of my potential failures?** This is especially pernicious in contexts that require experimentation, where failure—however temporary—is inevitable. This dynamic is exacerbated when the flattering judgments come from a manager or executive. That rockstar PM’s focus reorients from iteration and impact to seeking approval. She could spend a lot of energy trying to get an “Exceeds Expectations” on her next report card—I mean performance review—instead of focusing on influencing user behavior or being more effective in her work relationships. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/8c344f4a-e721-4f7a-8b1d-9e9654526d48_1517x1330.png) By contrast, observations of people’s actions and impacts point them to what is within their control—the actions they can change or keep doing and the potential impacts of those actions. They see how to increase their own ability to create their desired outcomes. This means that giving this kind of actionable feedback can actually *evoke* feedback recipients’ growth mindsets, or [as Satya Nadella frames it](https://nextbigideaclub.com/magazine/conversation-microsofts-ceo-on-the-power-of-being-a-learn-it-all/17851/), help them shift from know-it-alls to learn-it-alls. If you want greater openness and motivation, share impacts, not judgments! [Here’s how](https://actionablewisdom.beehiiv.com/p/how-to-motivate-your-team-for-free). **Acknowledge your own actions and their impacts** Judgments aren’t the only trigger for defensiveness. When you as the feedback giver have contributed to a situation but you aren’t noting that, the recipient’s attention can latch onto this gap. “Yeah, but what about *your* actions?” they wonder. You can preempt this reaction by simply and directly acknowledging your contribution. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/1b54dee2-c911-4433-b843-104d56822bd1_1457x1484.png) As a social species, humans unconsciously mirror the behavior we see in others around us. The crowd waits at a light, and so do we; we start jaywalking, and others are more likely to as well. This plays out interpersonally too. Far too often, we approach feedback conversations with the implicit attitude of “You are the one who needs to change.” From this angle, it’s no surprise that what we get back is the other person looking at us thinking, “No, *you* are the one who needs to change.” We tend to get what we give. But this human instinct is actually good news: if you want to lead a conversation and a relationship in a specific direction, just start moving in that direction yourself. The message switches to “I think we can both shift some things that will make it easier to change and move toward that Goal.” This is one of most disarming feedback approaches I’ve ever encountered—and one of the least-used. Once we have acknowledged our contribution and shown that we are monitoring our own behavior, the other person can release their focus on our actions, freeing up their attention to concentrate on their own actions. There’s an added benefit here for the feedback giver: shifts that I make in one relationship often apply to other relationships too. When I was frustrated with one client constantly rescheduling, I realized how I was contributing to and even encouraging the problem by always saying yes to requests. Quick check-ins with all my clients around minimizing changes dropped rescheduling requests to zero across the board. More levers, more leverage. ### N: Next actions What do we do with those levers we identified? Once feedback has been shared and received, the final step in an effective feedback conversation is defining next actions. These could be actions for the feedback recipient, the feedback giver, or both. The clearer and more concrete our action-and-impact observations are, the more obvious the potential next actions become. Sometimes it’s a very direct link: “Hey, when you don’t schedule interviews in the time frame you said you would, it slows down our hiring process. Can you schedule those this week and let me know if there are any blockers?” Other times, arriving at next actions is more complex and uncomfortable. Our insides are screaming, “This conversation is awkward. Get me out of here already!” But even an otherwise-great feedback conversation will often amount to nothing if we don’t close it clearly and concretely. There are a few ways of doing this that increase buy-in and follow-through. **First, ask for their ideas** In an often-maddening phenomenon, people will resist change that they themselves want if it appears to come at the expense of their autonomy. We may want to change, but we want to choose it. One simple way of supporting someone’s sense of autonomy is to ask for their ideas of what to change before offering our own. When the other person is really on board with the Goal and open to hearing what we’ve shared, given even a few moments of brainstorming, I have found that they regularly come up with useful options I had not considered. They also tend to be more motivated to pursue ideas they have generated themselves. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/a7c9b9af-7867-4feb-86cb-e783612f1820_1457x907.png) For feedback that is difficult to hear regardless of how it’s shared, a feedback conversation might actually be a series of conversations, so it’s often best to sleep on it and schedule a future time to return and commit to next actions (not just “later”). **Co-create through brainstorming together** If you are itching to share your own idea but also want to involve them, this option opens up the opportunity for both. It also lets you build on each other’s ideas. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/a779ab49-f7e7-47c0-b212-19e9f51a3422_1457x690.png) **Make specific requests—and make it easy to say no** Sometimes asking for the other person’s ideas can feel disingenuous when all you want is to propose your own. In that case, share your idea, and offer it as a suggestion or a request in a way that still honors their autonomy—in other words, not as a demand. To support this feeling, make it easy for them to say no to or edit the request. A [meta-analysis](https://www.researchgate.net/publication/344807911_The_effectiveness_of_the_But-you-are-free_technique_Meta-analysis_and_re-examination_of_the_technique) of 52 studies showed that “Feel free to say no” doubled the percentage of time that people said *yes* to a request. Supporting autonomy conveys real respect, which preempts resistance, engenders goodwill, and helps people think clearly about the suggestion on its own merits. If they actually say no, you will get to learn what’s blocking them and how to incorporate that in the next actions. (And if you’re feeling attached to your idea and actually want to demand it, remember that it’s the Goal you care about, not the way they get there.) It’s also important to ensure that your specific request includes what you want more of, not just less of. As psychologist Marshall Rosenberg, the author of *Nonviolent Communication*, said, “You can’t do a don’t.” Otherwise, when they find themselves in the same situation again, they’ll be unclear about what alternative behavior to engage in. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/f57e47a0-0e33-4194-89ba-ffb48e805918_1457x1102.png) **Frame it as an experiment** When someone is considering a new behavior, that means there’s an element of uncertainty. Understandably, there can be some hesitance about committing to do something they’ve never done before, no matter how much they follow the logic. Instead of committing to something new “forever,” explicitly acknowledge it as an experiment for a set time period. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/d2ef4a74-80f7-41a6-bf29-9311f2acd5b2_1457x850.png) **WWW:** ***Who*** **will do** ***What*** **by** ***When*****?** This question moves growth, progress, and whole companies forward faster than any other one I know. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/954bd6ef-863a-4891-b32c-1c66b7e33e6f_1457x693.png) “Let’s” and “should” are insidious impostors when it comes to action items. *Let’s*: Is that you or me? *Should*: Is that an idea we’re considering or a commitment? When you hear these words, use them as an automatic signal to pause and clarify who will do what by when. **Schedule a check-in** If you don’t set a clear check-in, you run the risk of not seeing the change you were hoping for and having to initiate a follow-up feedback conversation if you still want to close the gap. The discomfort of this is higher, so it’s easier to avoid, and the frustration festers into resentment and helplessness. If instead you establish a time and date to check in, you set up three more desirable possibilities: - An agreed-upon check-in time often increases the accountability and expectation of change. - The behavior may have changed, and you have a built-in moment to appreciate the impacts you’re seeing and the progress toward the Goal, encouraging even more development. This reinforces a growth mindset. - If the behavior hasn’t changed or hasn’t had the desired impact, you will recognize this sooner and be able to troubleshoot and course-correct faster. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/80049542-eea3-427c-8b89-3fefbaca8f78_1457x805.png) # AI in feedback That’s a lot to digest and consider when preparing and leading a feedback conversation. Fortunately, you’re not on your own anymore. While many people fear AI will erode human connection, I see the opposite happening with clients and in [our workshops](https://maven.com/actionablefeedback/managergpt): **People are using AI as a tool for expanding interpersonal skills and deepening human relationships.** Here are three ways AI can support you in using the GAIN approach and having much more effective, connective feedback conversations: 1. FeedbackGPT (see below): Help you prepare what to say 2. Voice Mode: Help you practice 3. Post-conversation retrospective: Help you reflect and improve ### 1. FeedbackGPT: Help you **prepare** what to say You can build custom GPTs that [give feedback for you](https://www.lennysnewsletter.com/p/how-to-become-a-supermanager-with) in some situations. But when you can’t replace yourself with it, AI can help you prepare much higher quality feedback. You can vent your raw, unfiltered material into any LLM and prompt it to craft more effective feedback. While you’ll often get smooth responses from an LLM, one big risk is that they won’t challenge you to examine your own contributions to the situation or help you learn. You can overcome this by asking for it specifically in your prompt. You can also use [FeedbackGPT](https://actionablewisdom.beehiiv.com/p/feedbackgpt-b453), a custom GPT I created using the GAIN principles, which takes you step by step through the reflection process and then crafts GAIN-framed feedback for you. ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/c28cc3e3-0b7b-4860-9e28-91503bca8881_512x375.gif) ### 2. Voice Mode: Help you practice As we discussed earlier in this post, feedback is a dialogue, not a monologue. As such, preparing what you are going to say is only half the equation. The rest is practicing how to actually engage another voice in an interaction, listen to their perspective, and respond in a way that integrates points from both parties. Feedback conversations are more powerful when everyone leaves with some perspective or takeaway they didn’t have before. But all these skills take practice. For the conversation above, Laura told me that while our preparation clarified her message, role-playing it together and on her own several times beforehand helped her approach the actual conversation with confidence and respond flexibly in the moment. The single greatest growth catalyst I have seen in clients and workshop participants comes when they practice, we give them immediate feedback and coaching on their practice, and then they do it again. This is now possible for everyone with the availability of 24/7 support for the practice-feedback-practice loop—and it extends beyond practicing what to say, to [how to listen](https://www.linkedin.com/posts/jackadamcohen_radical-idea-for-husbands-everywhere-stop-activity-7296180121894297601-DnmJ?rcm=ACoAAAMhRhgB8w9ICmC6Uistv-GiSYOH_tHzxB0). With a simple prompt, you can practice in chat form or engage Voice Mode. You could easily get in three or four simple practice reps in just 15 minutes and feel much more confident going into the actual conversation. #### **Here’s a sample prompt template:** ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/e95bb328-38bf-4f43-8a4e-66cf759b5d99_1457x2157.png) Alternatively, you can flip roles and ask AI to take on your role while you play the feedback recipient. This will model some possibilities for you with the side benefit of helping you access the feedback recipient’s perspective. It’s very easy to become completely absorbed in our own perspective and focused only on our side of the dance. I keep noticing that preparing and practicing with AI in these ways can help us consider both the other party’s potential reaction and their perspective, increasing our empathy in the process. ### 3. Post-conversation retrospective: Help you reflect and improve Once you have had the feedback conversation, you can continue to up your feedback game. If your call recorder joined, upload the transcript into a general LLM or FeedbackGPT and ask it for feedback or coaching on your feedback. Prompt: *You’re a world-class master executive coach with a specialty in the art and science of feedback. Analyze the transcript attached and give me feedback on this feedback conversation. For each of the growth areas, give me some specific, evidence-informed suggestions that world-class practitioners use in those areas. (Use only real evidence, not hallucinations.) Include time-stamped references and quotes.* This will yield insights about how to improve, but insights alone are often not enough for change. They need to be practiced. You can then return to #2 and ask AI to role-play the conversation, or a segment of it, as you focus on that skill. ## Recap: Unlocking growth with the GAIN framework Just as user feedback is indispensable for improving a product, teammate feedback is indispensable for improving a team and the people on it. Unfortunately, feedback is not always shared. One founder I coach, [Nischal Nadhamuni](https://www.linkedin.com/in/nishchaln/), summarized both sides of the feedback equation: On one hand, “The key to everyone’s growth is buried in someone else’s head—and they know it.” On the other hand, “People know *exactly* whatfeedback they want to give—they’re just withholding it.” Our work is to make it easier to give and get that feedback so we can grow ourselves, iterate our products, and improve the teams that build them. What’s more, the level of depth and connection that comes from sharing feedback skillfully can be electric. Here’s what two co-founders said after their whole executive team gave them feedback in a [Live 360](https://actionablewisdom.beehiiv.com/p/live-360s). ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/a79a7df3-e66c-4f1a-9987-b6e329b527a6_498x568.png) GAIN offers a simple, straightforward, science-based framework for communicating that immediately establishes a sense of shared purpose and—even if someone isn’t making progress toward that purpose right now—conveys that “I’m here, on your team, to help you get there.” ![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/bdea2422-920d-4adb-941a-bff9c8cf95ed_1456x1886.png)![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/45054717-5b85-4b7d-bd7d-643b2c86307a_1457x1286.png)![Image from Introducing the GAIN framework for feedback: an evidence-based approach to](https://substack-post-media.s3.amazonaws.com/public/images/505d9026-3631-414d-a527-475802d5e973_1457x1568.png) *Thanks, Jack! For more, check out his course [ManagerGPT: The AI Tools and Human Systems to Scale Yourself and Your Team Fast](https://bit.ly/3VK5fGt), and for a demo of both GAIN feedback and the AI tool Jack built to support it, check out his free 30-minute lightning lesson, “[FeedbackGPT: Give Feedback People Thank You For](https://maven.com/p/6f79f8/feedback-gpt-give-feedback-people-thank-you-for).”* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [38/46] A free year of Devin: the world’s most advanced autonomous AI software engineer *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Favorite AI and PM courses](https://maven.com/lenny) | [Favorite public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** ![Image from A free year of Devin: the world’s most advanced autonomous AI software engineer](https://substack-post-media.s3.amazonaws.com/public/images/caad4bfb-89ca-401e-a0cf-bc40d09870a5_2912x2267.png) The vision for the Lenny’s Newsletter [Product Pass](https://www.lennysnewsletter.com/p/productpass) is to give you access to the most cutting-edge and important tools and technologies, so that you can experience them—not just read about them. Beginning this week, I’ll be adding a new premium product to the collection each month. Some will be available to all annual subscribers, while others will be exclusive to [Insiders](https://www.lennysnewsletter.com/subscribe?plan=founding) (our premium annual tier). **Starting today, [Insider subscribers](https://www.lennysnewsletter.com/subscribe?plan=founding) get a free year of [Devin](https://devin.ai), the world’s most advanced autonomous AI software engineer.** The team behind Devin has never offered a deal like this before. This partnership is a significant investment from them. With this collab and more, I hope to get more people exposed to the future of how products will be built. This adds an additional $1,350 in value to your [Insider subscription](https://www.lennysnewsletter.com/subscribe?plan=founding) (for just $350/year)—on top of the more than 15 premium products you already have a year of free access to (worth over $10,000): **[Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass).** Not to mention this newsletter and a thriving private Slack community. [Subscribe to Insider](https://www.lennysnewsletter.com/subscribe?plan=founding) And there’s much more in store in the coming months. ### **What is Devin** Devin is an autonomous AI engineer. You assign it bugs, features, or complex refactors through Slack, Linear, Jira, or its website. It writes the code, then sends you a pull request to review. Tens of thousands of people are using Devin in production today, including top teams at Ramp, Nubank, Goldman Sachs, Citi, and Microsoft. [Watch on YouTube](https://www.youtube.com/watch?v=f9wR418HU4w) ### **Why I love Devin** It’s the most widely deployed enterprise-ready AI engineer on the market, and every forward-thinking builder I know who’s used Devin absolutely loves it: 1. [Claire Vo](https://www.linkedin.com/in/clairevo/) told me that Devin is the #2 contributor to her six-figure [ChatPRD](https://www.chatprd.ai/) business (behind only her, for now) and touches 100% of their PRs, by reviewing code, updating documentation, or writing all of the code. [Here’s Claire’s interview with Devin co-creator Scott Wu](https://www.youtube.com/watch?v=7m_xKFqSxTo). ![Image from A free year of Devin: the world’s most advanced autonomous AI software engineer](https://substack-post-media.s3.amazonaws.com/public/images/75d136d7-3d1c-4525-a55c-1e4ac5a09f40_1836x482.png) 2. [Sahil Lavingia](https://x.com/shl) has been proudly sharing that Devin is [Gumroad](https://gumroad.com/)’s #1 contributor of code, with over 1,500 merged PRs (averaging 10 per day). 3. [Dan Shipper](https://x.com/danshipper) said, “Working with Devin is familiar because it feels like adding a few junior engineers to your team. You toss them tasks, and they’ll get started with enthusiasm . . . It’s programming leverage—it’s productivity power.” [Here’s Dan’s chat with Scott Wu](https://youtu.be/ectyyex0IBo?si=CPiMhXm-uKbScnlC&t=1773) comparing Devin to Claude Code. [Watch on YouTube](https://www.youtube.com/watch?v=ectyyex0IBo) 4. Devin ranks #1 in the [Builder Arena](https://www.designarena.ai/builder), ahead of Cursor, Lovable, Replit, and Figma Make—a benchmark based on thousands of people blindly comparing the output of each AI’s product being given the same prompt. ![Image from A free year of Devin: the world’s most advanced autonomous AI software engineer](https://substack-post-media.s3.amazonaws.com/public/images/b27c1b94-27b7-442c-9df0-cc1d49ae3981_2578x1312.png) 5. Also, [Scott Wu](https://www.linkedin.com/in/scott-wu-8b94ab96/) (CEO of [Cognition](https://cognition.ai)) is something else: ### **How you might use Devin** Most people use Devin to take on their bugs and feature requests, but you can work with Devin to: 1. Scope new roadmap ideas 2. Update UI and add visual polish to your site 3. Handle QA on product changes 4. Keep internal documentation up to date 5. Take a first pass at data analysis requests 6. Take on the most tedious items on your backlog 7. Build SaaS integrations 8. Send daily summaries of shipped changes 9. Tackle complex migrations: [Nubank](https://nubank.com.br/) used Devin to migrate their over-6-million-line ETL monolith over the course of a couple of weeks, which otherwise would have been an 18-month and 1,000-engineer project. 10. Increase your unit test coverage: [Litera](https://www.litera.com/) used Devin to increase their test coverage by 40%, which reduced their regression cycles from three weeks to two days. Most people begin with Devin as an AI software engineer, handing off well-scoped software engineering tasks. They then expand into QA, product, analytics, and CX. The goal is to take on all of the lower-level work so that you have more time to focus on figuring out *what* to build and how to get it into the hands of users. Check out their [website](https://devin.ai) for more. ![Image from A free year of Devin: the world’s most advanced autonomous AI software engineer](https://substack-post-media.s3.amazonaws.com/public/images/f16380f5-6744-4264-bcd9-c36e743cd057_2752x1084.png) ### **How to redeem the deal** As an **[Insider](https://www.lennysnewsletter.com/subscribe?plan=founding)** subscriber, you get one free year of Devin Core with 50 ACUs a month, valued at $1,350. This is roughly equivalent to a full-time engineer working for a week straight, each month. **To redeem this deal:** 1. Become an **[Insider](https://www.lennysnewsletter.com/subscribe?plan=founding)** member (this is an upgrade from the regular annual membership). 2. **[Claim your free code here](https://lennysproductpass.com/).** ### **Additional offer details:** 1. **You must be a new customer of Devin** to take advantage of the free year. If you’ve already paid for Devin before (monthly or yearly), you won’t be able to get a free additional year of that specific product. 2. **Both existing and new subscribers** of Lenny’s Newsletter are eligible for this deal. Existing subscribers can [redeem the offer here](https://lennysproductpass.com/). 3. **Your one free year of the product begins when you redeem the deal on the partner’s website**,not when you purchase this newsletter subscription or claim your code. Enjoy! And much more to come. Sincerely, Lenny 👋 --- ## [39/46] Everyone should be using Claude Code more *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Favorite AI and PM courses](https://maven.com/lenny) | [Favorite public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *Subscribers get **19 premium products for free for one year**: [Lovable, Replit, Gamma, n8n, Bolt, Devin, Wispr Flow, Descript, Linear, PostHog, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, ChatPRD + Stripe Atlas](https://www.lennysnewsletter.com/p/productpass) ([terms apply](https://www.lennysnewsletter.com/i/168890236/important-offer-details-read-before-subscribing)). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/6b68fd37-43ed-49a8-889d-7f49e6102eb0_1456x970.png) Ever since [my chat with Dan Shipper](https://youtu.be/crMrVozp_h8?si=Kx_XeBfEM2pPJ1Ib&t=429), I couldn’t stop thinking about his hot take that [Claude Code](https://www.claude.com/product/claude-code) was the most underrated AI tool for non-technical people. A few weeks ago, I finally started playing around with it, and holy sh\*t, we’ve all been sleeping on Claude Code. **The key is to forget that it’s called Claude Code and instead think of it as Claude** ***Local*** **or Claude** ***Agent*****.** It’s essentially a super-intelligent AI running locally[1](#footnote-1), able to do stuff directly on your computer—from organizing your files and folders to enhancing image quality, brainstorming domain names, summarizing customer calls, creating Linear tickets, and, as you’ll see below, *so much more*. Since it’s running on your machine, it can handle much larger files, run much longer than the cloud-based Claude/ChatGPT/Gemini chatbots, and it’s fast and versatile. Claude Code is basically Claude with superpowers. **To inspire your own ideas, I’ve collected 50 of my favorite and most creative ways non-technical people are using Claude Code in their work and life. This list includes my own favorite use cases, and the best examples y’all shared with me on [X](https://x.com/lennysan/status/1960417604948123663) and [LinkedIn](https://www.linkedin.com/posts/lennyrachitsky_do-you-use-claude-code-or-some-other-cli-activity-7379238388391923713-nVnf?rcm=ACoAAABGvmoB4S920iEQfSFO_P91nw2wPqfPoic) of how you use Claude Code.** A huge thank-you to the more than 500 of you who shared your stories. 🙏 ## **But first, let’s install Claude Code on your computer** 1. **Open your Terminal app** 1. On a Mac, press **Command (⌘) + Space**, type **“Terminal”**, and hit **Return** 2. In Windows, press **Windows key + R**, type **“wt”**, and press **Enter** 2. **Install Claude Code** 1. On a Mac, run this command: *curl -fsSL https://claude.ai/install.sh | bash* 2. In Windows, run this command: *irm https://claude.ai/install.ps1 | iex* 3. **Launch Claude Code:** *claude* If you run into any trouble, just ask your favorite chatbot for help. Or better yet, install [Warp](https://www.warp.dev/) ([free with your newsletter subscription](https://www.lennysnewsletter.com/p/productpass)!), which replaces your local terminal app and automagically solves any issues you encounter trying to install stuff like Claude Code. That’s how I solved the problems I ran into, and I highly recommend you do the same. ## **Five ways I’ve been using Claude Code this month** #### 1. Clearing space on my computer Prompt: “*How can I clear some storage on my computer?*” I then discuss my options. ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/12f869f9-6519-4870-be16-dffe585eb475_1600x690.png) #### 2. Improving the image quality of screenshots Prompt: “*Improve the image quality of [filename]*”. I used this many times for the screenshots below. ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/dc76e64e-df51-4461-83e8-1bbd89b0b67c_1962x1598.png) #### 3. Downloading YouTube videos Prompt: “*Download this YouTube video: [URL]*”. Then I ignored all the warnings 🤫 ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/b4166da8-693b-432f-a509-284e4848e9de_2132x1260.png) #### 4. Downloading all of the images embedded inside a Google Doc Prompt: “*Download all of the images in high-res from this Google Doc: [URL]”.* This paired well with item #2. ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/2aee1f08-2f50-4ae9-b6db-0a1dec2e7a38_2394x1066.png) #### 5. Picking a random raffle winner from a Google Sheet of submissions Prompt: “*Pick a random row from this Google Sheet to select a winner for a giveaway.*” I used this for a recent Sora 2 giveaway in our subscriber Slack community. ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/6acf2721-9fb9-424b-9737-8ccc235e42c0_2170x770.png) # **50 creative ways non-technical people are using Claude Code** Out of the over 500 ideas you shared with me on [X](https://x.com/lennysan/status/1960417604948123663) and [LinkedIn](https://www.linkedin.com/posts/lennyrachitsky_do-you-use-claude-code-or-some-other-cli-activity-7379238388391923713-nVnf?rcm=ACoAAABGvmoB4S920iEQfSFO_P91nw2wPqfPoic), here are my favorites (with screenshots): ### **1. Brainstorming domain names, from [Ben Aiad](https://x.com/benaiad)** > “Just describe your project, and it’ll suggest creative options across multiple TLDs (.com, .io, .dev, etc.) while verifying what’s actually available to register.” ### **2. Finding high-quality leads, from [Jeff Lindquist](https://www.linkedin.com/in/jeff-lindquist-0497b625/)** > “I literally just typed: look at what I’m building and identify the top 5 companies in my area that would be good for a pilot for this. Then I go to LinkedIn and message them. If it’s not clear, I do this in the source directory of the code of my app so the first thing it does is figure out what it is that I’m building.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/01eff5de-1cb0-4601-897e-6c1b931a1ee6_2560x1704.png) ### **3. Same as above, but instead, scraping GitHub repos, from [Sergei Zotov](https://www.linkedin.com/in/szotov/)** > “My product masks sensitive data in code assistant queries. So Claude Code proposed the idea to find potential leads in the GitHub repos, by searching for the actual sensitive values in them (and whether in the repo we see some evidence of using coding agents). This was actually genius—not only does it filter out a lot of companies, but it also provides instant value to the lead. Here’s what it came up with: repos list, priority score, even LinkedIn URL.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/5be22cc6-7cd7-4366-9e8d-cc237e8542f4_1280x428.png) ### **4. Noticing when you’re avoiding conflict, from [Dan Shipper](https://x.com/danshipper)** > “I download all of my meeting recordings, put them in a folder, and ask Claude Code to tell me all of the times I’ve subtly avoided conflict.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/489eaeb3-c920-472e-9f0b-6745759e4867_1484x1853.png) *Pro tip: Use [Granola](https://www.granola.ai/) ([first year free](https://www.lennysnewsletter.com/p/productpass)!), and ask Claude Code how to download your meeting notes into a folder.* ### **5. Figuring out why your computer is running slow, from [Anthony Roux](https://www.linkedin.com/in/anthonyrouxfr/)** > “I sometimes use Claude Code for system diagnostics when my Mac slows down. > > I use it to check load averages, memory pressure, disk space, stuck processes, and swap activity, then it dives deeper to find what’s actually causing issues. It can calculate cache sizes, check Docker usage, find Time Machine snapshots eating space, etc. > > It is usually faster and more user-friendly than running all the commands and trying to extract the right numbers myself. It can explain what the analyses mean and why they matter, and suggests fixes with the actual commands while assessing the risk of running each of them.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/a522d04f-f7a7-4e50-8032-a8bcc8ec5392_1904x2082.png) ### **6. Cleaning up messy invoice files, from [Martin Merschroth](https://www.linkedin.com/in/martin-merschroth/)** > “I use Claude Code to sort my invoices for taxes. It reads each file in a messy folder, renames it to ‘YYYY-MM-DD Vendor - Invoice - ProductOrService.pdf’, and moves it into the right folder.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/48d52c72-a709-4b94-beb4-c70abebdbada_739x645.png) ### **7. Organizing files and folders across your computer, from [Justin Dielmann](https://www.linkedin.com/in/justindielmann/)** > “For me, staying organized is a huge chore. The cognitive load of figuring out where to store files and keeping everything clean and up to date was insane. My hack: I run Claude Code from my home directory and use it as my personal organization assistant. I’ll ask it things like: > > 1. ‘Find duplicate files and help me decide which to keep’ > 2. ‘Organize these downloads into proper folders’ > 3. ‘Review my directory structure and suggest improvements’ > 4. ‘Find old files I probably don’t need anymore’ > > It’s like having a thoughtful assistant who actually understands context and can make smart decisions about file organization. Game changer for reducing mental clutter.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/a0c4a412-51f2-48c5-8295-dc0624f5ad59_1136x1242.png) ### **8. Building a slide for your child, from [John Conneely](https://www.linkedin.com/in/johnbconneely/)** > “I built my own DIY subagent last week to help me build a slide tower for my son 😀” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/cc626b43-9eef-4145-a899-7c88d0c2a28c_1116x1566.png) The finished product: ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/79b18afd-e4d7-4c8a-bf9d-260a2f5d2ac7_1280x720.png) ### **9. Organizing scattered thoughts, from [Helen Lee Kupp](https://www.linkedin.com/in/helenleekupp/)** > “I’m a mom who voice-records ideas during morning stroller walks, not a developer. The terminal interface? Overwhelming at first. The word ‘Code’ . . . but what if I don’t have a ‘coding project’? After 3 weeks of struggling to organize my scattered thoughts, I tried it anyway. And discovered something wild: Claude Code isn’t about coding at all. It’s about having an AI that manages your entire process—whatever the goal might be. > > How I use it: > > → Fed it rambling voice notes from stroller walks > → It organized them into coherent research themes > → Wrote a full article in *my* exact voice (pulled from my own examples!) > → Created LinkedIn versions automatically (this post is one of them!) > → Everything saved and ready to publish (including grabbing a screenshot of the template repository that I’m adding here!)” [[Here’s the repo](https://github.com/WomenDefiningAI/claudecode-writer)] ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/d39ecb30-620e-4df9-bba6-7dd9d2e47990_1724x1500.png) ### **10. Writing a job description, from [Justin Bleuel](https://x.com/JustinBleuel)** > “I used it to generate a full job description, hiring plan, interview plan, and rubric for a new role at Clay. > > I made a folder with internal Notions of our hiring material for PMs, added similar role JDs from other companies, then in planning mode asked to generate the same collateral for this new role.” ![Image](https://substack-post-media.s3.amazonaws.com/public/images/9392251d-0b64-45e5-9d46-05fa1c7ed491_1513x893.jpeg) ### **11. Synthesizing transcripts of calls with customers, from [Derek DeHart](https://www.linkedin.com/in/derekdehart/)** > “My current Claude Code jam is synthesizing transcripts of calls with customers to compile evidence that supports or invalidates a running tally of assumptions/requirements/hypotheses/whatever. Given MCPs to interact with other tools in our productivity stack—Fireflies, Linear, Notion, etc.—it’s become my hub for ongoing product research and development.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/14db451f-0dad-4236-b0da-efcef7ee88dd_1280x720.png) ### **12. Improving your writing, from [Teresa Torres](https://www.linkedin.com/in/teresatorres/)** > “I now write all of my content with Claude Code in VS Code. We iterate on an outline, it helps me improve the hook, it conducts research for me and adds citations to my outline, and it reviews and gives feedback on each section as I write. It has completely changed the way I write.” [[Much more on Teresa’s process here](https://www.producttalk.org/21-ways-to-use-ai-at-work/)] ![Screenshot of a writing workspace with three panels: a markdown article draft on the left, a list of AI use cases in the middle, and grammar and style feedback on the right.](https://substack-post-media.s3.amazonaws.com/public/images/07200b8d-a92d-473e-bb43-3743d22e5985_2000x752.png) ### **13. Working with audio files, from [Dan Heller](https://www.linkedin.com/in/thedanheller/)** > “I’m working with multiple audio files. I use Claude Code to manipulate them, convert the sample rates, rename them, and translate them from Portuguese to English.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/bab43e55-862f-48af-9428-ec49136af776_1280x739.png) ### **14. Creating “self-driving” documentation, from [James Pember](https://www.linkedin.com/in/jamespember/)** > “The most interesting use case we’re playing with is something I call ‘self-driving documentation.’ Basically, how can we give an Agent the responsibility of figuring out how/where our documentation can be better and more comprehensive. We’ve been experimenting with using Claude Code together with [Playwright](https://playwright.dev/) to automatically explore our software independently, identify knowledge gaps in our documentation, and then create those changes itself. Very promising!” [[More here](https://jamespember.substack.com/p/self-documenting-software)] ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/2001bcaa-90d7-48a0-84d7-79f21dbf01e8_2560x1706.png) ### **15. Creating a self-improving feedback loop, from** **[Gang Rui](https://www.linkedin.com/in/limgangrui/)** > “I created a slash command that analyzes my journal entries + Git commits (for the past 7 days; usually I use this weekly), spots gaps between what I said vs. did, and suggests system improvements. Like having a COO that learns from my patterns.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/61b9353c-e8a1-4c17-90a7-64fcb3e09112_1432x1774.png) ### **16. Getting inspiration from competitors’ ads, from [Sumant Subrahmanya](https://www.linkedin.com/in/sumantus/)** > “Extract ads from competitors to find the problem, use case, or copy/asset that’s working for them, and then repurpose it for my ads. Claude Code built out these scripts that would screen-grab all ads running on the ad library, and it’s super cool to watch it navigate to the browser and grab all screenshots in an almost ‘agentic’ way.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/2cb3d9d0-51d6-4c6b-9d59-654b87d25b97_1280x1385.png) ### **17. Automatically creating changelogs, from [Manik Aggarwal](https://www.linkedin.com/in/aggarwalmanik/)** > “I use Claude Code to create user-facing changelogs. I ask Claude Code to scan all commits from a specific time period, then pull in my changelog guidelines. It drafts a clean, structured changelog that usually needs few quick edits. What earlier took me hours is now down to 10–15 minutes. Most of our changelog output is created with Claude Code, with a final polish done by me.” ![Image from Everyone should be using Claude Code more](https://substack-post-media.s3.amazonaws.com/public/images/7e0b5a53-032a-4b0a-9737-d37cd4797527_2560x1596.png) ### **18. Building presentations, from [Hank Yeomans](https://x.com/HankYeomans)** > “All my slide work is done in Claude Code as HTML, then imported into PPT. > > First, I make a couple of slides that I need with some simple but well-prompted data, while also prompting to create them as html pages. Otherwise they might get created as SVGs or ‘slop AI images.’ I work with Claude Code making any adjustments. > > When I get it the way I want it, I ask Claude Code to make a template that I can use to ensure all future slides added are the same format, branding, look, and feel. > > This template becomes an .md file (I’ve put a snippet of that file below). With html you can be extremely specific—‘change this wording to say that’—and then you can use an MCP to view the ‘slides’ to show what you mean as well for greater context. With this, I have added several more slides as well as changed others using the template to keep things the same. Yes, hallucinations happen, that has to be accepted going in. Someone who knows what they want and can articulate what they want will not have too much trouble. > > When I want to put these into PPT, I just do well-thought-out screenshots. But honestly, If I’m using these for a customer or some presentation, I just open a browser and I have ‘previous’ and ‘next’ buttons built into them so that I can click through. Claude Code creates these as interactive html ‘slides’ so they react to the mouse.” ![Image](https://substack-post-media.s3.amazonaws.com/public/images/29355a05-9769-415a-b7ab-0257fbd13b6a_680x423.jpeg) ### **19. Doing social media research, from [Danny Shmueli](https://x.com/dannyshmueli)** > “Using Claude Code as a social media research assistant. I built 4 custom subagents that scan Reddit and X for developer pain points about AI coding tools. One command → finds relevant threads → drafts authentic replies. > > We have a marketing GitHub repo with subagents in .claude/agents/: > > - Reddit replier: tracks AI coding frustrations > - Reddit promoter: finds productivity/burnout posts > - X specialist: hunts Claude Code pain points > > Each knows exact keywords and communities to monitor.” ![Image](https://substack-post-media.s3.amazonaws.com/public/images/12966ce4-1aee-4754-8cc6-6fbdcafe7fbf_408x418.png) ### **20. Roadmapping, from [Abhi Chandwani](https://x.com/abhi_chandwani)** > “Because it has access to my repos, I am able to get a pretty good view on the impact vs. effort across the roadmap. Basically, hero story/press release thinking for new launches. > > I find it very useful, again because it is super rich in context. > > I run this through a repo that I have named {https://project.name}-product. See structure in the image below. Regularly Git push with verbose Git commit messages to create context for Claude to run some part of my decisioning workflows in future. > > > https://CLAUDE.md has paths to my engineering repos for architectural review, tradeoffs, and decisions. Broad structure below, and [here’s my prompt](https://x.com/abhi_chandwani/status/1960434310759993555).” ![A directory structure displayed in a file explorer. Folders named latent-product, bugs, execution, press-release, product-flywheel, user-research, and vision are visible. A file named CLAUDE.md is highlighted with a star icon next to it.](https://substack-post-media.s3.amazonaws.com/public/images/e23707e1-deef-44ca-a883-6933312a4068_678x370.png) ### **And all kinds of stuff:** 21. “Fully uninstalling Adobe products from their computer. Notoriously hard to do.” via [Jonathan Stiansen](https://www.linkedin.com/in/stiansen/) 22. “Automating Jira. I take a call transcript with my eng team, write up action items, create tickets in Jira using MCP—hands-off with the most annoying product in the world!” via [Terry Lin](https://x.com/itsmeterrylin) 23. “I use it to summarize recent tickets in Intercom and create bug reports in Asana, including the ticket link, user information, and steps to reproduce. I believe Claude could handle this directly, but CC feels easier to use.” via [Eren Gündüz](https://x.com/merengunduz) 24. “I use it for the majority of my PM tasks: custom commands for refining Linear issues, explaining features/codebase to me, writing release notes, discussing UX improvements, and polishing the UI. For good or bad, I talk to Claude more than anyone else. [Demo](https://share.cleanshot.com/mmxwTzgk).” via [Trist Adlington](https://x.com/trist_adlington) 25. “I just joined a startup as CPO, so to make conversations with my dev team the most productive, can ask questions about what our code can/can’t do before asking them.” via [Brandon Dorman](https://x.com/brandon_edu) 26. “I host a weekly figure-drawing event where we hire a model. I sent an email to my model list asking for a few dates when they are available, and let Claude create a schedule for the next few months by reading the replies. I give it access to Gmail like [this](https://code.dblock.org/2025/07/30/using-claude-code-with-google-sheets.html).” via [Daniel Doubrovkine](https://x.com/dblockdotorg) 27. “I have several Wi-Fi routers, and there were problems with connections and speed between them and my devices. Using MCP [Playwright](https://github.com/microsoft/playwright), it logged in and configured everything.” via [Chingis Alekenov](https://x.com/Alekenov) 28. “I made an audio recording of my doctor’s appointment. Took the audio, chunked, transcribed, and now I have an assistant using this as context.” via [Iliya Valchanov](https://www.linkedin.com/in/iliya-valchanov/) 29. “My daughter needed to present her artwork for her school application. I took photos of her work and asked [Warp](https://www.warp.dev/) to resize the photos and add them all to a PDF. Would have take me a lot of time to do this manually.” via [Melvin Vivas](https://x.com/donvito) 30. “I created ‘Claude CEO,’ which connects to my Gmail, Brex, Mercury, and Linear each day and gives me a comprehensive report on what I need to focus on each day.” via [Dennison Bertram](https://x.com/DennisonBertram) 31. “I used Claude Code as a *monster* research unit. Get it to parse arXiv or any other databases with reports and research papers. Compile thousands of papers to find a needle-in-a-haystack problem/solution. Find commonalities among papers, weird cross-connects, and new discoveries. Nerdy stuff, but I love it.” via [Troy Assoignon](https://x.com/troyassoignon) 32. “I use it to review my Slack conversations from the past day and identify conversations I participated in where I learned something or solved a problem and, turn that into an organic (personal-experience-based) social post that others can benefit from the learnings.” via [Chad Boyda](https://x.com/chadboyda) 33. “I’ve definitely used it to reverse-build documentation and requirements. We were building so fast on this side project, we were making decisions on the fly but losing the context and location of those decisions. I was crudely copy/pasting materials and conversations into Claude Code and leveraging its ability to reference the code itself to draft a PRD for better internal discoverability, decision logging, and source listing.” via [Megan Salisbury](https://www.linkedin.com/in/masalisbury/) 34. “I recently used Cursor (with Claude) to resize a 23MB GIF that was too large for most online converters. Also, formatting an NTFS SD card so my Mac could read it. Super cool.” via [Akilesh Bapu](https://www.linkedin.com/in/akileshbapu/) 35. “Storing all contracts and being able to then query these easily (e.g. which customers have X clause).” via [Pulkit Agrawal](https://www.linkedin.com/in/agrawalpulkit/) ### **For more, here are some of my favorite much-more-advanced use cases:** 36. [Refining your backlog in Linear](https://x.com/JucelyMarie/status/1960828753317339220) 37. [Managing user research, concepts, test results, and strategy](https://x.com/SamZoloth/status/1960418474779697326) 38. [Building a command center for your attention](https://x.com/raizamrtn/status/1960454186757448087) 39. [Turning product discovery notes into PRDs, and optimizing PRDs for coding agents](https://x.com/_HelenaT/status/1960448248088551892) 40. [Coming up with the first drafts of product messaging](https://www.linkedin.com/feed/update/urn:li:activity:7379238388391923713?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7379238388391923713%2C7379244841391554560%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287379244841391554560%2Curn%3Ali%3Aactivity%3A7379238388391923713%29) 41. [Running standups, sprint retros, and PRD reviews, and building a career OS](https://www.linkedin.com/feed/update/urn:li:activity:7379238388391923713/?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7379238388391923713%2C7379260143655342080%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287379260143655342080%2Curn%3Ali%3Aactivity%3A7379238388391923713%29) 42. [Running a branding studio](https://www.linkedin.com/feed/update/urn:li:activity:7379238388391923713?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7379238388391923713%2C7379536238078328832%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287379536238078328832%2Curn%3Ali%3Aactivity%3A7379238388391923713%29) 43. [Running your marketing](https://www.linkedin.com/feed/update/urn:li:activity:7379238388391923713?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7379238388391923713%2C7379244841391554560%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287379244841391554560%2Curn%3Ali%3Aactivity%3A7379238388391923713%29) 44. [Running your product](https://www.linkedin.com/feed/update/urn:li:activity:7379238388391923713?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7379238388391923713%2C7379259023898038272%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287379259023898038272%2Curn%3Ali%3Aactivity%3A7379238388391923713%29) 45. [Running your entire life](https://www.linkedin.com/feed/update/urn:li:activity:7379238388391923713?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7379238388391923713%2C7379382232454680577%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287379382232454680577%2Curn%3Ali%3Aactivity%3A7379238388391923713%29) Have you found a fun way to use Claude Code? Leave a comment! [Leave a comment](https://www.lennysnewsletter.com/p/everyone-should-be-using-claude-code/comments) *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 [1](#footnote-anchor-1) Claude Code does not actually run “locally” on your machine—the AI/LLM processing happens in the cloud—but you run the commands locally (directly on your machine), it has access to all of your local files and computer interfaces, and its specifically designed to feel “local.” --- ## [40/46] A builder’s guide to living a long and healthy life *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[My favorite AI and PM courses](https://maven.com/lenny) | [My favorite public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *Annual subscribers get a **free year** of 17 premium products: **[Devin, Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, and ChatPRD](https://www.lennysnewsletter.com/p/productpass)** (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** Since turning 40, I’ve felt a lot less invincible. For the first time in my life, my annual bloodwork results weren’t 100% healthy. A few months ago, I broke my pinkie toe on the edge of a wall. Last month, I sprained my wrist trying to adjust an A/C unit. Last week, I banged up my knee after slipping on a staircase. Then I took a toxins screen and learned I’m half-man, half-plastic: ![Image from A builder’s guide to living a long and healthy life](https://substack-post-media.s3.amazonaws.com/public/images/681aff87-f5a9-4981-9a96-d69c0624d98e_2044x484.png) So over the past year I’ve started getting serious about my health: tracking my nutrition, exercising, experimenting with supplements, focusing on my sleep, etc. Along that journey, I came across [Justin Mares](https://justinmares.substack.com/). Unlike the health folks everyone knows—Andrew Huberman, Rhonda Patrick, Bryan Johnson—[Justin](https://www.linkedin.com/in/justinmares/) is a full-time builder. He co-founded [Kettle & Fire](https://kettleandfire.com/), [Perfect Keto](https://perfectketo.com/), [Surely](https://www.drinksurely.com/), and now [Truemed](https://truemed.com/), and he’s basically a health nerd who spends hundreds of hours researching what to buy for himself and his family and shares what he learns in blog posts and tweets. When I decide what products to buy and which brands to trust, I’ve found myself *constantly* referencing his recommendations, more than anyone else’s. But I’ve always wanted more. So I pitched Justin on putting together a comprehensive and specific list of his favorite products and brands—the safest, least toxic, and highest-quality products he himself buys for clothing, sleep, food, toxin mitigation, and more. This is what you’ll find below. **And as a bonus, at the end of the post, we’ve compiled a handy bullet-list summary of every product and brand mentioned.** Note, except for a couple that Justin explicitly calls out, neither Justin nor I is an investor in any of these companies, and there are no affiliate deals involved. This is just a well-meaning post to help you all live a long, healthy life . . . so that you can keep building beloved products for many years to come. A huge thank-you to Justin for spending endless hours compiling this list. I will be referencing it *frequently*. *For more from Justin, check out his [newsletter](https://justinmares.substack.com/) and [Truemed](https://www.truemed.com/) (buy health products at a steep discount using your HSA). They’re also [hiring a marketer](https://www.truemed.com/careers)!* *You can also listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from A builder’s guide to living a long and healthy life](https://substack-post-media.s3.amazonaws.com/public/images/a5218dbd-ce54-4b6f-8c73-3340a862544c_1456x970.png) In 2011, I was working for my first company while also taking a full college course load. As a committed Tim Ferriss acolyte, I read a blog post about polyphasic sleep. For a whole month, I slept 3.5 hours a night and took a 20-minute nap every four hours, even if that meant leaving class or coming late to a fraternity party. This was insane. After a month of polyphasic sleep—and two crash-outs where I slept 18 hours in a row to recover—I realized it wasn’t for me. But what *was* for me, and what I’ve mostly stuck to in the 14 years since, is the paleo diet, which I decided to do after reading about it in the same blog post. I went paleo for two weeks my junior year of college (yes, I was the weird kid who would pass on the pizza and beer). And during this two-week experiment, my acne disappeared, I got leaner, slept better, and just had more energy. As it turns out, what you put in your body affects how you feel! If you’re not healthy, you’re not able to perform at your best, period. This is why so many in Silicon Valley are so obsessed with sleep tracking, diet, and other tools to help improve performance at work. Unfortunately, the West is in the [throes of a chronic disease crisis](https://justinmares.substack.com/p/the-great-american-poisoning). Americans today are the sickest population of humans to ever exist. Nearly 50% of adults have prediabetes or diabetes, 73% are obese or overweight, and the richest American men [live 15 years longer](https://jamanetwork.com/journals/jama/fullarticle/2513561?guestAccessKey=4023ce75-d0fb-44de-bb6c-8a10a30a6173) than the poorest, almost entirely due to chronic disease burden. Diabetes alone has an economic cost of around [$106 billion](https://diabetesjournals.org/care/article/47/1/26/153797/Economic-Costs-of-Diabetes-in-the-U-S-in-2022). In this post, I won’t cover traditional health tips (exercise, get eight hours of sleep, etc.), though they’re crucial. Instead, I want to cover some of the lesser-known health gotchas that consistently sap your health. I’ve spent the past decade of my career building companies in this world: first with [Kettle & Fire](https://kettleandfire.com/) (now a $100M+ annual revenue brand) and now with [Truemed](https://truemed.com/). And I’ve spent hundreds (thousands?) of hours searching for products that are the most effective both at driving real health outcomes and at avoiding many of the common toxins, microplastics, and other unhealthy compounds we encounter on a daily basis. Here are my most recommended products: ### Sleep Getting good sleep is among the most critical things you can do to improve your health and set yourself up for peak work performance. There’s a ton of literature on why sleep is so important, but for the purposes of this post I’ll assume you’re already a believer. In my view, the 80/20 of sleep health boils down to three things: 1. Exercise during the day 2. Get morning sunlight 3. Optimize your sleep setup (no external light, quiet, low CO2, cold) as much as possible Beyond the basics, I’ve found a few products that really help me get my best sleep. I love my [Eight Sleep](https://www.eightsleep.com/), and have found their pod works *incredibly* well to help me both increase my deep sleep and avoid wake-ups. For those who have trouble falling and staying asleep, magnesium [has been shown](https://pubmed.ncbi.nlm.nih.gov/39252819/) to radically improve sleep quality. I think the [Momentous magnesium L-threonate](https://www.livemomentous.com/products/magnesium-threonate) supplement is the best out there. If you’re really, really struggling to get to sleep, I have a few friends who swear by the peptide Epitalon . . . though it’s technically a research chemical, so I can’t point you anywhere to get it. Ask your doctor. Or buy it illegally online, but beware of sourcing! Studies have found that [mattresses can be a source of harmful chemicals](https://www.consumerreports.org/babies-kids/childrens-health/mattresses-can-be-source-of-harmful-chemicals-in-kids-rooms-a5263703680/) (especially foam ones), as they emit phthalates, benzophenones, and other compounds that have been linked to asthma, developmental issues, and reproductive harm. Sadly, many of these chemicals have not been thoroughly tested (thanks to [the insane way we regulate chemicals in the U.S.](https://justinmares.substack.com/p/our-insane-approach-to-regulating)), but the few studies we do have are concerning. You may or may not know this, but our current approach to chemical regulation relies on industries to run their own tests and make their own assertions that the novel, never-before-seen chemicals they’re inventing and putting in our products are safe and sound. When a pharma company invents a drug that humans take, it goes through a rigorous FDA approval process that takes a decade and is tested for safety. When a chemical company invents a compound that your body can’t break down and ends up in food, the water supply, and our bodies, regulators require . . . nearly nothing. How does this make sense? It gets worse! That’s our approach to *new* chemicals. But for the chemicals there *have* been safety testing for—the thousands that we know are harmful at some dose but where the dose makes the poison—how are exposure limits determined? To determine exposure limits, scientists follow a relatively simple process. First, they’ll expose a lab animal, often a rat, to small amounts of said chemical. They’ll then sit and watch for behavioral changes: Does the rat slow down, look like it’s in pain, or start wobbling a bit? They then continue to increase the dosing of the suspected toxin until they observe behavioral changes in the rat. At that point, they take the dose they exposed the rat to and multiply it by a “safety factor” (usually 100) to account for the difference in weight between rats and humans. And that’s it. Anything below the dose number that’s spit out is now considered . . . safe! To ensure you’re not inhaling toxic compounds that mattresses often off-gas (like VOCs, benzene, and formaldehyde) while sleeping, I highly recommend the [Woolshire pillow](https://thewoolshire.com/products/wool-pillow), the [Avocado](https://www.avocadogreenmattress.com) mattress and pillows, and [Coyuchi sheets](https://www.coyuchi.com/collections/sheet-sets). Besides Avocado, I have heard good things about [Essentia](https://myessentia.com/) and [Naturepedic](https://www.naturepedic.com/). And lastly, if you have $80,000 burning a hole in your pocket, I’m sure you can’t go wrong buying [Hästens](https://hastens.com/us/) (though I wouldn’t know). Apparently, Drake bought eight of these mattresses for his Toronto home. ### Clothing Your skin is the largest organ in your body, and it consistently absorbs compounds and chemicals that it’s exposed to. Yet we rarely think about what’s in the clothes we spend nearly all our time wearing. Even though many clothes use dyes and chemicals known to be hazardous, there’s no such thing as a nutrition label, or even a disclaimer, as to what might be in the clothes you put on your body. Turning fibers into clothing is a complicated and chemically intensive process. Thousands of chemicals are used to make clothing, some 10% of which have been shown to [disrupt the immune system, increase cancer risk, mess with hormones, and create reproductive issues](https://www.kemi.se/en/publications/reports/2014/report-6-14-chemicals-in-textiles). Even worse, the fun, stretchy performance fabrics you love to wear probably have the highest level of chemical treatment, and heavily utilize PFAS (forever chemicals) and PBDEs (chemicals used as flame retardants). This could be one reason why PFAS are [found in 97%](https://www.atsdr.cdc.gov/pfas/data-research/facts-stats/index.html) of Americans’ blood. It’s unfortunately not just the chemicals, dyes, and other additives that make your clothing choices important for your health. Extremely common materials like polyester can [lower sperm count](https://www.eviemagazine.com/post/polyester-underwear-could-be-causing-widespread-infertility-in-men), yet it’s by far the most popular material used in underwear. These fabrics, and the dyes and chemicals on them, can fairly easily enter the body, especially in areas where your skin is most permeable: your nether regions, feet, and armpits. I strongly recommend *at minimum* wearing toxin-free, organic cotton or linen socks and underwear. My favorite brands are [Industry of All Nations](https://industryofallnations.com/collections/underwear) and [Pact](https://wearpact.com/) for underwear ([Nads](https://nadsunder.com/) as runner-up), and Pact for socks. For shirts, I like [Paka](https://www.pakaapparel.com/)’s stuff, especially their T-shirts, as they use a blend of 85% organic cotton and 15% alpaca fibers. [Faherty](https://fahertybrand.com/) has a surprisingly good selection of organic tees that I’m into. And lastly—though I probably shouldn’t share my secrets—[Wax London](https://waxlondon.com/collections/organic-cotton) is my go-to fashionable brand that has tons of organic options. ### Food This is the area I feel most passionate about. The Standard American Diet (SAD) is terrible. Right now, many Americans get 55% of their calories from [ultraprocessed foods](https://www.npr.org/sections/shots-health-news/2025/08/07/nx-s1-5495308/ultra-processed-food-upf-rfk-cdc). The typical American diet is often bereft of nutrients, is riddled with toxins, and has created the sickest population of humans to ever exist on this planet 🇺🇲. In large part, this is because we have a food system that is uniquely permissive in what it allows in our foods. Worse, international companies (like Mondelez) actually have *American versions of the same product*, where the American version is simply more processed! ![Image from A builder’s guide to living a long and healthy life](https://substack-post-media.s3.amazonaws.com/public/images/870283f5-4fde-45f4-981e-5f11a90fce77_768x679.png) When it comes to sourcing foods, the biggest things to optimize for are (1) nutrient density and (2) toxin avoidance. The most nutrient-dense foods are sourced more locally, which inherently means eating seasonally. The farther your food travels, the less nutrient-dense it is. One study found that after nine days of travel, spinach was [90% less nutrient-dense](https://www.chicagotribune.com/dining/ct-xpm-2013-07-10-chi-most-produce-loses-30-percent-of-nutrients-three-days-after-harvest-20130710-story.html) than it was at harvest. Here in Austin, I am fortunate to live near an amazing grocery store ([Radius](https://www.eatradius.com/)), but many major cities have excellent farmers markets or grocery stores to buy local. Beyond nutrient density, it’s important to mitigate common toxins like pesticides (glyphosate, atrazine), phthalates, microplastics (though it’s nearly impossible), and PFAS. That means buying organic where possible—especially the most-sprayed fruits and vegetables, called the “[dirty dozen](https://www.ewg.org/foodnews/dirty-dozen.php)”—and, again, buying whole, unprocessed foods that require fewer chemicals to ensure some amount of shelf life. There are certifications (like glyphosate-free) that are fairly good signs that you’re buying from a brand that cares about sourcing, but these are few and far between. Beyond those principles, there are a few foods from companies I feel comfortable recommending: 1. **Protein bars/snacks:** I absolutely love these [Maui Nui meat sticks](https://mauinuivenison.com/products/snack-membership?Title=Default+Title). Maui Nui has the best-sourced meat I am aware of, anywhere in the world. (We partnered with them at Kettle & Fire.) They actually hunt and harvest the animals they make the jerky from, and it shows in the taste and nutrient density of their meat sticks. It should be no surprise, but when animals live a longer time, forage, exercise, and are outside for their natural lifetimes, well, they tend to be more nutrient-dense and protein-rich: exactly what Maui Nui products are. I prefer these sticks to a protein bar most of the time. But if I’m running low, I absolutely love the [Jacob bar](https://eatjacob.com/) in a pinch. Just be sure to [avoid David bar](https://x.com/jwmares/status/1904218913413672988). 😉 2. **Bone broth:** You probably knew I would say this, but I drink at least a carton of [Kettle & Fire](https://kettleandfire.com) bone broth each day. We made the first [wild-harvested bone broth](https://www.kettleandfire.com/pages/mauinui) (with Maui Nui), which is my go-to. But all our flavors are good! 3. **Beef:** People should be eating far more organ meats, because they’re such a great source of key amino acids and other nutrients. I love Force of Nature and their [ancestral blend](https://forceofnature.com/products/regenerative-beef-ancestral-blend), which makes it easy to get organ meats in your diet while also buying some of the highest-quality regenerative beef out there. [White Oak Pastures](https://whiteoakpastures.com/) also does an incredible job from a sourcing standpoint. 4. **Coffee:** This is a tough one to source locally. I buy a lot of coffee from the [Groundwork](https://www.groundworkcoffee.com/collections/regenerative-organic-certified) team, who have worked hard to build one of the only regenerative coffee supply chains in the world. Bonus points if you want to stir in some [$400 manuka honey](https://manukahoneyofnz.com/products/mgo-1500-manuka-honey-umf-28?currency=usd) (which is apparently lightly psychoactive, though I haven’t personally tried it). At this point in my life, there are very few national food brands that I support, as I tend to prefer buying locally and seasonally where possible. Scaling a food business while maintaining high standards is extremely difficult, and, to be frank, most companies out there don’t do a good job. Even Whole Foods’s beef [has a high degree of phthalates!](https://www.eatradius.com/beef-plastic-testing/) ### Supplements Many supplements are a waste of time and money. Much (most?) of what you’ll find on Amazon is probably fake, overhyped, or doesn’t have enough effective ingredients to do anything. A recent consumer test found that 4 of 6 creatine gummy brands [contained no creatine](https://www.wired.com/story/creatine-gummies-dubious-claims/) at all! For almost everyone, getting core nutrients and amino acids from whole food sources is *far* superior to getting them from supplements. That said, many folks are still missing some of the building blocks the body requires to thrive. That’s where supplements can come in, to . . . *supplement* the diet. Some I like and recommend: 1. **Protein powder:** Protein is key, and most Americans don’t get anywhere close to the recommended 0.36 grams per pound of body weight. For those looking to increase protein intake, [Equip](https://equipfoods.com/) is probably my favorite protein powder, with the [Momentous whey](https://www.livemomentous.com/products/grass-fed-whey-protein-isolate-powder-limited-edition-flavors) a close second. I would recommend avoiding plant protein powders, as they (1) basically don’t work and are [not bioavailable](https://systematicreviewsjournal.biomedcentral.com/articles/10.1186/s13643-022-01951-2#:~:text=Proteins%20also%20inherently%20differ%20in,PDCAAS%20for%20different%20protein%20sources.), and (2) are often made from pea or soybean protein, which is among the most pesticide-sprayed crops out there. For example, a [Mamavation test](https://mamavation.com/food/pea-protein.html) found Orgain’s organic pea protein powder had glyphosate levels roughly *1,000 times* higher than recommended. Beyond protein for building muscle, Americans also criminally underconsume collagen and gelatin, both of which are key building blocks for healthy joints, skin, hair, and gut. I have four or five servings of collagen per week (on top of my daily bone broth) and recommend the collagen from Equip, Momentous, [Lineage](https://lineageprovisions.com/products/creatine-monohydrate), and a company I started years ago, [Perfect Keto](https://shop.perfectketo.com/products/grass-fed-collagen-with-mct) (if you want the extra boost that comes with coconut fats). 2. **Creatine:** Creatine is a naturally occurring amino acid stored in muscles and the brain that helps with energy creation. It’s been safely used by bodybuilders since time immemorial and is increasingly associated with a host of [mental performance benefits](https://www.nature.com/articles/s41598-024-54249-9). Pretty much the only creatine I would buy is [Momentous](https://www.livemomentous.com/products/creatine-monohydrate), as they are one of the only non-Chinese-sourced brands out there. 3. **Nattokinase:** This is a new compound that I am very bullish on. Heart disease is the leading cause of death in the U.S., and 70% to 80% of heart attacks are caused by atherosclerotic plaque. Certain dosages of nattokinase (at least 10,000 fibrinolytic units) were found to be more effective than statins at reducing arterial plaque, a significant contributor to [heart](https://pmc.ncbi.nlm.nih.gov/articles/PMC9441630/) [disease](https://pubmed.ncbi.nlm.nih.gov/28763875/#full-view-affiliation-2). For the many folks whose labs show high cholesterol or concerning ApoB/Lp(a) markers, I think taking nattokinase preventively is a good idea. I’ve been taking [Toku](https://tokuhealth.com/), which has the clinically effective dose plus vitamin K2 (I also invested in the company). A friend [found recently](https://x.com/jwmares/status/1962904842281893898) that his ApoB dropped 16% and his LDL came down 20% after 60 days of taking Toku. Josh (the founder of Levels) [found similarly](https://x.com/joshuasforrest/status/1973062469036683487)! 4. **B vitamins and MTHFR:** If you’re like me—and 30% of other people—you may have some variety of the MTHFR mutation. This mutation impairs the body’s ability to detox and can lead to higher levels of homocysteine, an amino acid in the blood that, at elevated levels, is linked to increased risks of heart disease, blood clots, strokes, and cognitive decline. I take this [B complex](https://www.amazon.com/Thorne-Methyl-Guard-Plus-methylation-homocysteine/dp/B00O5AHC4S/) from Thorne daily to help with methylation (basically, how your body clears toxins) and give my body support to detox and function the way it should. 5. **Organ supplementation:** As I’ve mentioned, organ meats and connective tissue contain key amino acids and nutrients that are not found in muscle meats like steak or chicken breast. Unfortunately, most Americans primarily eat cuts of muscle meats, and thus don’t get many key amino acids in their diets. I get it: eating straight kidney and liver can be gross! Many people (aka me) prefer supplementing with organs in capsule form. My favorite [organ supplement](https://mauinuivenison.com/products/organ-blend-supplement) is from Maui Nui (a partner of my company Kettle & Fire), which I believe sources the best and most ethical animal protein in the world. Outside of those daily staples, I have a whole host of supplements I’ll take for specific life events, often from brands like Momentous or Thorne, which both have good sourcing practices. For example, when I’m sick, I follow the [Huberman protocol](https://www.hubermanlab.com/episode/how-to-prevent-treat-colds-flu) and take 600 to 900 milligrams of N-acetylcysteine (NAC), along with 100 mg of zinc, per day for three to five days. I take [AHCC](https://www.amazon.com/Premium-Kinoko-Platinum-AHCC-Supplement/dp/B00HG0YZMG/ref=sr_1_6_pp) and [astaxanthin](https://www.amazon.com/Pure-Synergy-SuperPure%C2%AE-Astaxanthin-Antioxidant/dp/B07KX3NZJ5/ref=sr_1_6_pp?dib=eyJ2IjoiMSJ9.RCID4tv3DFCQwIWFVzZA_22htPLJuRaXepz1XeDNiKAgy4p5vfIKZP6sGyNJD5hFJXqvQFTOpy8C_lEj7Oo0a0UfggDsEyQ7TSMQ_tYJ4dksab1_BZZdmSa5kIk_wybWHDd1Wg9HbjDWnDhfs_PDMBpnpWz1kyUbCNEccx4DaY4yq6PPyrbKqvAl1RsNhMjPrIrBepiFrNSYfslF_yqP2GMq9lexbpzkrExnOHYBvS7PihMY6w4kulumQ-hkYGLPd_t6FqTnjFdRv9PnnzgaRgVln-7t8jmbTEofVpm5Ru4.mtPHiGccPFeIAq40eQTDxzHJ-_glHBPnZDeNow4n-h4&dib_tag=se&keywords=astaxanthin&qid=1758035441&sr=8-6) to help my body fight off the infection. I also mega-dose vitamin C and try to spend at least 30 minutes in the sauna. Though this regimen seems to help, nothing is better than not getting sick at all. That’s why whenever I fly, I use [Viraldine](https://www.amazon.com/VIRALDINE-Povidone-Iodine-Designed-Congestion-applications/dp/B0CTWWS1DK) before getting on the plane to prevent catching anything (including feelings). ### Toxin mitigation In my view, environmental toxins are the trans fats of our generation. [PFAS](https://www.vox.com/2022/8/25/23318667/pfas-forever-chemicals-safety-drinking-water), phthalates, endocrine-disrupting chemicals, pesticides, and heavy metals are everywhere. They’re in your clothing, detergents, soap, shampoo, paint, furniture, water, and air. As of this writing, 92% of Americans have measurable phthalates in their body and 97% have PFAS in their blood. These chemicals have been shown to affect [testosterone levels](https://ehp.niehs.nih.gov/doi/10.1289/ehp.1205118), [anxiety](https://www.sciencedirect.com/science/article/pii/S0160412020318493#:~:text=Experimental%20data%20show%20that%20early,Gray%20and%20McNaughton%2C%202003%29.), [cancer risk](https://www.epa.gov/pfas/our-current-understanding-human-health-and-environmental-risks-pfas), and even [sperm count](https://pmc.ncbi.nlm.nih.gov/articles/PMC9986484/). [Some researchers believe](https://www.nytimes.com/2022/05/19/science/early-puberty-medical-reason.html) that they even have a role to play in the fact that girls are hitting puberty one to two years sooner than they were 40 years ago. In large part, we are here because the FDA has totally dropped the ball on regulating new chemicals. The European Union, on the other hand, takes a “do no harm” approach and assumes that new chemicals may cause harm. So they regulate them! The EU currently bans more than 1,300 chemicals and compounds that the U.S. allows. The FDA takes the opposite approach: Sure, humans have never before been exposed to these classes of chemicals, but whatever—let’s assume they’re safe. The U.S. has over 40,000 chemicals allowed in our food, water, and environment, with minimal pushback from the FDA. Between this and the 50 years it took the FDA to act on [the whole trans fats thing](https://justinmares.com/why-im-ignoring-the-nutrition-experts/) (and then, only after multiple lawsuits were brought against them), I think it’s safe to say that the FDA’s approach to chemical regulation has lots of room for improvement. Anyway, back to toxins. They’re everywhere (in breast milk, our blood, our urine), have tremendous health impacts, and are highly likely causing many of the chronic conditions that plague Americans today. Detoxing your environment is critically important but almost impossible to do completely. That’s why I think an 80/20 approach makes a lot of sense here: cut out toxins in the things you’re regularly exposed to, and use detox tools (like sauna, or fancy new blood cleaning approaches like [Inuspheresis](https://www.inuspheresis.com/)/[Proxima](https://www.proxima.health/)) to attack the remaining 20%. For many, this means cleaning up your home environment, your office environment, and your kitchen. Most of what I’ve learned about this I learned after having the [Lightwork](https://lightworkhome.com/) (think: functional medicine doctor for your house) team do a thorough test and review of my home. Here are the things I’d buy to mitigate toxin exposure in all of those spaces: 1. **Air purifiers:** Most microplastics are inhaled, not ingested via water or food. On the high end, I like the [Jaspr](https://jaspr.co/), but if you’re looking for a great everyday purifier, the [PuroAir](https://getpuroair.com/) is also quite good. 2. **Water purifier:** Our drinking water is riddled with PFAS, glyphosate, and many other toxins. [A recent study](https://www.ewg.org/news-insights/news-release/2025/04/cancer-causing-chemicals-drinking-water-put-122m-americans-risk) found that 35% of Americans live in areas where their water exceeds the limit for one or more carcinogens, and that [half of all water](https://www.sciencedirect.com/science/article/pii/S0160412023003069?via%3Dihub=) is contaminated with PFAS! To filter out as much as possible, I can’t recommend the [Rorra](https://rorra.com/products/countertop-system) enough (I even invested in it). It’s the best countertop system available, and has the best showerhead filter to boot. If you’re looking for a whole home system (which I have), the [Aquasana](https://www.aquasana.com/) is excellent—though if you’re in the market for a system that will have you going, “I cannot believe how much I just spent on this,” the [Ophora system](https://www.ophorawater.com/whole-home/) is incredible. 3. **Pesticide mitigation:** I am becoming more and more concerned about pesticides. They’re everywhere, and by far the best thing you can do is improve your food sourcing by buying organic, local, and seasonal wherever possible. To the extent that that’s not an option, I use [this cleaner](https://www.amazon.com/Renovera-Vegetable-Infant-Friendly-Powdered-Produce/dp/B07R3W8H41/ref=sr_1_6?dib=eyJ2IjoiMSJ9.7ckRjZuxqNsmcIyStMrrXpAzOIRLzBn5pK-NheGfpCW3tOfJyHpjBeUUCNV7yBLN1WgrRlm2XJo_HjN8nB7Fwh89tZOqwy0FqX6badqwutdzP_955CAkcfJxKbgosqeI8mN8PK-kYMzjrVn8ZRJVxIzX1I2pk4vd3OXFmRQmysJMhCUuSK8Mu6_lgmKI62YbVnggH6GxTsjS0bmlJrmnOBRaashycwL8zNBp2g_5OvuoM7AbY5NTIzPtPf1mYtT2RmgdZ5GTOdxzDu2c4q57wmbobh4oa6VoNisjINIVVls.GwBbqY-rk4UyrUW2xYSYDMssqD3bGtNWzRvvdwi8ea0&dib_tag=se&keywords=pesticide%2Bremoval&qid=1754577982&sr=8-6&th=1) on much of my produce, and am going to start experimenting with one of [these crazy devices](https://www.amazon.com/Powerscale-Generator-Vegetable-Multipurpose-Pesticides/dp/B0D8DM18HV/ref=sr_1_17) that removes much of the pesticide residue found on fruits and veggies. 4. **Detoxing your kitchen:** This is crucial but can be relatively simple. Most of the compounds you’re looking to remove are plastic items that will shed plastics and endocrine-disrupting chemicals like BPA, BPS, and phthalates. Remove plastic storage containers, plastic wrap, plastic cutting boards, plastic-coated pans, and so on, as a first step. I’d recommend replacing them with glass containers and ceramic pans ([Caraway](https://www.carawayhome.com/) is great for both), [beeswax wraps](https://www.amazon.com/Bees-Wrap-Assorted-Sustainable-Alternative/dp/B0126LMDFK/ref=sr_1_7?dib=eyJ2IjoiMSJ9.-6b6IdDUrHXN828SJ9DA6ti-e8Dl293N7a5T7Lj0_lGJdlOYOLcjmajLiRvwRm4a8FJFP0XdODb0iXLJl3UZLPlMVjn6ohUxVif3AZ2JmoPQQu86u4PIobVDvZv0_iDaHjEypAX-gjUsEJIIFbtKv1e7uIr6dzpUhsQK6gmqip5MFSYl6xmAzth3DP4pm8EOFvVlW45szdnJxLT4hwYTHHKBH7PLNrW70OinDQtTHA5634kAWPptxIZEd61Z7rTGLQwfSRgiNh0tL3rCGUasx_X7sXNKIZnyYCj3ax7RszI.IU6yXYPwmgZi-founYFc2aw1BdZd3AEkrLW8ssVkSlM&dib_tag=se&keywords=beeswax+wrap&qid=1754578385&sr=8-7), and a [teakwood](https://teakhaus.com/collections/cutting-boards) cutting board (no microplastics). A bit further off the beaten path, I increasingly notice that I feel a *lot* better when I’m not in lights that emit a lot of blue light or have a very high flicker rate. There’s a simple way to see just how often your lights are flickering: find a light in your home or office, and film it in slo-mo mode using your phone. Then watch the video—that flicker rate is happening all the time, stressing you out, all just below your conscious perception. I’ve replaced (first just a few, and now all) the bulbs in my house with [these circadian bulbs](https://thehealthyhome.shop/collections/lights). If you’re skeptical of the impact, try replacing one lamp with a circadian light bulb, and you’ll see a huge difference. I have people remark on how relaxed my house feels *all the time* when they walk in in the evening. If you want to go really deep on the topic, [this post on light](https://andybromberg.com/all-light) is exceptional. But trust me, once you start removing blue light from your environment in the evenings, you’ll never go back. Lastly, there are electromagnetic fields (think Wi-Fi signals, Bluetooth, cellphone radiation, power infrastructure). Andy Bromberg of [Lightwork](https://www.lightworkhome.com/) has another banger post on the [skeptic’s guide to EMFs](https://andybromberg.com/rational-guide-to-emfs). I think EMFs are a bigger deal than society thinks, but likely less harmful than many of the tinfoil-hat crowd believes. Still, they’re very much worth mitigating, which is simple to do: 80% of the battle is moving Wi-Fi routers away from your bed and workstations. To get to 90%, turn your Wi-Fi off at night (using a simple [smart plug](https://www.amazon.com/Govee-WiFi-Outlet/dp/B08731J1L4/ref=sr_1_6)), and you’ll mitigate a huge amount of potential exposure. ### Everyday items A few everyday items I enjoy that I think are worth having: 1. **Deodorant:** I’ve used and loved [Primally Pure](https://primallypure.com/collections/deodorant) for years now, though today there are plenty of good options that are free of heavy metals and actually effective. [Native](https://www.nativecos.com) is also decent. 2. **Blue-light blockers:** I dig the [Ra Optics](https://raoptics.com/) blue-light blockers for nighttime wear, or to wear if I’m staring at a screen for many hours. [Chroma](https://getchroma.co/collections/blue-light-blocking-glasses) also makes excellent blue-light products, though I find them slightly less fashionable and comfortable to wear. 3. **Soap/shampoo:** You can’t go wrong with [Dr. Bronner’s](https://www.drbronner.com/). 4. **Cleaning products:** [Branch Basics](https://branchbasics.com/) is the best by a mile. Huge fan. Skin-care-wise, I tend to like a few products but not go crazy. On the daily, I use [Mirror Skin](https://mymirrorskin.com/) as a base layer and then add a [mineral sunscreen](https://karigran.com/products/essential-spf) (when I remember, which is not always 😬). I also like [this tallow balm](https://www.shwallyhome.com/collections/our-balms/products/shwally-face-and-body-balm), although I can’t say I have seen amazing results or anything from it. For washing my face, I use the [Meteorite scrub](https://alitura.com/collections/skincare/products/the-meteorite-facial-scrub?selling_plan=4148527355) fairly often, as well as [this cleanser](https://alitura.com/collections/skincare/products/alitura-pearl-cleanser) on the daily. — Sadly, staying healthy in our current environment is hard. I am hopeful this changes in the coming decades, but until it does, you can take concrete steps to avoid the poor average health outcomes in the U.S. I know it’s helped me: since making many of these changes, I’ve had more energy and been better able to handle the startup journey. I hope this comprehensive list of everything I’ve learned over the years is helpful in your journey toward health. # Handy summary of every product/brand mentioned above #### Sleep 1. [Eight Sleep](https://www.eightsleep.com/) 2. [Momentous magnesium L-threonate](https://www.livemomentous.com/products/magnesium-threonate) 3. [Woolshire pillow](https://thewoolshire.com/products/wool-pillow) 4. [Coyuchi sheets](https://www.coyuchi.com/collections/sheet-sets) 5. [Avocado mattress](https://www.avocadogreenmattress.com) 6. [Essentia mattress](https://myessentia.com/) 7. [Naturepedic mattress](https://www.naturepedic.com/) 8. [Hästens](https://hastens.com/us/) #### Clothing 1. [Industry of All Nations](https://industryofallnations.com/collections/underwear) 2. [Pact](https://wearpact.com/) 3. [Nads](https://nadsunder.com/) 4. [Paka](https://www.pakaapparel.com/) 5. [Faherty](https://fahertybrand.com/) 6. [Wax London](https://waxlondon.com/collections/organic-cotton) ### Food 1. [Maui Nui meat sticks](https://mauinuivenison.com/products/snack-membership?Title=Default+Title) 2. [Jacob protein bar](https://eatjacob.com/) 3. [Kettle & Fire bone broth](https://kettleandfire.com) 4. [Force of Nature beef](https://forceofnature.com/products/regenerative-beef-ancestral-blend) 5. [White Oak Pastures meats](https://whiteoakpastures.com/) 6. [Groundwork coffee](https://www.groundworkcoffee.com/collections/regenerative-organic-certified) 7. [Radius](https://www.eatradius.com/) #### Supplements 1. [Equip protein](https://equipfoods.com/) 2. [Momentous](https://www.livemomentous.com/products/creatine-monohydrate) 3. [Lineage](https://lineageprovisions.com/products/creatine-monohydrate) 4. [Perfect Keto](https://shop.perfectketo.com/products/grass-fed-collagen-with-mct) 5. [Toku](https://tokuhealth.com/) 6. [Thorne B complex](https://www.amazon.com/Thorne-Methyl-Guard-Plus-methylation-homocysteine/dp/B00O5AHC4S/) 7. [Maui Nui Organ Blend](https://mauinuivenison.com/products/organ-blend-supplement) 8. [AHCC](https://www.amazon.com/Premium-Kinoko-Platinum-AHCC-Supplement/dp/B00HG0YZMG/ref=sr_1_6_pp) 9. [Astaxanthin](https://www.amazon.com/Pure-Synergy-SuperPure%C2%AE-Astaxanthin-Antioxidant/dp/B07KX3NZJ5/ref=sr_1_6_pp?dib=eyJ2IjoiMSJ9.RCID4tv3DFCQwIWFVzZA_22htPLJuRaXepz1XeDNiKAgy4p5vfIKZP6sGyNJD5hFJXqvQFTOpy8C_lEj7Oo0a0UfggDsEyQ7TSMQ_tYJ4dksab1_BZZdmSa5kIk_wybWHDd1Wg9HbjDWnDhfs_PDMBpnpWz1kyUbCNEccx4DaY4yq6PPyrbKqvAl1RsNhMjPrIrBepiFrNSYfslF_yqP2GMq9lexbpzkrExnOHYBvS7PihMY6w4kulumQ-hkYGLPd_t6FqTnjFdRv9PnnzgaRgVln-7t8jmbTEofVpm5Ru4.mtPHiGccPFeIAq40eQTDxzHJ-_glHBPnZDeNow4n-h4&dib_tag=se&keywords=astaxanthin&qid=1758035441&sr=8-6) 10. [Viraldine](https://www.amazon.com/VIRALDINE-Povidone-Iodine-Designed-Congestion-applications/dp/B0CTWWS1DK) ### Toxin mitigation 1. [Inuspheresis](https://www.inuspheresis.com/) 2. [Proxima](https://www.proxima.health/) 3. [Lightwork](https://lightworkhome.com/) 4. [Jaspr](https://jaspr.co/) 5. [PuroAir](https://getpuroair.com/) 6. [Rorra](https://rorra.com/products/countertop-system) 7. [Aquasana](https://www.aquasana.com/) 8. [Ophora](https://www.ophorawater.com/whole-home/) 9. [Renovera](https://www.amazon.com/Renovera-Vegetable-Infant-Friendly-Powdered-Produce/dp/B07R3W8H41/ref=sr_1_6?dib=eyJ2IjoiMSJ9.7ckRjZuxqNsmcIyStMrrXpAzOIRLzBn5pK-NheGfpCW3tOfJyHpjBeUUCNV7yBLN1WgrRlm2XJo_HjN8nB7Fwh89tZOqwy0FqX6badqwutdzP_955CAkcfJxKbgosqeI8mN8PK-kYMzjrVn8ZRJVxIzX1I2pk4vd3OXFmRQmysJMhCUuSK8Mu6_lgmKI62YbVnggH6GxTsjS0bmlJrmnOBRaashycwL8zNBp2g_5OvuoM7AbY5NTIzPtPf1mYtT2RmgdZ5GTOdxzDu2c4q57wmbobh4oa6VoNisjINIVVls.GwBbqY-rk4UyrUW2xYSYDMssqD3bGtNWzRvvdwi8ea0&dib_tag=se&keywords=pesticide%2Bremoval&qid=1754577982&sr=8-6&th=1) 10. [PowerScale](https://www.amazon.com/Powerscale-Generator-Vegetable-Multipurpose-Pesticides/dp/B0D8DM18HV/ref=sr_1_17) 11. [Caraway](https://www.carawayhome.com/) 12. [Beeswax wraps](https://www.amazon.com/Bees-Wrap-Assorted-Sustainable-Alternative/dp/B0126LMDFK/ref=sr_1_7?dib=eyJ2IjoiMSJ9.-6b6IdDUrHXN828SJ9DA6ti-e8Dl293N7a5T7Lj0_lGJdlOYOLcjmajLiRvwRm4a8FJFP0XdODb0iXLJl3UZLPlMVjn6ohUxVif3AZ2JmoPQQu86u4PIobVDvZv0_iDaHjEypAX-gjUsEJIIFbtKv1e7uIr6dzpUhsQK6gmqip5MFSYl6xmAzth3DP4pm8EOFvVlW45szdnJxLT4hwYTHHKBH7PLNrW70OinDQtTHA5634kAWPptxIZEd61Z7rTGLQwfSRgiNh0tL3rCGUasx_X7sXNKIZnyYCj3ax7RszI.IU6yXYPwmgZi-founYFc2aw1BdZd3AEkrLW8ssVkSlM&dib_tag=se&keywords=beeswax+wrap&qid=1754578385&sr=8-7) 13. [Teakhaus](https://teakhaus.com/collections/cutting-boards) 14. [Circadian bulbs](https://thehealthyhome.shop/collections/lights) 15. [Smart plug](https://www.amazon.com/Govee-WiFi-Outlet/dp/B08731J1L4/ref=sr_1_6) #### Everyday items 1. [Primally Pure deodorant](https://primallypure.com/collections/deodorant) 2. [Native deodorant](https://www.nativecos.com) 3. [Ra Optics blue-light glasses](https://raoptics.com/) 4. [Chroma blue-light glasses](https://getchroma.co/collections/blue-light-blocking-glasses) 5. [Dr. Bronner’s soaps](https://www.drbronner.com/) 6. [Branch Basics cleansers](https://branchbasics.com/) 7. [Mirror Skin serum](https://mymirrorskin.com/) 8. [Kari Gran sunscreen](https://karigran.com/products/essential-spf) 9. [Shwally tallow balm](https://www.shwallyhome.com/collections/our-balms/products/shwally-face-and-body-balm) 10. [Alitura Meteorite scrub](https://alitura.com/collections/skincare/products/the-meteorite-facial-scrub?selling_plan=4148527355) 11. [Alitura Pearl cleanser](https://alitura.com/collections/skincare/products/alitura-pearl-cleanser) *Thanks, Justin!* *For more from [Justin](https://www.linkedin.com/in/justinmares/), check out his [newsletter](https://justinmares.substack.com/) and [Truemed](https://www.truemed.com/). They’re also [hiring a marketer](https://www.truemed.com/careers)!* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [41/46] Part 2 of how to get the most out of your product pass—and welcome, Stripe Atlas, to the bundle! *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[AI/PM courses](https://maven.com/lenny) | [Public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** ![Image from Part 2 of how to get the most out of your product pass—and welcome, Stripe](https://substack-post-media.s3.amazonaws.com/public/images/1585f86b-cd3d-45c9-b983-9008312a87f0_1457x971.png) First of all, some exciting news: **Starting today, [Insider subscribers](https://www.lennysnewsletter.com/subscribe?plan=founding) get 40% off [Stripe Atlas](https://stripe.com/atlas).** I want y’all to build amazing things, and not get bogged down in legal and tax hell. Stripe Atlas magically automates the entire startup incorporation process. **For just a few hundred dollars, they retrieve your EIN, issue founder stock, and file your 83(b) tax elections. Within two business days of registering, you’ll be able to open a bank account, fundraise, and start charging customers—even before your EIN has been issued.** In addition, Atlas startups get world-class company legal documents, the same breadth and caliber you’d get if you spent thousands on a top-tier law firm. Cursor, Lovable, and Linear were all incorporated through Atlas (along with 80,000 other startups), and it’s the no-brainer first step to making your startup a real thing. [Learn more about Stripe Atlas here](https://stripe.com/atlas) and [get the deal here](https://lennysproductpass.com/). This brings the total value of the [Lenny’s Newsletter product pass](https://lennysproductpass.com/) to $15,546. For just $200-$350/year. That’s also what I’d call a no-brainer. **With this additional puzzle piece, more and more your newsletter subscription becomes your product builder’s starter pack: all the guidance, community, and products you need to build and grow a world-class product (and career). For just $200-$350/year.** ![Image from Part 2 of how to get the most out of your product pass—and welcome, Stripe](https://substack-post-media.s3.amazonaws.com/public/images/8ff85855-5718-414d-b5df-ea3bae4f8691_2912x1880.png) And this will only get better. Over the next few months, we’ll be adding several more premium products that’ll be available to all annual subscribers (not just Insiders). Next is the second part of my comprehensive [guide](https://www.lennysnewsletter.com/p/how-to-get-the-most-out-of-your-product) for getting the most out of the product pass products. Below are all the ways that **[Lovable](https://lovable.dev/lenny), [Bolt](https://bolt.new/), [Granola](https://www.granola.ai/), [ChatPRD](https://www.chatprd.ai/lenny), [Superhuman](https://superhuman.com/), [Raycast](https://www.raycast.com/), and [Perplexity](https://www.perplexity.ai/pro)** have changed my life and are transforming people’s lives at work and at home. Again, my goal here isn’t to sell you the products ([you already get them for free](https://www.lennysnewsletter.com/p/productpass)!) but to help you get the most value possible out of your paid subscription. Enjoy. # **1. [Lovable](https://lovable.dev/lenny)** #### **What it is** Not only the fastest-growing company in history ($100 million ARR just eight months after launch, faster than Cursor 🤯), Lovable is an AI-powered vibe-coding platform that enables anyone to build apps and websites by chatting with AI. PMs can turn sketches, screenshots, and ideas into live interactive prototypes. Designers can refine the UX inline. Engineers review the code in their GitHub repos. As of last month, Lovable includes cloud and AI features that allow you to build AI-native, full-stack apps with zero configuration or tedious work required. #### **Why I love it** Whenever I’m vibe coding something, I try a bunch of the tools in parallel. Lovable has been producing the best stuff for me lately. #### **What I use it for** A bunch of personal use cases. I built [a YouTube preview tool](https://youtube-thumbnail-preview.lovable.app/), [a YouTube thumbnail downloader](https://yt-thumbnail-grab.lovable.app/), [a strikethrough text tool for X](https://twitter-strike-through.lovable.app/), and [a tool to download all of the images inside a Google Doc](https://gdoc-images-grab.lovable.app/). So fun! #### **How other people are using it** 1. Product teams: turning PRDs or screenshots into live, interactive prototypes to validate ideas 2. Marketing teams: launching landing pages; turning customer collateral into interactive websites with trackable CTAs 3. Sales teams: building ROI dashboards that quantify product value 4. Finance teams: taking FP&A out of spreadsheets and into live, easy-to-use dashboards [Here’s a fun directory](https://lovable.dev/projects/featured) of what people have built with Lovable. #### **Learn more** Check out their [website](https://lovable.dev/), and [my chat with Lovable CEO Anton Osika](https://www.youtube.com/watch?v=DZtGxNs9AVg). # **2. [Bolt](https://bolt.new/)** #### **What it is** Also one of the fastest-growing companies in history (0 to $40 million ARR in five months 🤯), Bolt is an enterprise-grade AI-powered vibe-coding platform that helps PMs, founders, and designers turn ideas into fully functional prototypes using AI and natural language—no code required. Bolt is used by 72% of Fortune 500 product teams, and even allows you to build using your company’s native design system. #### **Why I love it** What makes Bolt unique in my mind is that, unlike other popular vibe-coding platforms, which rely on their own homegrown agents, Bolt integrates with state-of-the-art frontier agents, like Claude Code (with OpenAI Codex and Gemini CLI), so you always get access to the most advanced coding agents, without switching tools. In other words, if you love Claude Code, you’ll love Bolt. #### **What I use it for** Also a bunch of personal use cases. [A feed of the latest AI news (updated daily)](https://daily-ai-news-feed-2nt6.bolt.host/), [a dial of how stressed I am based on workload that I can share with my wife](https://lennys-stress-level-xv83.bolt.host/), [a simple app to tell me if my son needs a jacket at school today](https://toddler-jacket-weath-f61o.bolt.host/), and [a tool to learn basic Spanish](https://spanish-learning-app-plkw.bolt.host/). #### **How other people are using it** 1. [Managing and reducing subscription costs](https://x.com/jakub_works/status/1938680359840792984) 2. [Building video games (and building tools that build them)](https://game-frame.site/) 3. [Helping people learn better](https://x.com/noratutor/status/1945974205465403565) 4. [A lawyer created an app to transform complex legal documents into clear, actionable insights](https://serene-sunflower-becadb.netlify.app/) 5. At a recent hackathon, a solo builder named Adrian created [Tailor Labs](https://www.tailorlabsai.com/), an AI-powered video editor #### **Learn more** Check out their [website](https://bolt.new/), and [my chat with Bolt CEO Eric Simons](https://www.youtube.com/watch?v=L22DtAHLmzs). # **3. [Granola](https://www.granola.ai/)** #### **What it is** An AI notepad designed for people in lots of Zoom/Teams/Meet meetings (i.e. everyone). As you take notes, it silently transcribes everything said in the background. Once the meeting is complete, Granola enhances your notes, summarizing what everyone said, and makes it easy to share (and search) what was said. Best of all, Granola doesn’t join your calls. Instead, it works behind the scenes using your laptop audio. There’s an iPhone app for in-person meetings and phone calls too. #### **Why I love it** I was blown away by both the UX and the quality of the notes it generates. You can use it to quickly share meeting notes with colleagues, recall decisions you’ve made in past meetings, analyze themes over time, and even [notice when you’re avoiding conflict (with help from Claude Code)](https://www.lennysnewsletter.com/i/175662739/noticing-when-youre-avoiding-conflict-from-dan-shipper). The product is so thoughtfully designed in every way, and everyone I introduce it to gets hooked. It’s quickly become the obvious-tool-to-use for AI meeting note-taking. #### **What I use it for** Every single meeting I have. #### **How other people are using it** 1. [Linear’s CEO uses Granola to stay focused](https://x.com/karrisaarinen/status/1922668975164461079) 2. [Intercom’s founder uses it to stay on top of all his meetings](https://x.com/destraynor/status/1907909497399734459) 3. [Getting coached in the style of a top CEO coach](https://x.com/sarahdrinkwater/status/1973653176101511297) 4. [Searching for something someone said in any of your meetings](https://x.com/nakul/status/1973473942724813245) 5. [For stressful non-work situations](https://x.com/jaymehoffman/status/1973121013568381239) #### **Learn more** Check out their [website](https://go.granola.ai/V1LQczH) and their cool new [Recipes feature](https://www.granola.ai/blog/say-hello-to-recipes). # **4. [ChatPRD](https://www.chatprd.ai/lenny)** #### **What it is** An AI platform built specifically for product managers. It writes product specs, standardizes your documents across your team, and integrates with the best-in-class PM tools, including other product-pass products like Lovable, Bolt, Gamma, Magic Patterns, and Linear. It’s especially great at taking raw ideas, turning them into clear PRDs, and then sharing that with your favorite vibe-coding tool. #### **Why I love it** There’s no other AI tool out there that’s this laser-focused on both making PMs’ lives better *and* helping everyone else on the team play PM. It’s also [a first-class citizen in Linear](https://linear.app/integrations/chatprd) now, meaning you can use it to improve task descriptions, break down work into actionable sub-issues, and get product feedback (by simply mentioning @chatprd in comments). *Neat*. #### **What I use it for** Since I’m not building products with teams these days, I don’t use ChatPRD much, but I’ve heard nothing but rave reviews of it from readers. And having worked closely with [Claire Vo](https://clairevo.com/) on [her pod](https://www.youtube.com/@howiaipodcast), the [Lenny & Friends Summit](https://www.youtube.com/watch?v=93fCvFkY1Lg), and all kinds of other side projects, I know anything she builds will be top-notch. #### **How other people are using it** 1. [Turning messy notes into “a banger PRD in no time”](https://x.com/marchattonhere/status/1966129003061723646) 2. [Automating your software development lifecycle](https://x.com/nummanthinks/status/1962141253992161373) by connecting it to Linear 3. [Planning features for agentic coding tools](https://x.com/JoschuaBuilds/status/1956240158405144934) 4. [Using Lennybot and ChatPRD together](https://x.com/robin_faraj/status/1955316655401127983) 5. [Helping you improve your structured thinking](https://www.linkedin.com/posts/max-vorhauer_one-of-my-favorite-ai-tools-in-the-recent-activity-7379485447695052800-Ckk1?rcm=ACoAAABaMoABXs298Ex3dZ9oSQYFUA_dcPeNiXI) #### **Learn more** Check out the [website](https://www.chatprd.ai/) and [this rad templates library](https://www.chatprd.ai/templates). # **5. [Superhuman](https://superhuman.com/products/mail)** #### **What it is** The most beautifully designed and thoughtfully crafted email productivity tool, specifically focused on helping you get to inbox zero. Superhuman’s research shows that the average user gets through their inbox *twice* as fast (vs. using Gmail or Outlook) and saves four hours a week. #### **Why I love it** If you want to see what a well-crafted consumer product looks like, study every pixel of Superhuman. #### **What I use it for** Email! #### **How other people are using it** 1. [Drafting automatic replies with AI](https://superhuman.com/products/mail/ai?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863) *[(](https://superhuman.com/products/mail/ai?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863)*[in your](https://superhuman.com/products/mail/ai?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863)[own voice and tone)](https://superhuman.com/products/mail/ai?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863) 2. [Collaborating on emails with your colleagues](https://superhuman.com/products/mail/startups?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863#:~:text=the%20business%20forward.-,Team%20Comments,-No%20more%20discussing) 3. [Creating team snippets of commonly used phrases, paragraphs, or even whole emails](https://superhuman.com/products/mail/startups?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863#:~:text=with%20your%20team.-,Snippets,-Accelerate%20your%20team) 4. [Booking meetings and sharing your availability](https://blog.superhuman.com/find-time/?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863) 5. [Automatically filtering and organizing incoming email, smartly](https://superhuman.com/products/mail/startups?hsa_acc=3060134131&hsa_cam=22926086053&hsa_grp=&hsa_ad=&hsa_src=x&hsa_tgt=&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=22916203863) #### **Learn more** Check out their [website](https://superhuman.com/products/mail), and [my podcast chat with founder Rahul Vohra](https://www.youtube.com/watch?v=0igjSRZyX-w). # **6. [Raycast](https://www.raycast.com/)** #### **What it is** It replaces your Mac’s native Spotlight tool (Cmd-space) and makes it 100 times more powerful. Just watch the video above to see the kinds of things it can do. It’ll blow your mind. #### **Why I love it** Similar to Granola, it’s one of these obvious-tools-to-use. If you ever use Spotlight, switch to Raycast immediately. #### **What I use it for** I mostly use it to quickly launch (and switch between) apps, do math, and find random system settings, but I’m also uncovering all the other wild stuff Raycast can do: position windows on my screen, set timers, create snippets of text I use often, extract colors from any image, talk to ChatGPT, and smartly search emojis. ![Image from Part 2 of how to get the most out of your product pass—and welcome, Stripe](https://substack-post-media.s3.amazonaws.com/public/images/1d81c0d7-7af6-4e15-b39d-c529a55b43be_1548x1008.png) #### **How other people are using it** 1. [Snippets for standardized email replies](https://www.raycast.com/community-stories/sania-saleh) 2. [Really great file search](https://www.raycast.com/community-stories/simon-kubica) 3. [Launching your most-used apps via hotkeys](https://www.raycast.com/community-stories/gavin-nelson) 4. [Creating custom shortcuts to Windows Management commands](https://www.raycast.com/community-stories/pawel-grzybek) 5. [Auto-join all your meetings without having to click any links](https://www.raycast.com/community-stories/james-mulholland) [Tons more here](https://www.raycast.com/community-stories). #### **Learn more** Check out their [website](https://www.raycast.com/) and their [store of extensions](https://www.raycast.com/store). # **7. [Perplexity](https://www.perplexity.ai/pro) (makers of Comet!)** #### **What it is** Perplexity is a leading AI-powered answer engine (with in-line citations, built-in deep research, and a focus on accuracy), but, most exciting of all, they recently launched [Comet](https://www.perplexity.ai/comet), the world’s most popular AI-powered browser. #### **Why I love it** Comet has been my default browser for months, and I really love it. #### **What I use it for** Browsing! #### **How other people are using it** 1. [For finance teams](https://www.youtube.com/watch?v=HT7ZEILjs-U) 2. [For business development](https://www.youtube.com/watch?v=iNqe1S5XWUQ) 3. [Automating entire workflows with Comet](https://www.youtube.com/watch?v=lqAHw6TwLsk) 4. [Creating financial reports using Perplexity Labs](https://x.com/AravSrinivas/status/1928220532614537318) 5. [Searching and analyzing subreddits for ad angles using Comet](https://x.com/NathanSnell/status/1943095214932943291) #### **Learn more** Check out [Perplexity](https://www.perplexity.ai/pro) and [Comet](https://www.perplexity.ai/comet). Finally, a question: What’s a product you’d absolutely love to see me add to the product pass? Leave a comment. [Leave a comment](https://www.lennysnewsletter.com/p/part-2-of-how-to-get-the-most-out/comments) *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [42/46] Ecosystem is the next big growth channel *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[AI/PM courses](https://maven.com/lenny) | [Public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *Annual subscribers get a **free year of 17+ premium products**: [Devin, Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, ChatPRD + Stripe Atlas](https://www.lennysnewsletter.com/p/productpass) (while supplies last). [Subscribe now](https://www.lennysnewsletter.com/subscribe?).* I’m always on the lookout for alpha to share with you on what the most successful companies are doing differently—so you can do it as well. This post is the epitome of that. [Emily Kramer](https://www.linkedin.com/in/emilykramer/) has identified a pattern in how top AI (and non-AI) startups are able to break through the noise and grow unlike at any other time in history. Once you see it, you immediately rethink your growth strategy. I did! After reading the first draft of this post, I got a whole new level of understanding for why my [product pass](https://lennysproductpass.com/) has been such a success (fun fact: it doubled my growth, and is the most win-win-win idea I’ve ever concocted). Below, Emily shares how you can implement this strategy yourself, along with dozens of real-world examples of how top companies are executing it. For more insights from Emily, check out her excellent [newsletter](https://newsletter.mkt1.co/), and find her on [LinkedIn](https://www.linkedin.com/in/emilykramer/) and [X](https://x.com/emilykramer?lang=en). *P.S. You can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from Ecosystem is the next big growth channel](https://substack-post-media.s3.amazonaws.com/public/images/93162fea-515c-4810-a010-6bb96eaa2d67_1456x970.png) You’ve built something great. Maybe even shipped it faster than you thought possible, thanks to AI. But here’s the catch: everyone else has that same speed advantage. So you turn to distribution. But here’s the second catch: the B2B channels you’ve long relied on for growth are increasingly becoming less effective, for that same reason. - **Inbound:** Search traffic is declining due to LLMs, and social is flooded with derivative content. - **Outbound:** With AI SDRs and easier access to contact data, everyone’s inboxes are overflowing. - **Product virality:** We’re overwhelmed with new products (thanks, AI!), so with endless choice, virality is much harder to achieve. - **Events:** While people crave authentic connection in the AI and hybrid-work era, the explosion of trade shows, micro-conferences, and webinars of mixed quality has led to event fatigue. - **Lifecycle emails:** With inboxes saturated with cold outbound, even lifecycle emails focused on driving engagement and retention get tuned out. So in this new competitive, AI-driven landscape, how do you get your great product noticed? #### **The answer is** ***ecosystem*****. Instead of going directly to your prospects, go through intermediaries who already have access and trust with your audience.** [Tom Orbach](https://www.linkedin.com/posts/tomorbach_building-an-audience-from-scratch-is-stupid-activity-7376883159348428800-6cFa?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU), the director of growth marketing at Wiz, said it best: “Why start at zero when you can start at 10,000?” ![Image from Ecosystem is the next big growth channel](https://substack-post-media.s3.amazonaws.com/public/images/0fb09e9e-0690-42a7-b342-2608f03bf3e1_1540x736.webp) You’re likely familiar with the individual tactics within an ecosystem strategy: influencer and creator relationships, channel partnerships, developer relations, communities, product integrations, and customer marketing. But the growth unlock doesn’t come from any one of these activities on their own. Instead, a flywheel emerges when you properly implement this overarching strategy: your ecosystem partners create and distribute content, you amplify and repurpose their efforts, together you drive greater reach and credibility, new customers and partners come on board, and the cycle strengthens with each turn. Now your ecosystem is a powerful extension of your GTM team. ![Image from Ecosystem is the next big growth channel](https://substack-post-media.s3.amazonaws.com/public/images/7919a2aa-fe3c-460d-9c72-5bf12811abbc_1456x1238.png) ## Ecosystem is a force multiplier *and* differentiator When you look at the fastest-growing companies today, you probably see AI-first companies known for their lean teams and great products. But I see something else: companies that wouldn’t have achieved these growth trajectories without their ecosystems. #### [Supabase](https://supabase.com/) The Postgres development platform and database became the default backend for vibe-coding products like Lovable. As these companies grew, so did Supabase. In addition to these “integration” partners, Supabase’s open source community shared docs, tutorials, and videos, further driving growth and building credibility. → [Supabase went from 1 million to over 4.5 million](https://www.linkedin.com/feed/update/urn:li:activity:7381727584788811778?updateEntityUrn=urn%3Ali%3Afs_updateV2%3A%28urn%3Ali%3Aactivity%3A7381727584788811778%2CFEED_DETAIL%2CEMPTY%2CDEFAULT%2Cfalse%29) developers in under a year, which would not have been possible without two sets of ecosystem players: the open source community and the awareness from integration partners. > *“Community and virality aren’t ‘intangible.’ Supabase shows how to turn organic pull (GitHub stars, YC adoption, vibe-coding buzz) from developers into structured funnels, lifecycle paths, and monetization.” —[Aaron Cort, operating partner at Craft Ventures, investors in Supabase](https://www.linkedin.com/in/aaroncort?miniProfileUrn=urn%3Ali%3Afsd_profile%3AACoAABeN5BwB8WucJrGbyT_XnJEEtG-WLLOhPII)* #### [Clay](https://www.clay.com/) Clay grows through close partnerships with “GTM engineer” power users—consultants, agencies, and in-house operators. These Clay creators make videos and tutorials that drive adoption and awareness. → Without this creator ecosystem showing what’s possible, Clay’s flexibility could have been a liability instead of an advantage. #### [Lovable](https://lovable.dev) Lovable helps builders quickly create apps that are naturally shareable—a classic driver of viral growth (think “Built with Webflow” or “Made in Typeform” badges). But Lovable accelerates this virality through community-driven initiatives like [hackathons](https://www.youtube.com/playlist?list=PLbVHz4urQBZmqUUa7Kh2y5NqpatC1KloT) and their [Discord, which just surpassed 100,000 members](https://www.linkedin.com/posts/lovable-dev_lovables-discord-community-just-passed-100000-activity-7372355679162007552-flWK?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU). → Lovable’s product is excellent, but there are many other vibe-coding products out there. Lovable’s investment in building a brand alongside their creators has catapulted them above the rest. > *“For years, the marketing playbook was predictable, but not anymore. All the attention has shifted to the creator economy. **The goal is no longer to produce everything in-house but to empower creators to generate content about you**,and then repurpose it across paid and organic.” —[Elena Verna, head of growth at Lovable](https://www.elenaverna.com/p/9-ways-growth-is-different-in-ai)* #### [Gamma](https://gamma.app/) Gamma built an ecosystem of micro-influencers across TikTok, Instagram, X, and LinkedIn to show off presentations built with Gamma’s AI design tool. They paid them not just to post but offered a bonus for virality. They also extensively catalogued successful hooks and formats to make this scale. → Gamma attributes **over half of their growth** to influencers, and founder Grant Lee breaks down [the details in this post](https://x.com/thisisgrantlee/status/1966874658680303880). #### [Vercel](https://vercel.com/) Vercel’s huge developer community builds on [Next.js](https://next.js) (which Vercel leads work on), and those projects naturally scale into Vercel enterprise deals. → Without the Next.js ecosystem and community contributions, Vercel likely never would have had the credibility or reach to move upmarket. #### [ElevenLabs](https://elevenlabs.io/) **ElevenLabs** turned users into contributors through its Voice Library. Voice actors upload voices and earn money when they’re used, which attracts more creators. These voices also power viral content like “[AI presidents](https://www.youtube.com/watch?v=9d06Tgj--Sg)” on YouTube and TikTok. Viral hits drive press coverage, too. → [ElevenLabs hit $100 million ARR with just 50 employees](https://www.nytimes.com/2025/02/20/technology/ai-silicon-valley-start-ups.html). Without its paid creator ecosystem (over *[$1 million to creators in 2025](https://elevenlabs.io/voice-library)*), the product itself would have fewer voices and it wouldn’t have spread as quickly. ![Image from Ecosystem is the next big growth channel](https://substack-post-media.s3.amazonaws.com/public/images/3bdd10d6-1cfb-44f9-b0c8-beb6fb78259e_1456x2244.png) You’ll see ecosystem flywheels behind nearly every fast-growing AI startup, from **[Baseten](https://www.baseten.co/)**’s strategic partnerships with cloud providers, to **[Midjourney](https://www.midjourney.com/)** embedding its product in Discord from day one, to **[Anthropic](https://www.anthropic.com/)** building [MCP](https://www.anthropic.com/news/model-context-protocol) to make it easier for developers to connect tools and data to Claude and other LLMs. ## Why ecosystem works I initially became obsessed with the power of ecosystems firsthand at [Carta](https://carta.com/solutions/law-firms/) in 2018, where lawyers as channel partners drove massive growth, and at Astro in 2017, where building for the Slack platform led to [Slack acquiring Astro, in its largest acquisition at the time](https://qz.com/work/1392936/slack-has-made-its-biggest-acquisition-to-date). Back then, these plays felt like outliers. But today, this strategy is essential. An ecosystem is fast becoming the rule for growth, not an exception. **Here’s why:** #### 1. Your ecosystem makes every channel more effective The best reason to invest in an ecosystem is that, when done well, it makes every channel better by serving as both a distribution path and a source of content. **Ecosystem partners fuel inbound, add value and credibility to your outbound messages, accelerate product virality, make events more enticing, and ground lifecycle marketing in real use cases.** > *Example: HubSpot partnered with a bunch of marketing influencers [to promote their new framework, “Loop Marketing,”](https://www.linkedin.com/posts/emilykramer_hubspotmediapartner-activity-7376256981201862656-xjEb?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU) when they announced it at their Inbound conference this year. This combination spurred lots of discussion, driving a bunch of organic Linkedin posts too. All of these efforts came together to make it feel like the Loop Marketing framework was everywhere that week (at least if you were a B2B marketer).* #### 2. Your ecosystem knows how to speak to your audience (maybe better than you do) AI has made it easy to pump out content, making it harder than ever for people to know who and what to trust. This means they are turning to voices they already value and respect (like Substack writers!). Ecosystem players like these already know how to speak to your audience. Your brand simply can’t build that trust overnight—if ever—without these partners. You see this especially on LinkedIn, where individual people (employees, customers, or partners) perform much better than brands—eight times better from employees vs. brands, according to [LinkedIn](https://www.linkedin.com/pulse/5000-engagement-boost-power-employee-generated-content-sabina-brdnik-bipnf/)*.* > *Example: Vanta may be best known for its clever [play-on-words billboards](https://www.lennysnewsletter.com/i/145424815/making-an-impression-how-do-i-ensure-my-campaign-is-memorable), but you’ve likely also seen them sponsoring business Substacks (like Lenny’s) or podcasts. Since their product centers on trust, it’s smart to partner with high-trust creators who their (compliance-minded) audience already knows. It’s clear that a significant share of their paid spend now goes not just to billboards but to creators too. Fun fact: Vanta is one of Lenny’s top 3 podcast sponsors.* #### 3. GTM teams are smaller now and need leverage AI was supposed to make GTM teams more efficient—and in some ways it has. But [shrinking](https://www.chiefmarketer.com/gartner-39-of-cmos-plan-to-reduce-labor-costs-and-cut-agency-allocations/) GTM teams are scrambling to get efficiency through AI while also finding creative, compelling ways to differentiate. Your ecosystem is an accelerant for a lean team, like bringing on a contracted team of expert marketers. Working with partners, customers, and communities may not be as turnkey as paid search, for instance, but with the right workflows (and a boost from AI), the ROI can be extremely high. > *Example: [Navattic](https://www.linkedin.com/posts/natalie-marcotullio_for-our-most-recent-launch-i-wanted-to-activity-7355587957677858816-meh_?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU) built a team of marketing advisors, which makes the team seem much larger than it is. They set up long-term or advisory arrangements with micro-influencers who get early access to products, give product feedback, and also regularly post on LinkedIn as part of the relationship.* #### 4. You have a ton of options within an ecosystem strategy Ecosystems aren’t limited to products you integrate with or creators with big LinkedIn followings. And if one ecosystem partner starts showing diminishing returns, you aren’t out of luck. There’s a broad network of individuals and companies that can reach your audience. If a person or company, no matter the size or reach, has spent time building up the audience you want, you can figure out how to work with them. > *Example: If you have a niche audience, you can often find organizations and trade associations to partner with. [Clio](https://www.clio.com/partnerships/bar-associations/) works with state bar associations to reach lawyers, and [Upstart](https://www.upstart.com/) has long been a preferred partner with the [NAFCU](https://www.businesswire.com/news/home/20210616005046/en/NAFCU-Services-Announces-Upstart-as-a-Preferred-Partner-for-their-AI-Lending-Platform-for-Credit-Unions), a regional credit union trade association.* ## How to implement an ecosystem strategy Growth strategies are never one-size-fits-all, but most B2B startups today can grow faster with a well-designed ecosystem strategy. With audience attention fragmented and harder than ever to earn—and startups hitting revenue milestones faster—you can’t afford *not* to leverage the people, companies, and platforms that already have your audience’s trust. Because there are so many types of partners and ways to collaborate, there’s opportunity here for nearly every company. **The question isn’t** ***if*** **you should build an ecosystem strategy, it’s** ***who*** **your best partners are,** ***how*** **to activate them specifically, and** ***how*** **to reshape your team and process to prioritize that work.** ![Image from Ecosystem is the next big growth channel](https://substack-post-media.s3.amazonaws.com/public/images/da18866a-b312-4d30-b5af-86fb02540385_1456x1206.png) #### 1. Recognize that an ecosystem is not a side project The most effective ecosystem strategies aren’t bolted on or handled by a single partner marketer. You need buy-in at the senior leadership level. Often, product needs to be deeply involved in the process in order to design integrations, build referral mechanics, create products or features for channel partners specifically, and embed themselves as product experts in communities. > *Example: At [Tracksuit](https://tracksuit.com), an always-on brand tracking tool, agency partners are both power users and advocates. Agencies drive so many referrals, the team is doubling down on this flywheel by building products to support agencies’ internal use cases.* #### 2. Map your ecosystem Competitive research is a common exercise, but “complement” research is very often skipped or not even thought about. By this, I mean identifying and mapping all the players in your ecosystem. The key is to look beyond the obvious and consider all the players who already have reach and credibility with your audience. The 2x2 at the start of this section breaks down the main types of ecosystem partners. You can use this to help map your own ecosystem. The most outsize returns typically come from the top half of the grid, since relationship-driven partnerships tend to result in the highly resonant creative. Many relationships are bespoke, so depending on the partner, their placement on the grid may shift. #### 3. Make the partnership worth it: look for win-win-wins Every partnership adds another cog in the wheel, so it needs to be worth the effort. My personal bar for partnerships is that they must be a “win-win-win” for you, the partner, *and* the prospect, never one or two out of three. > *Example: With [Lenny’s Product Pass](https://www.lennysnewsletter.com/p/productpass), startups get promoted through Lenny’s Newsletter, subscribers get free tools, and Lenny gets more paid subscribers.* Remember that money and distribution aren’t the only ways to make a partnership worth it. Value can come in the form of shared data, shared event costs, special product feature access, etc. Also, when choosing partners, remember that it’s not just about their audience size. Audience overlap matters too. I look at two filters, borrowed from the advertising world: - **Coverage:** What percentage of your total target audience do they reach? Higher coverage means you can reach more of your TAM (total addressable market) through this partner. - **Composition:** What percentage of their audience overlaps with yours? Higher composition means less waste—you aren’t reaching a bunch of people who don’t care about your product at all. - **High coverage and high composition:** If you can find a partner who ranks high for both, and you have shared values and goals, it could be a match made in B2B heaven. It can still be beneficial to work with a partner where there is only high coverage of your audience or only high composition of your audience, but you’ll want to use different strategies. #### 4. Start narrow, then expand Don’t try to launch every kind of partnership at once. Pick one program that fits your product and audience and prove it works. If it is a win-win-win for you, the ecosystem partner, and your audience, go deeper or expand the program. But as you scale, don’t lose the human relationship and authenticity that created the success in the first place. > *Example: [Arrows](https://arrows.to) rebuilt their standalone customer onboarding product exclusively for HubSpot users. They’ve grown almost 40x since then and are now using it to expand into Salesforce and other ecosystems. [CEO Daniel Zarick](https://www.linkedin.com/feed/update/urn:li:activity:7374531049785102336?updateEntityUrn=urn%3Ali%3Afs_updateV2%3A%28urn%3Ali%3Aactivity%3A7374531049785102336%2CFEED_DETAIL%2CEMPTY%2CDEFAULT%2Cfalse%29) said, “Nobody would care about us if we didn’t try to make a mark with HubSpot first.”* #### 5. Remember: an ecosystem strategy only works when you’re leveraging multiple channels If you don’t know where to start on ecosystem marketing or how to expand your efforts, think about using it to breathe new life into channels you’re already using. Here are some examples: **Inbound:** Use a community to get people excited to talk about your product. They may continue that conversation on other platforms, like Reddit, and that content has a higher chance of surfacing in an LLM than your own. *Lenny had [Ethan Smith, the CEO of Graphite](https://www.linkedin.com/in/ethanls/), on [his podcast](https://www.linkedin.com/posts/lennyrachitsky_the-reddit-strategy-for-llm-mentions-1-activity-7376338655411568640-qAKn?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU) talking about the Reddit LLM strategy.* **Outbound:** Boost a sponsored influencer post on LinkedIn as a Thought Leader Ad from your company, use engagement with the post as a “signal,” then connect and follow up in a DM. *Here’s an example of a [sponsored post I did with an agency](https://www.linkedin.com/posts/emilykramer_sponsored-activity-7303084395828051969-pKPT?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU). They then boosted it and followed up with promising leads.* **Product virality:** Drive product adoption by offering an educational course, then make sure students show off what they learn. *[AirOps does this with their three-week AI marketing course](https://www.airops.com/cohort), and [GC AI](https://gc.ai) does this via [Maven courses](https://maven.com/ceciliaz/prompting-crash).* **Events:** Run programming in your trade-show booth with partners to get people to show up, without the gimmicks and throw-away-able swag. *Mercury hosts [expert sessions](https://www.linkedin.com/posts/mercuryhq_good-morning-from-techcrunch-disrupt-come-activity-7256709231838408704-V3GX?rcm=ACoAAAC36HYBYwE6pfdDyi_36enIwk6IrfnVTyU) directly from their booth at TechCrunch Disrupt.* **Lifecycle:** Instead of sharing generic use cases, highlight how top customers use the product through videos or educational events. ## Don’t just build products, build ecosystems AI has leveled the playing field for building and distributing products. What once felt like an advantage—shipping faster or producing more content—has become table stakes. The real differentiator now is how you stand out and build a moat in an AI-saturated world. An ecosystem is that lever. Rather than a side project or a single partnership, it’s a company-wide strategy that unites partners, customers, creators, influencers, and communities into an extension of your team. You win by orchestrating all of these players into a flywheel that compounds credibility, distribution, and growth. Done right, an ecosystem strategy doesn’t just add another tactic to the mix—it makes every channel better. For today’s lean(er) GTM teams, it’s the force multiplier you can’t afford to ignore. **The companies breaking through today are going beyond building great products. They’re building ecosystems around them. That’s what will separate the winners in this next era of growth.** *Thanks, Emily! For more from Emily, check out her [newsletter](https://newsletter.mkt1.co/), and find her on [LinkedIn](https://www.linkedin.com/in/emilykramer/) and [X](https://x.com/emilykramer?lang=en).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [43/46] A holiday gift guide for tech people with taste 🤌 *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[My fav AI/PM courses](https://maven.com/lenny) | [My fav public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *Become an annual subscriber and get an unbelievable deal: **A full free year of 17+ premium products, including [Devin, Lovable, Replit, Bolt, n8n, Wispr Flow, Descript, Linear, Gamma, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, ChatPRD, and Stripe Atlas](https://www.lennysnewsletter.com/p/productpass) (while supplies last). [Learn more](https://www.lennysnewsletter.com/p/productpass).*** For the fifth consecutive year, I’m excited to present my holiday gift guide. The theme this year is taste 🤌:high-quality non-obvious stuff. No slop. I’ve included more high-end and handmade products this year than in previous years, but there are also still plenty of affordable items. As always, nothing below contains affiliate links, I’m not an investor in any of these companies, and I’m not benefiting from recommending any of these products (except one, which will be clear). This list is simply a collection of things I love and think you’ll love too. For more ideas, check out my previous gift guides from [2024](https://www.lennysnewsletter.com/p/lennys-newsletter-holiday-gift-guide-e6b), [2023](https://www.lennysnewsletter.com/p/a-new-parent-gift-guide-for-product) (for new parents), [2022](https://www.lennysnewsletter.com/p/lennys-newsletter-holiday-gift-guide), and [2021](https://www.lennysnewsletter.com/p/lennys-holiday-gift-guide). Did I miss something absolutely awesome? Share it in the comments 🙏 [Leave a comment](https://www.lennysnewsletter.com/p/a-holiday-gift-guide-for-tech-people/comments) Enjoy! ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/d4c9fa0b-ef85-49a8-ae68-a22b91bf4194_2912x1940.png) ### **For the home** 1. **[Handcrafted Modern maple donut bowl](https://www.handcraftedmodern.co/products/medium-maple-wood-donut-bowl):** My wife got us this, and I was like “wtf is it?” But then I realized how much a beautiful wood piece adds to your space. [The artist makes lots of other great stuff, too](https://www.handcraftedmodern.co/makers/elise-mclauchlan). ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/b6252421-205a-415e-99e0-a8832b4a4498_1600x1205.jpeg) 2. **[Coyuchi 100% organic cotton sheets](https://www.coyuchi.com/products/organic-crinkled-percale-sheet-set-undyed?variant=50911855411492):** Pricey, but oh so nice. You spend so many hours in your sheets, imho it makes sense to splurge here. ![Organic Relaxed Linen Sheet Set](https://substack-post-media.s3.amazonaws.com/public/images/81f64fd5-1724-46c0-916f-7db72ddd6d97_600x453.jpeg) 3. **[Deadline candle](https://sublime.metalabel.com/candle?variantId=1):** “Transforms your most dreaded moments into sacred ritual.” ![Finite time. Infinite potential. ](https://substack-post-media.s3.amazonaws.com/public/images/14c70f61-6d6d-4248-947f-efc6bb587409_1200x1600.jpeg) 4. **[Matic](https://maticrobots.com/):** [Wirecutter](https://www.nytimes.com/wirecutter/reviews/matic-robot-vacuum/), [Wired](https://www.wired.com/review/matic-robot-vacuum/), [Gizmodo](https://gizmodo.com/the-matic-robot-vacuum-charms-and-sucks-in-a-good-way-2000638109), [The Verge](https://www.theverge.com/tech/816645/matic-robot-vacuum-review), and [ZDNet](https://www.zdnet.com/home-and-office/kitchen-household/this-robot-vacuum-broke-the-industry-mold-heres-my-verdict-after-months-of-testing/) have all reviewed it and said it’s the best smart vacuum ever made. I agree. We use ours constantly. They are raising prices on December 2 (from $1,095 to $1,245—tariffs 🫠), so ordering before then will save you $150. They also offered me a deal to share with this community: use [this link](https://maticrobots.com/) to get a free 6-to-9-month supply of replacement HEPA bags. Their team sent me a unit to try out when they were just getting started, and I loved it so much I’ve been telling everyone about it. ![Matic](https://substack-post-media.s3.amazonaws.com/public/images/6ad9f1dc-92f9-458e-8a56-20f798352710_1308x872.webp) 5. **[Living Tea seasonal tea club](https://www.livingtea.net/products/seasonal-tea-club-gift-subscription?variant=40597401141315&selling_plan=2122285123):** A friend got this for me, and it’s so nice. They include background on each tea, how to best steep them, and all kinds of extras that make the experience feel special. [These stainless-steel infusers](https://www.fivemountainstea.com/shop/Gifts__Accessories/5MIBL001.html) are really handy (especially if you’re wary of tea bags these days). We also love [Leaves and Flowers](https://leavesandflowers.com/collections/all) and [Rishi](https://www.rishi-tea.com/collections/gift-sets) teas. 6. **[Koshi chimes](https://soundhealinglab.com/collections/chimes-and-shakers/products/koshi-chimes):** The sound of these is unique and delightful. “Simply move the chime gently holding it by its cord to produce a crystalline, relaxing sound that mesmerizes and calms.” ![Koshi Chimes - Made in France - Koshi Chimes](https://substack-post-media.s3.amazonaws.com/public/images/a91540b7-a390-42f5-9056-89a096c23114_1600x1214.jpeg) 7. **[Plant-based desk lamp](https://maju.design/?v=7516fd43adaa):** You may have noticed one of these in my podcast background. We bought it at the West Coast Craft fair in SF, but they also sell them at [SFMOMA](https://museumstore.sfmoma.org/products/lily-cloud-omni-santal-ash-lamp?variant=45466879295666). I don’t recommend the ones with legs, though, as they tend to come off. [These are also cute](https://store.moma.org/collections/lighting/products/space-s-mini-portable-lamp-orange). As are [these classic Noguchi lamps](https://nymag.com/strategist/article/steal-my-noguchi-lamp.html?gad_source=1&gad_campaignid=21471568684). ![Processed with VSCO with c3 preset](https://substack-post-media.s3.amazonaws.com/public/images/19f7f427-25cf-4bee-b62b-4d3c59066833_600x800.jpeg) 8. **[Angels Horn vinyl record player](https://www.amazon.com/ANGELS-HORN-Bluetooth-Turntable-Turntables/dp/B09VK9DLYR?th=1):** Includes built-in speakers (and Bluetooth!), so it’s a really easy, versatile, and affordable way to enjoy records. We’ve had this for years and love it. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/7696fe54-822c-44c4-a7d6-c379f9540685_1373x1369.jpeg) 9. **[Graf Lantz Bierfilzl merino wool round coasters](https://www.amazon.com/Graf-Lantz-Bierfilzl-Square-Coasters/dp/B0BXYQ1VXF?th=1):** I can’t get enough of these around the house. Lots of color options, and the [square ones](https://www.amazon.com/Graf-Lantz-Bierfilzl-Square-Coasters/dp/B09LKZHDRH?th=1) are nice too. ![Merino Wool Round Coaster Multi 6 Pack in Rainbow | Graf Lantz](https://substack-post-media.s3.amazonaws.com/public/images/bd4b3af4-8c6f-4202-a646-abe6f3a56b16_900x1125.jpeg) 10. **[Vera salt](https://verasalt.co/):** Microplastic-free salt. After learning that I was [half-man, half-plastic](https://www.lennysnewsletter.com/p/a-builders-guide-to-living-a-long) and also that sea salt is one of the most microplastic-laden foods one can consume ([thanks, Huberman](https://ai.hubermanlab.com/clip?sids=chunk_3456144) 🥴), we discovered this product (through [Superpower](https://superpower.com/) marketplace), and it’s now our go-to salt. Pretty, healthy, and yummy. Pairs well with this plastic-free [Klean Kanteen](https://www.kleankanteen.com/products/plastic-free-water-bottle-27oz?variant=860232191) water bottle (which is not as fun as other bottles but easy to clean). ![Flake Spring Salt (180g) - Microplastic Free – Vera Salt](https://substack-post-media.s3.amazonaws.com/public/images/5e508308-bb96-4734-899a-81e83cd99519_1445x963.jpeg) ### **For parents and kids** 1. ***[Charts for Babies](https://www.amazon.com/Charts-Babies-Picture-Book/dp/1419785184)*** **[by Michelle Rial](https://www.amazon.com/Charts-Babies-Picture-Book/dp/1419785184):** MY WIFE HAS A NEW BOOK COMING OUT. I know I’m biased, but it’s genius. It’s also cute as hell, will warm your heart, and will incept your kids into liking STEM. It’s coming out in April, and [you can pre-order it here](https://www.amazon.com/Charts-Babies-Picture-Book/dp/1419785184). If you haven’t seen [her first book](https://www.amazon.com/Am-Overthinking-This-Over-answering-questions/dp/1452175861/ref=sr_1_3?dib=eyJ2IjoiMSJ9.W6mGAhve3sUVTgLNelMtqfvnZVB3R1LNvZNxXtn9N4g5FRd_4YBibF5SJ4kwFM7URKi2vCnufB1z3M0SOnWtYfl9ymbui6Y7ZT1JQownA-UJRSQZaBynMVMLX_p9qDtY.z8c5kf1m1g3XDL-N4gxe5kidNA6GAcTVwma6mWSyeok&dib_tag=se&qid=1762985370&s=books&sr=1-3&text=Michelle+Rial), that one makes a great gift too (for adults). ![Charts for Babies A Picture Book](https://substack-post-media.s3.amazonaws.com/public/images/55950225-9b78-44db-ab1b-1c8cfd635c81_920x920.jpeg) 2. **[Emerging Artist beanie for babies](https://shop.whitney.org/products/emerging-artist-baby-hat):** We gifted this to our nephew when he was younger, and it’s never not funny. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/35d1e9db-947e-48b9-a94f-f4b11757b6ef_1000x655.png) 3. **Kids’ [guitar](https://loogguitars.com/products/loog-pro-acoustic-natural) or [piano](https://loogguitars.com/collections/loog-piano):** They come with guides to help you learn, and they are really high-quality. My wife and I even practice on it while we’re playing with our son because we’re both learning piano. ![loog piano](https://substack-post-media.s3.amazonaws.com/public/images/107eb3f6-33bb-4424-821e-a8cab5c80c2c_1024x1024.jpeg) 4. **[Yoto mini player](https://us.yotoplay.com/yoto-mini):** A neat screen-free audio player for kids. Our son loves his so much. He rotates between the Beatles, Raffi, Queen, and Spanish [cards](https://us.yotoplay.com/collections/best-sellers) while he walks around dancing, listening to it (rather than staring at a screen). How can you not love that? Sir Paul McCartney and Chan Zuckerberg Ventures are investors, to encourage more screen-free entertainment options for kids. Many people love the [Tonies](https://us.tonies.com/) as an alternative. ![Yoto 101: Everything You Need to Know About a Yoto Player - Everyday Reading](https://substack-post-media.s3.amazonaws.com/public/images/7bd9fadb-a60d-418a-849c-7133706cf0ba_2560x1707.jpeg) 5. **[Personalized ring](https://www.catbirdnyc.com/famous-letter-gold-ring.html#306=98) ([or necklace](https://www.catbirdnyc.com/tiniest-name-necklace.html)):** For your BFF, mom, lover, or just get it for yourself. As a fun twist, make it the joke name you call your kid/pet. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/a88a8e0c-8683-4be4-8c5f-a894bc097476_1020x402.png) 6. **Personalized [pennant](https://oxfordpennant.com/collections/customizable) or [blanket](https://littlemoonjumper.com/blankets/stroller-blankets/?sort=bestselling):** Is there anything kids/parents/humans love more than their name on stuff? ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/cc3c1cbf-ea07-4762-a8c4-bc3d1210de5e_1914x570.png) 7. **[Coyuchi handstitched organic toddler blanket](https://www.coyuchi.com/products/pebbled-handstitched-organic-toddler-quilt-undyed-w-warms?variant=50911873368356):** Nontoxic naps for your little one. 8. **[Kids’ Adirondack chair](https://www.hineighbor.com/collections/outdoor-chairs/products/low-chair-jr?variant=39959166976080):** Extra-cute if you place it near your adult-sized chair on the porch. ![A studio photo of Low Chair Jr. Default Title](https://substack-post-media.s3.amazonaws.com/public/images/2f41cf53-849a-46e0-978b-3f365c5d2823_800x800.jpeg) 9. **[Organic floppy brown teddy bear](https://www.bellalunatoys.com/products/senger-organic-cotton-brown-teddy-bear?variant=7265131266094):** This teddy bear is relatively expensive, but it’s a rare non-polyester option for something your kid will be snuggling all night. It’s made in Germany from certified organic cotton and pure organic wool. [They have other animals too](https://www.bellalunatoys.com/collections/handmade-toys/organic-toys). ![Senger - Organic Brown Teddy Bear - Brown bear - Germany - Bella Luna Toys](https://substack-post-media.s3.amazonaws.com/public/images/6bae884c-5dc3-426f-995c-eed4e57e1304_1000x1000.jpeg) 10. **[John Klassen prints](https://www.gallerynucleus.com/detail/13584/):** There are also [originals](https://www.gallerynucleus.com/artists/jon_klassen), but most are sold out. [More prints here](https://www.gallerynucleus.com/artists/jon_klassen?category=prints). IYKYK 🔺 ![I Want My Hat Back - Pg. 25-26 - Drama [PRINT], Jon Klassen](https://substack-post-media.s3.amazonaws.com/public/images/eb149af7-97b1-498b-b240-73c5734b71a3_750x513.jpeg) ### **For gadget lovers** 1. **[ModRetro Chromatic](https://modretro.com/products/chromatic-tetris-bundle?variant=49807115223342):** A modern remake of the original Game Boy by Palmer Luckey (founder of Anduril and Oculus). It feels great, plays great, and makes a unique gift for gamer friends. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/e0c82f8d-ee9e-4ca6-861b-d3ec4bd9fe0c_1000x667.jpeg) 2. **[Peak Design tech pouch](https://www.peakdesign.com/products/tech-pouch?Color=Eclipse&Size=Small):** They also make a [larger version](https://www.peakdesign.com/products/tech-pouch?Size=Regular&Color=Eclipse), and [their toiletry bag](https://www.peakdesign.com/products/wash-pouch?Size=Regular&Color=Coyote+X-Pac%E2%84%A2) is great. ![Person holding an Eclipse Tech Pouch Small with electronic accessories on a wooden surface](https://substack-post-media.s3.amazonaws.com/public/images/5b55ef7d-1299-4457-83f1-8af5ed72fabf_1024x1024.jpeg) 3. **[Aranet CO2 monitor](https://www.aranet.com/en/home/products/aranet4-home):** Did you know high CO2 impairs your cognition? [Research](https://pmc.ncbi.nlm.nih.gov/articles/PMC4892924/) shows that brain function decreases by 15% when CO2 levels are over 1,000 ppm, and by over 50% above 1,400 ppm. Freaky shit. I’ve got one of these in my office and open a window anytime it gets into the yellow zone. ![Aranet4 HOME US](https://substack-post-media.s3.amazonaws.com/public/images/923f1dfe-e908-4542-ad49-47ad110df468_1100x824.bin) 4. **[Belkin MagSafe 3-in-1 charger](https://www.amazon.com/dp/B0DL8CK73P?th=1):** I saw these at a hotel once, immediately got it, and love it. Better than any other nightstand charger I’ve tried. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/4b2af550-4892-487f-b14c-4d8f242c44ae_924x785.jpeg) 5. **[Herschel tech backpack](https://herschel.com/shop/backpacks/kaslo-daypack-tech?v=11289-07070-OS):** This has been my go-to bag for years. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/0748b920-bd9e-4c46-b08b-6ee75de3e45a_1456x1304.png) 6. **[Sony WH-1000XM5 noise-canceling wireless headphones](https://www.amazon.com/dp/B0CFRWFTQQ?th=1):** These have become my workhorse headphones. Grab [this cute stand](https://www.amazon.com/dp/B0DZHXKZNP?th=1) if you use headphones at your desk. My wife prefers the [Bose QuietComfort](https://www.amazon.com/Bose-QuietComfort-Wireless-Cancelling-Headphones/dp/B0CCZ1HQ39?th=1) because, as the name suggests, they are comfy. 7. **[iPhone case with built-in stand](https://www.amazon.com/dp/B0D6G99C1H?th=1):** I learned about this from [Kevin Rose’s newsletter](https://www.kevinrose.com/). There’s also [an Anker version](https://www.amazon.com/dp/B0D7GV2JGT?th=1), but I like this one better. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/8a1bfcc0-8c0a-417f-ad5c-3f06c50dbe86_1365x1500.jpeg) 8. **[Hooga amber book light](https://www.amazon.com/dp/B07Q87WJ7M?th=1):** I originally got it to change the baby’s diaper during the night while keeping the lights dim and super warm. Now use it nightly while reading in bed. Lasts forever, charges easily, and has multiple brightness settings. 9. **[Amazon Kindle Colorsoft](https://www.amazon.com/All-New-Amazon-Kindle-Colorsoft-Signature-Edition/dp/B0CN3XR57P?th=1):** Did you know they make Kindles in color? Thanks for the tip, [Laura Fingal-Surma](https://x.com/urbanistvc). 10. **[AirPods](https://www.apple.com/airpods/):** For all your friends who have “AirPods #1,” “AirPods #2,” . . . “AirPods #5” in their Bluetooth devices list. ### **For your stressed founder friend or family member** 1. **[Function Health](https://www.functionhealth.com/gifting) or [Superpower](https://superpower.com/gift) membership:** Make sure they aren’t pushing themselves *too* far. 2. **[Spirit Rock gift certificate](https://spirit-rock.secure.retreat.guru/program/gift-certificate/):** I did a 10-day silent retreat there a number of years back, and it was life-changing. [Check their upcoming retreats here](https://www.spiritrock.org/calendar?programType=retreats). 3. **[The Anti-Anxiety Notebook](https://shop.therapynotebooks.com/products/anti-anxiety-notebook): “**Utilizing Cognitive Behavioral Therapy, a rigorously-tested and widely-used treatment, you’ll develop the skills to identify, challenge, and change unhelpful thought patterns so you can feel better.” ![The Anti-Anxiety Notebook](https://substack-post-media.s3.amazonaws.com/public/images/9cba646c-0bfb-40f0-8ced-afcf995969e2_900x900.jpeg) 4. **[Japanese Hot Spring Minerals bath salts](https://jinenstore.com/collections/best-selling-gifts/products/jinen-japanese-hot-spring-minerals-bath-salts-hinoki-500g):** 🧘🧘🧘 5. **[Leaves and Flowers Sleep Tea](https://reliquarysf.com/collections/leaves-and-flowers/products/sleep-tea):** 💤💤💤 6. **[Rice wax Japanese candles](https://jinenstore.com/collections/best-selling-gifts/products/daiyo-rice-wax-japanese-candles-earth-colors): 🕯️🕯️🕯️** ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/9bb4f37c-1efa-490b-969b-69e4e4da8f6e_1500x878.png) 7. **[Candle holder](https://www.mbuenopottery.com/shop/p/candle-holder):** Goes well with [beeswax tea lights](https://bluecorncandles.com/products/pure-beeswax-tea-light-candles), and creates a really nice light effect. 8. ***[Runnin’ Down a Dream: How to Thrive in a Career You Actually Love](https://www.amazon.com/Runnin-Down-Dream-Thrive-Actually/dp/0593799666)*** **[by Bill Gurley](https://www.amazon.com/Runnin-Down-Dream-Thrive-Actually/dp/0593799666):** I haven’t read this yet, but I’m 100% sure it’ll change many people’s lives, considering the impact [his talk](https://www.youtube.com/watch?v=xmYekD6-PZ8) on the same subject has had on people I know. 9. **[Squishy stress ball set](https://www.amazon.com/dp/B01HN3RYDC?th=1):** I’ve got fidgety hands, so I often need something to play with. You may see me holding these during my podcast chats sometimes. “Free from BPAs, phthalates, and latex.” 10. **[“Things will work out” keychain](https://www.peopleiveloved.com/collections/objects/products/things-will-work-out-keychain):** A comforting reminder to have around. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/3f8e5beb-90a3-4a21-8101-3603189a45c1_1624x822.jpeg) ### **For the PM who just cashed out their OpenAI stock** 1. **Donate** to **[GiveDirectly](https://www.givedirectly.org/)**, **[St. Jude Research Hospital](https://www.stjude.org/)**, **[SF-Marin Food Bank](https://donate.sfmfoodbank.org/page/32140/donate/1?ea.tracking.id=DigAd2526-PMG&ea.tracking.id=DigAd2526-PMG&gad_source=1&gad_campaignid=22907547271)**, or your favorite charity. 2. **[Evil eye charm](https://www.shopesqueleto.com/products/evil-eye-hamsa-gold-charm?pr_prod_strat=e5_desc&pr_rec_id=cd53d792c&pr_rec_pid=8086385262779&pr_ref_pid=8086385131707&pr_seq=uniform) or [stoneware](https://www.mquan.com/collections/objects/products/eye-new-eye-indigo?variant=12128027836530)** 3. **[100-pound Parmesan wheel](https://flamingoestate.com/products/brads-parmesan-wheel):** “Crafted by hand from the milk of cows grazing high in the alpine meadows of Benedello di Pavullo, these monumental Parmesan wheels are aged by Giorgio Cravero, whose family has been perfecting the art of affineurship in Bra, Italy, since 1855. Simply put—the best Parmigiano Reggiano that exists.” ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/16b29601-3316-457f-8f73-5f494366b6cc_1544x1326.png) 4. **[A $4,000 wind chime](https://communedesign.shop/products/furniture-music-wind-chimes):** Part of Chris Kallmyer’s “collection of furniture made to intone atmospheric music and breezy melodies that promote social well-being through sound.” ![Chris Kallmyer for Commune Wind Chime](https://substack-post-media.s3.amazonaws.com/public/images/34a32df3-e310-4981-bd89-2f8d34e3d1e8_1445x964.jpeg) 5. **[Jony Ive’s $5,000 lantern](https://www.balmuda.com/lovefrom-balmuda/):** “The simple form is the result of meticulous design and engineering. Available in polished stainless steel with gold lens guard and component details. Easy to maintain, disassemble and repair, and to recycle at the end of a lifetime of use. A limited edition of 1,000 pieces.” ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/c17d09f2-4671-4deb-92a7-607dc4eadf77_1504x1504.jpeg) 6. **[A really nice coffee machine](https://home.lamarzoccousa.com/product/linea-micra/):** Make your home feel like your dream coffee shop. [There’s also one for $20k](https://home.lamarzoccousa.com/product/strada-x1/). [And a $4k coffee grinder](https://weberworkshops.com/products/eg-1?variant=31624830353461). [And this](https://weberworkshops.com/products/the-sg-1) 🤣 ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/720ea557-69bc-4700-9fc1-bacf4d65dc61_2560x1707.jpeg) 7. **[A really nice set of adjustable dumbbells](https://www.snodesport.com/products/ad-80-quick-adjusting-dumbbells):** “In its simplest form, strength means utilizing muscle to generate force. And if you are interested in living a long and healthy life and playing with your great-grandkids someday, then muscle mass should be a priority. Never in the history of human civilization has a 90-year-old said, ‘I wish I had less muscle.’” —[Peter Attia](https://peterattiamd.com/category/exercise/strength-muscle-mass/) ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/46ac4bf9-35dd-4fe4-b15e-ff766f8f2833_1200x815.png) 8. **[A really nice watering can](https://www.shopterrain.com/shop/haws-fazeley-flow-watering-can-05l?cj_flex=Content&cj_publishercid=3512519&cjdata=MXxOfDB8WXww&cjevent=f96198a3c34811f0801001ae0a1cb82b&cm_mmc=cj-_-affiliates-_-Wirecutter+Inc.-_-11554124&color=028&type=STANDARD&size=0.5L&quantity=1):** Wirecutter:“The highest-scoring can in our houseplant testing is also the smallest and most expensive. But it’s well-made—by a 130-year-old British company—and a delight to use.” ![Haws Fazeley Flow Watering Can, 0.5L #1](https://substack-post-media.s3.amazonaws.com/public/images/ffa2127d-3004-4e3f-9f9a-0590e0cd88e3_1450x2175.jpeg) 9. **[A really nice candle holder](https://origin-made.com/collections/store/products/candeia?variant=39858954272848):** “Consists of a mouth-blown glass shade paired with either a brass or alpinina stone base option. Candeia’s design was informed by gentle memories from the designer Rui Alves’ childhood.” ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/eb23255b-d044-405d-9058-0aa7ad3194ac_914x1024.jpeg) 10. **[Poetry written by humans](https://store.newyorker.com/products/a-century-of-poetry-in-the-new-yorker):** No slop here. ![A Century of Poetry in The New Yorker 2 ](https://substack-post-media.s3.amazonaws.com/public/images/b25ddc6f-375a-4b2f-ad1c-08d2f2cd6a9d_1000x740.png) ### **Other fun stuff** 1. **[Books Are Magic hat](https://www.booksaremagic.net/item/nQhV0zs_31GYHpOYUyRrDw/lists/LRuDNzlF1pHc/):** Signal to the world your love for books and Brooklyn bookstores. ![Burnt Orange - BAM Books Hat by Books Are Magic | Books Are Magic](https://substack-post-media.s3.amazonaws.com/public/images/52617310-06c0-47b4-ae69-02617ead0839_1200x824.jpeg) 2. ***[The New Yorker](https://www.newyorker.com/v2/offers/tnyagift01002?ot=%7B%220%22%3A%224e168c05-8cca-4eee-8e6f-d620129bf17d%22%7D&ct=0)*** **[subscription with tote](https://www.newyorker.com/v2/offers/tnyagift01002?ot=%7B%220%22%3A%224e168c05-8cca-4eee-8e6f-d620129bf17d%22%7D&ct=0):** Classy yet practical. 3. **[Small Works expertly framed limited-edition prints](https://smallworksprojects.com/collections):** “Small Works has made frames for the Whitney Museum in New York and mega-gallery Gagosian, and just about every major artist and art-hanging institution in the Bay, from SFMOMA to Tartine. They make the frames that make the art world swoon.” —*[The San Francisco Standard](https://sfstandard.com/2025/10/13/artists-collectors-obsessed-man-s-frames/)* 4. **[Kat and Roger pottery](https://cedarandhyde.com/collections/kat-and-roger):** “Their work combines classic shapes and graphic surface patterns with earthy natural clay textures.” ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/e4a6e396-a599-4a78-85e2-91c062b56706_1200x1800.jpeg) 5. **[Milk + Honey breast-milk jewelry](https://www.milkandhoney.jewelry/):** This may creep some people out, but it was a hit in our household. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/2caa0720-491b-4222-a230-b4c1c63adc81_2500x1667.webp) 6. **[Made In stainless-steel cookware](https://madeincookware.com/collections/stainless-clad):** We started with one pan and are slowly replacing every pan with Made In. 7. **Organic cotton clothing** from **[Wax London](https://waxlondon.com/collections/organic-cotton), [Faherty](https://fahertybrand.com/), [Pact](https://wearpact.com/),** **[Industry of All Nations](https://industryofallnations.com/collections/underwear)**, and **[Nads](https://nadsunder.com/):** My favorite clothing brands from [Justin Mares’s guide to living a long and healthy life](https://www.lennysnewsletter.com/p/a-builders-guide-to-living-a-long). 8. **[Zennyth startraveler gel pens](https://www.amazon.com/gp/aw/d/B0B45RTFQH/?_encoding=UTF8&pd_rd_plhdr=t&aaxitk=69cfb0e855eab9b2a8ac4f9b29f5a229&hsa_cr_id=0&qid=1763333232&sr=1-1-9e67e56a-6f64-441f-a281-df67fc737124&pd_rd_w=FGnU5&content-id=amzn1.sym.9f2b2b9e-47e9-4764-a4dc-2be2f6fca36d%3Aamzn1.sym.9f2b2b9e-47e9-4764-a4dc-2be2f6fca36d&pf_rd_p=9f2b2b9e-47e9-4764-a4dc-2be2f6fca36d&pf_rd_r=V0GT0B40KHS7MAX6C1H2&pd_rd_wg=GjPQZ&pd_rd_r=d88c156b-1ce9-4590-82f3-e70218b2dd67&th=1):** My go-to pens. I found them recommended on TikTok, of all places. You may have seen me holding one during my podcast. I also love [Muji pens](https://www.amazon.com/Muji-Point-Black-0-5mm-Japan/dp/B07B6W5668/ref=ast_sto_dp_puis). My wife loves [this notebook](https://md.midori-japan.co.jp/en/products/mdnote/) ([available here](https://komorebistationery.com/products/midori-md-notebook-a5?variant=47699751043355)). 9. **[Blissoma SPF 30 broad-spectrum mineral facial sunscreen](https://blissoma.com/photonic-spf-30-broad-spectrum-zinc-oxide-mineral-face-sunscreen-moisturizer-sensitive-skin-acne/):** My go-to (face) sunscreen. ![Photonic - Light Shifting Solution SPF 30 Broad Spectrum Facial Sunscreen + Moisturizer](https://substack-post-media.s3.amazonaws.com/public/images/c9cfbffd-ec73-49f9-a455-9c994216b4d0_882x985.jpeg) 10. **[Owala FreeSip insulated stainless-steel water bottle with straw](https://www.amazon.com/Owala-Insulated-Stainless-BPA-Free-Conquest/dp/B085DV8T75/ref=fabric-ww-slds-dp-fsdpnewarrivals-fa-xcat-fixed-desktop_d_sccl_1_1/136-3168637-1219309?pd_rd_w=S2fiw&content-id=amzn1.sym.bcfe8fa4-e659-4035-91f3-d1293f01d987&pf_rd_p=bcfe8fa4-e659-4035-91f3-d1293f01d987&pf_rd_r=8A0C24PN1N5EF08BC8XY&pd_rd_wg=7ykYB&pd_rd_r=aaa8c4a0-a9e2-4252-86ca-fc4ea5212dca&pd_rd_i=B0FDSL18S8&th=1):** My go-to water bottle. Make sure to replace the plastic straw with these [Koala stainless-steel straws](https://koalastraw.com/). ### **Returning champions** These are the products from past editions that drive nonstop thank-yous from readers for helping them discover, and which I also continue to love, use, and recommend: 1. **[Corrugated cardboard cutter](https://www.amazon.com/gp/product/B008RIS0UY?th=1):** Breaking down Amazon boxes will never be the same. ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/3bb78830-2a5a-40a6-ab59-e4012d523d1b_891x208.jpeg) 2. **[Manta silk sleep mask](https://mantasleep.com/products/manta-silk-mask?variant=45539961766041):** A twist! I’ve long recommended the [WAOAW sleep mask](https://www.amazon.com/WAOAW-Sleep-Sleeping-Blocking-Blindfold/dp/B09712FSLY), but I’ve recently switched to this because I found that the WAOAW left black marks on my pillow when I left it there for a bit, which doesn’t seem good. These Manta ones look ridiculous but work damn well and are really comfortable. You can go with [silk](https://mantasleep.com/products/manta-silk-mask?variant=45539961766041) or [their standard fabric](https://mantasleep.com/products/manta-sleep-mask-pro), or even [a pair with sound](https://mantasleep.com/products/manta-sleep-mask-sound-new-genhttps://mantasleep.com/products/manta-sleep-mask-sound-new-gen). For a simpler option, my wife likes [these](https://www.amazon.com/dp/B0CGR6M8VW?th=1). ![Image from A holiday gift guide for tech people with taste 🤌](https://substack-post-media.s3.amazonaws.com/public/images/8bbe4c64-63d0-4f20-9db3-8fdaa3ea00ac_1400x980.png) 3. **[Aura frames](https://auraframes.com/gifting):** Digital picture frames have historically been super-lame, but this one is different. It feels like an actual picture frame and looks good on your shelf. One delight feature is that your photos appear immediately when the reciever opens up the frame. A perfect gift for grandparents and in-laws. ![Walden 15" | Aura's largest, wall-worthy HD digital picture frame](https://substack-post-media.s3.amazonaws.com/public/images/aee24dae-0113-48e4-a5de-0cfcc7b211f7_800x800.jpeg) 5. **[Aarke stainless-steel hot water kettle](https://aarke.us/products/aarke-kettle):** I’m a tea guy, and I’ve tried a bunch of electric kettles. This has been the best by far. Not cheap, but beautiful, fast, and as plastic-free as I can find (there’s a small black plastic or silicone piece at the very top). [Aarke makes lots of other nice things too](https://aarke.us/). 6. **[Glerups slippers](https://www.glerups.com/collections/slip-ons):** I’ve been using the same pair of Glerups for years, and they still look great and feel great. [Wirecutter loves them too](https://www.nytimes.com/wirecutter/reviews/best-slippers/). We got a couple of extra pairs for guests (no shoes in the house, you animals). [They even make them for kids](https://www.glerups.com/collections/children). Many readers also love [UGG slippers](https://www.ugg.com/men-footwear/ascot-leather-distressed/1171296.html?dwvar_1171296_color=DDC), [Bombas gripper slippers](https://bombas.com/products/womens-gripper-slipper-bootie-sherpa-lined?variant=soft-taupe-heather&size=m), [Topdrawer Kolo house shoes](https://topdrawershop.com/collections/house_shoes), and [Birkenstock Zermatt Shearling](https://www.birkenstock.com/us/zermatt-shearling-felt/zermatt-cozyhomeshearling-woolfelt-0-latex-u_3732.html). 6. **[Electric tire/ball inflator](https://www.amazon.com/HOTO-Inflator-Portable-Compressor-Motorcycle/dp/B0CQ1RCJV7):** You don’t have to get this exact model or brand, but man, once you go electric, you don’t go back. Works with bikes, cars, balls . . . anything that needs to be inflated. Keep it in your car in case your tire pressure gets low. [Here’s a cute portable one](https://www.amazon.com/HOTO-Portable-Electric-Schrader-Motorcycle/dp/B0DCG1JGKN/ref=sr_1_4?crid=3SDNE7570J5M4&dib=eyJ2IjoiMSJ9.p8-0jxbvQPawwaBkY_2fKDcjgLn-EjwILcJh0vwXt5j9vVdhe8c3z4qHWI9ceViure2tKMsL3GMFHMY6icRhb9iGpp036FjxDM1cNAWaEb-x1LsMWmp5jIgbXaNVV0kK_OR8ipOdTmqtI0wPLlq0dxFC52F0ocbtwSZyXsKUdceYt0NA29y43tHXyWYiL4URjfGumSn2up3v_Iq4dIuy4vNE1lWuc5leSQhDsjJNBxw.bYERujJsLJEP7UBC7VeWb0OGN4MtrOEMX8yyisPBabM&dib_tag=se&keywords=hoto+tire+inflator&qid=1762984587&sprefix=hoto+tire+inf%2Caps%2C187&sr=8-4) from HOTO, but there [are other (less cute but cheaper) brands](https://www.amazon.com/s?k=electric+tire+inflator+for+bike&crid=30GQCVQINXI42&sprefix=electric+tire+inflator+for+bike%2Caps%2C186) that make these. [HOTO makes all kinds of other cool stuff too](https://hototools.com/). (Thanks, [Anirudh](https://x.com/anirudh_twt/status/1857536319922057605), for the tip.) 7. **[Owala kids’ insulated stainless-steel tumbler](https://www.amazon.com/Owala-Lian-Li-LANCOOL-MESH/dp/B0C768CWDF?th=1):** I haven’t found a better cup for toddlers. Good stocking stuffer. Tip: Replace the plastic straws with these [stainless-steel straws](https://www.amazon.com/dp/B07GQ7VR73). 8. **[DrTung’s Smart Floss](https://www.amazon.com/dp/B016790E7Q?th=1):** Another twist! I’ve long recommended [Cocofloss](https://www.amazon.com/COCOFLOSS-Coconut-Oil-Delicious-Dentist-Designed-Cruelty-Free/dp/B082BF4TYW?th=1),but with my focus on reducing microplastics, I saw [Bryan Johnson](https://blueprint.bryanjohnson.com/blogs/news/my-oral-health-protocol) recommend this, and now it’s my favorite. Makes a great stocking stuffer. ### **Some bonus recommendations from readers (thanks, all!)** 1. [Kodak Ektar H35 Half Frame film camera](https://www.bhphotovideo.com/c/product/1714484-REG/reto_project_rk0101_reto_kodak_h35_half.html) via [Robert Bye](https://x.com/RobertJBye) 2. [Map gifts](https://evanapplegate.com/map-gifts) via [Evan Applegate](https://x.com/youwillmakemaps) 3. [Analog starter kit](https://ugmonk.com/collections/analog/products/analog-starter-kit-black-steel-with-cards?variant=44404663222422) via [David Kyle Choe](https://x.com/davidkylechoe) 4. [Our Place Wonder Oven](https://www.amazon.com/dp/B0CLTK2JF8?social_share=cm_sw_r_cso_cp_apin_dp_5SGP9102CXDEW857S6NP&titleSource=true&th=1) via [Abby Grills](https://x.com/AGrillz) 5. [Hatsukokoro Kurogane Bunka knife](https://cutleryandmore.com/products/hatsukokoro-kurogane-bunka-41739) via [Madhav Shroff](https://x.com/Maddy_Shroff) 6. [Horl 2 knife sharpener](https://www.horl.com/us/en/sharpeners/horl2/) via [Jeremy Michael Cohen](https://x.com/jeremymcohen) 7. [Bearaby cotton weighted blanket](https://bearaby.com/products/the-napper) by [Claire Vo](https://x.com/clairevo) 8. [Nordace universal travel adapter](https://nordace.com/en/product/nordace-universal-travel-adapter-with-usb-charging-ports-type-c-usb/) via [Emily Ross](https://x.com/emilyross) 9. [Everyday cozy Larimer crew socks](https://www.smartwool.com/en-us/wool-apparel-and-socks/everyday-socks/everyday-cozy-larimer-crew-socks/SW001890.html) via [Hristo Vassilev](https://x.com/hristo_vassilev) 10. [Anker Prime power bank](https://www.amazon.com/dp/B0F66LNB8D) via [Nick Huber](https://x.com/nhuber) 11. [Prismacolor Premier colored pencils](https://www.amazon.com/dp/B00006IEEU?social_share=cm_sw_r_cso_cp_apin_dp_NZKV655CE78VNA07V309&titleSource=true&th=1) via [Jackie Bavaro](https://x.com/jackiebo) 12. [Guidecraft kitchen toddler tower](https://guidecraft.com/products/guidecraft-classic-kitchen-helper-stool) via [Marissa Goldberg](https://x.com/mar15sa) If there’s something else you absolutely love that people should know about, leave a comment 👇 [Leave a comment](https://www.lennysnewsletter.com/p/a-holiday-gift-guide-for-tech-people/comments) *Happy holidays! And have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [44/46] A year free of PostHog ($16,500 value): The all-in-one analytics, experimentation, feature flag, surveys, session replay, error tracking, data warehouse, LLM analytics platform *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[Fav AI/PM courses](https://maven.com/lenny) | [Fav public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *This Giving Tuesday, I’m partnering again with [GiveDirectly](https://givedirectly.org/lenny) and over two dozen other Substack writers to send cash to every family across three villages in rural Rwanda. If we’re successful, over **800 families** in extreme poverty will each receive $1,100, no strings attached, to spend on what they need most. **All donations today, December 2, 2025, will also be 1.5x matched** (while funds last). **Please donate 👇*** [Give now and get matched](https://givedirectly.org/lenny) ![Image from A year free of PostHog ($16,500 value): The all-in-one analytics](https://substack-post-media.s3.amazonaws.com/public/images/09ab6ac0-f9ab-4d44-8d79-9b2c3f2a6b0d_1457x971.png) I’m excited to welcome [PostHog](https://posthog.com/) to the [Lenny’s Newsletter Product Pass](https://lennysproductpass.com/) 🎉 **All annual Lenny’s Newsletter subscribers now get one full year’s worth of PostHog Scale** ***and*** **2x the free-tier credits, a $16,500 value.** This deal is available to *all* annual subscribers, and brings the total value of the [Product Pass](https://lennysproductpass.com/) to **over $25,000**. Simply absurd. *This is the first time PostHog has ever offered a deal like this, and you won’t find it anywhere else. Like all Product Pass deals, you must be a new PostHog customer to take advantage of the deal. See [Product Pass terms here](https://www.lennysnewsletter.com/i/168890236/important-offer-details-read-before-subscribing).* PostHog helps you build successful products by giving you literally every piece of SaaS that you need: **analytics**, **experimentation**, **feature flags**, **session replay**, **error tracking**, **data warehouse**, **LLM analytics,** **surveys**, and [more](https://posthog.com/products)—all in one platform. In the words of the founders, “We keep adding new capabilities at a pace that’s borderline unreasonable.” ![PostHog – We make dev tools for product engineers](https://substack-post-media.s3.amazonaws.com/public/images/0ba1efd8-6b7f-447b-bafd-d99042ab5d18_1200x632.png) The all-in-one platform strategy that PostHog is executing is really interesting to me. I rarely see it done this well, because it’s hard to do. But if you pull it off—like they are—you can build a generational business. Being able to follow an issue from a session recording, to its impact in analytics, to shipping a fix as a feature flag, to testing a variant, to collecting feedback with surveys—that’s the holy grail. Side note: I stupidly had a chance to invest in PostHog early on and passed because I didn’t think this strategy would work 🤦. I’m just happy to get to work with them on this collaboration now. Oh, they also just launched [PostHog AI](https://posthog.com/ai), which integrates with all of the products and lets you ask questions about your data, magically create dashboards, build surveys using natural language, summarize session replays, and more. [More how-to videos here](https://youtube.com/playlist?list=PLnOY1RYHjDfzBX5wsSUHwLj91xuGnH5Ci&si=WS2ElJSPw5UE7h-v). In addition to all the wonderful products PostHog offers, there’s so much to love about how the PostHog team operates: 1. Their [website](https://posthog.com/) is 🤌 2. Their [story](https://posthog.com/handbook/story) is inspiring. 3. Their [vibes](https://posthog.com/community/profiles/27732) are hilarious. 4. Their [company blog](https://posthog.com/blog) is one of the most genuinely useful and high-signal-to-noise company blogs out there. A model for how to do content marketing. Some examples: 1. [I’ve consistently underestimated how important communication is as a CEO](https://posthog.com/blog/ceo-diary-6) 2. [How we got our first 1,000 users](https://posthog.com/founders/first-1000-users) 3. [How to do startup sales with no experience](https://posthog.com/founders/startup-sales-strategy) 4. [Job interview questions engineers should ask (but don’t)](https://posthog.com/newsletter/job-interview-questions-engineers) 5. [How not to be boring](https://posthog.com/blog/brand) Over 300,000 companies use PostHog, including 65% of every YC batch, Lovable, Supabase, ElevenLabs, and even Y Combinator itself. Twice the free credits gives you tons of room to scale (90% of their users never exceed their already-generous free tier), and the Scale add-on gives you all the features you need to grow: priority support, white labeling, unlimited projects, HIPAA BAA, SAML, SSO + advanced permissions, audit logs, extended session replay retention, and more. If you’re already a subscriber, [redeem your free account here](https://lennysproductpass.com/). Otherwise 👇 Increasingly, your Lenny’s Newsletter subscription is becoming your product builder’s starter pack: **all the tools, guidance, and community you need to build, launch, and grow a world-class product (and career). For just $200-$350/year.** ![Image from A year free of PostHog ($16,500 value): The all-in-one analytics](https://substack-post-media.s3.amazonaws.com/public/images/418562e3-0649-407d-898f-d3774d9134a7_2912x1880.png) *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [45/46] How to spot a top 1% startup early *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[AI/PM courses](https://maven.com/lenny) | [Public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *Annual subscribers get a **free year of 19 premium products**: [Lovable, Replit, Gamma, n8n, Bolt, Devin, PostHog, Linear, Wispr Flow, Descript, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, ChatPRD + Stripe Atlas](https://www.lennysnewsletter.com/p/productpass) (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** After reading the early draft of this post, my approach to investing and giving career advice immediately shifted. Thank you, Bob, Cristina, Soleio, Rasmus, and Sean, for sharing your incredible insights with us 🙏 *For more from [Terrence Rohan](https://trohan.com/) (i.e. one of my absolute favorite seed investors), find him on [X](https://x.com/tmrohan?lang=en) and [LinkedIn](https://www.linkedin.com/in/terrencerohan/). Also, don’t miss his previous beloved guest post: [Raising a seed round 101](https://www.lennysnewsletter.com/p/raising-a-seed-round-101).* *P.S. You can listen to this post in convenient podcast form: [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from How to spot a top 1% startup early](https://substack-post-media.s3.amazonaws.com/public/images/1f1cf63c-0c3e-4ee0-942c-7f9392a3095d_1456x970.png)![Image from How to spot a top 1% startup early](https://substack-post-media.s3.amazonaws.com/public/images/b3237a8e-9fc8-4fc2-a20c-eae06d44b6e7_1178x338.png) There are endless posts and podcasts about how investors pick startups. But Lenny and I were curious about a quiet class of employees who seem just as good as—if not better than—the most famous VCs at spotting generational companies before they blow up. How do these rare folks keep joining world-changing companies before most of the world even notices them? To find out, we interviewed five people whose resumes include some of Silicon Valley’s most remarkable companies: **Palantir**, **OpenAI**, **Facebook**, **Stripe**, **Linear**, **Figma**, **Notion**, **Slack**, **Box**, **Spotify**, and **Dropbox**. ![Image from How to spot a top 1% startup early](https://substack-post-media.s3.amazonaws.com/public/images/2a5af163-9243-4a03-b132-feb10d3c7901_1638x1676.png) Each joined at least two of these companies early—an extraordinary feat, especially since they committed as full-time employees, not diversified investors. Their “hit rate” is phenomenal. **We were curious: What did they see? How did they choose? Are there lessons to take from their experiences?** Across their stories, we saw three distinct factors that mattered most. Though originally written for job seekers, these insights apply much more broadly—for founders, investors, or anyone trying to recognize greatness early. Here’s what to look for. ## **1. Ambition bordering on “ludicrous”** This was the most novel takeaway for us: One of the clearest markers of a future generational company, according to our interviewees, is **ambition**. It came up again and again. Bob pointed out that “**both Palantir and OpenAI were considered ludicrous when the companies were first started.** Joining Palantir in particular seemed like a very risky proposition. But actually it wasn’t risky at all. If the company failed, at worst, I’d wasted a year of my life and would have to go back to my PhD. But if the company succeeded, it would be life-changing.” Soleio “was surprised by the **ferocity** and **ambition** of the early Facebook team. They all seemed too smart to be working on ‘social networking’ but had a lofty idea for where the internet was ultimately headed and how Facebook might completely reshape it.” Sean added, “If a company’s thesis is marked by **extraordinary** **ambition**, it’s probably worth paying attention.” “The key tell,” Bob said, “was always the **ambition of the goal.**” Rasmus explained, “The logic here is simple: If everyone says, ‘Yes, that’s clearly a great idea, and you have direct competitors on day one, you are definitely late to the game. Even if you excel and go above and beyond expectations, the chance of making a meaningful difference in this world is small-ish. However, if someone has sailed across the sea of exploration, waded through the bog of research, and is still going on about an idea, there’s a small chance that they are ahead of the rest of us and see something I’ve yet to see.” #### **How to judge ambition** What are signs that the founder’s ambition is big enough? **The founder sounds a bit crazy.** Sean shared, “For Meter, to start from scratch to make your own hardware across many platforms . . . it really indicated that this company is taking a huge swing. Or they’re just **crazy**. Along those same lines, when I met Dylan Field [CEO and co-founder of Figma] on a bus to a developer conference in 2013, he was just starting Figma. I remember talking to him about what he was building, what he wanted to do with it, and how it seemed **completely insane** to do in the browser. It was clear Dylan was not going to stop, no matter what, and we know how that has gone.” **People laugh at them.** Rasmus recalled, “Spotify, a little group of ‘nobodies’ in Sweden, said, ‘Let’s build a catalog of all the music in the world and give everyone access.’ People **laughed at us**—I mean that literally, as in record-label officials **laughing at Daniel** [Ek, CEO and co-founder], asking them to give us a chance. The key here was that businesspeople thought it was a bad, bonkers idea while friends thought it sounded like a perfect future.” **No one has ever attempted this idea before.** Rasmus added, “It may be a bit of a cliché at this point, but **if** **no one else is trying to do what a company is trying to change in this world, then there might be something interesting going on**, **especially if it’s ambitious.”** To close, in the words of Bob, “At one point at Palantir, [co-founder and CEO Alex] Karp said, ‘**We want Palantir to be the most important company in the world**, not the most valuable one. But if it’s really important, it’s going to be valuable too.’” ## **2. Founders, founders, founders** Building on the above insight, every interviewee emphasized the same point: the founders matter above all else. Not as one variable among many—this was *the* variable. Cristina (early Stripe, Notion, Linear) said it outright: “**The founders (and early team)—nothing matters more than this to me.** I’m going to work hard, and I want to win, but I want to do it with people whom I want to see win too. When I joined Stripe, I joined more because I thought the people were special. I had more conviction about the company itself later.” Sean (early Slack, Box, Meter) echoed her: “**Quality (and authenticity) of founders** have always been the most important variable to me.” Rasmus (early Spotify, Figma, Dropbox) distilled it even further: “**People and mission.** Who and why (not as much ‘how’).” Bob (early Palantir, OpenAI) added, “The common pattern was an incredibly **ambitious goal combined with a credible team.”** There’s that ambition again. #### **How to judge founders** Joining a company with great founders is easy to say and hard to do. What exactly makes a founder great? Many people call out intelligence, and we have a whole section highlighting the importance of ambition. But our interviewees mentioned some less obvious traits. 1. **Ability to adapt** Bob pointed out that “both Palantir and OpenAI started with an unworkable initial strategy that perhaps the world was correct to mock. But both iterated over time to something that worked. **It’s more important to be able to learn quickly than to have a good strategy.**” Cristina shared the same sentiment: “It’s so difficult to build a company, and there’s so much you need to learn going from one stage to the next and scaling. **Founders who truly love to learn and look at company-building as a learning experience are quite predictive of whether founders will build durable, special companies.** I remember saying early on at Stripe that Patrick and John [Collison, co-founders] didn’t just want to build a great product; they wanted to build a great company. And in looking back, I remember their apartment had books piled to the ceiling—they knew so much about so many different topics, and they always sought advice from others. They had a learning mindset, and I think they still do.” Sean described this as “**rate of change**.”Soleio called it “**clock speed**,” adding, “The common thread I’ve observed with all of these companies is how fast they operated and how extraordinarily capable their early employees were. Are they building their ideas in real time or does it take months to see their visions crystallize in software?” 2. **Ferocity** Rasmus looked for “clear **passion** from the people who are ‘the company.’ Passionate and interesting people have been a core aspect of my professional life.” Cristina said she “loved Ivan’s [Zhao, founder and CEO of Notion] **intensity**” and has always been drawn to “intensity plus intelligence.” And as you’ll recall from above, Soleio “was surprised by the **ferocity** and **ambition** of the early Facebook team.” 3. **Founder-market fit** Sean always asked himself, “Do these people seem like they’re doing what they’re meant to be doing, and is there no question that no one else can do it as well as them?” Bob was optimistic about Palantir’s chances because **“**it was founded to take the ideas behind the analysis software that solved fraud at PayPal and apply them to the intelligence community.” To close, in the words of Cristina: “If the three most important things in real estate are location, location, location, **the three most important things in startups are people, people, people**.” ## **3. Judging today’s product is a trap** You’d think that for legendary product companies like Stripe, Figma, Slack, Notion, and Facebook, you could tell how special they were going to be by how good their early product was. It turns out this way of thinking is a trap. Soleio said that when he first logged in to Facebook, “**I remember being disappointed**. The version their team had described was light-years ahead of what I saw that day.” Likewise, Figma was **more prototype than product** the day Dylan laid out his vision to me for building a collaborative design platform.” Cristina had a similar perspective: “Many of the companies I’ve joined were developer products or products that were meant for teams, so I couldn’t truly try the product myself, as I’m not a developer or didn’t have a team use case for it. So in general, **I discount my own thoughts about a product in those cases.**” Sean told us that **“in the earliest days of Slack, it was rough around the edges**. **To quote Stewart [Butterfield, co-founder], [it was a giant piece of shit](https://www.youtube.com/shorts/RaFX5rnKKtY).** The bulk of the vision was there in that beta period from 2013 to 2014, but still awaiting refinement.” Rasmus rightly pointed out, “**Almost every product I’ve worked on started out as one thing but was something quite different at the time of consumer success.** Spotify was going to be a video streaming platform and Figma a meme generator.” So even if the product is great, it may have to radically change anyway. #### **How to judge the product** Instead of focusing on the product today, focus on the mission, the magic, and customer love—in other words, where the product is *going*. **1. Mission/vision** Rasmus explained that “**the mission of a company is unlikely to change** and it’s the reason we all show up, so you’ve got to be down with whatever that is.” Soleio said, “**The vision the Facebook team had described a few hours earlier was light-years ahead** of the pre-Photos, pre-News Feed version of Facebook I found that day.” Likewise with Figma. “In both cases, special companies have a crisp articulation of how future systems should look and how they break the mold of today’s user expectations. **Their products, especially at the earliest stages, lag far behind their ambitions**.” There’s that word *ambition* again. Bob recalled that he “didn’t initially have a lot of conviction that either Palantir or OpenAI would be world-changing. But I had conviction that **they were doing something interesting** **and that I would learn a huge amount in one year if I was pushing on the right goal**.” **2. Strong customer pull** Instead of using your own judgment, ask what customers are saying about the product. Cristina told us that “with Stripe, **I never asked to even see metrics. I was instead observing what developers said about the product** in forums I could access (Hacker News, Twitter, etc.). I deferred to others (e.g. I asked a friend who was a data scientist at Watershed why they switched over to Linear from a competitor) to better understand the product from their perspective. **Even if it was early and the market was niche, I cared that people loved it enough to share it.**” She added, “Find a company that’s building a product so good that its users can’t stop talking about it.” Soleio said that when he told his brother he was meeting the team behind Facebook, his brother replied, “Oh dude, Facebook is bigger than God out here! It’s blowing up.” Wisely, he listened. **3. A demo that leaves you feeling like you just glimpsed the future** Finally, Soleio had a fantastic metaphor for what to look for: “There’s an iconic scene in the original *Jurassic Park* where the tour stops so John Hammond and his guests can witness a raptor hatching. The group crowds around as a dinosaur emerges from its shell, while Hammond gushes about always wanting to be present for births. Even in this slimy, harmless form, you sense the world has shifted—an apex predator has entered it. **I call these ‘*****Jurassic Park*** **moments’ at startups: early prototypes or demos that leave you breathless and feeling like you are being given a clear glimpse of the future. Even though it’s ugly and small, you can sense this product will grow up and one day eat things**.” ## **Now it’s your turn** When looking for your next role, it is interesting to note that none of these people found their jobs through applications or recruiters. Their paths were unpredictable, even serendipitous. Cristina met Stripe’s Patrick Collison at a barbecue. Rasmus got a cold call from Spotify’s Daniel Ek in 2006. Soleio was invited to Palo Alto after winning an online design award. Bob already knew half of Palantir’s founding team from PayPal. ​Sean [replied to Stewart’s Twitter thread](https://x.com/seanrose/status/482659147410706432?s=20) about a feature request. “You can’t manufacture luck,” Sean said. “But you can optimize for serendipity. Be genuine. Use what they’ve built. Reach out with real interest.” Rasmus put it simply: “Say yes to people who make you curious. That’s where the best adventures start.” When an opportunity does present itself, remember the golden advice of these illustrious early employees. 1. Look for ambition—does it ooze from the founders, team, and product roadmap? 2. Study the founders. Are they not only intelligent, but fast learners, with an undeniable passion for what they are building? Does it seem that they are uniquely fit for building this? 3. And the product—if it is rough, do customers love it? And does it point to something bold and in the new future? If you answered yes to these questions, you might have just found the next rocketship. *Thanks, Terrence! For more, find [Terrence](https://trohan.com/) on [X](https://x.com/tmrohan?lang=en) and [LinkedIn](https://www.linkedin.com/in/terrencerohan/). Also, don’t miss his previous guest post: [Raising a seed round 101](https://www.lennysnewsletter.com/p/raising-a-seed-round-101).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 --- ## [46/46] How to build your PM second brain with ChatGPT *👋 Hey there, I’m Lenny. Each week, I tackle reader questions about building product, driving growth, and accelerating your career. For more: **[Lennybot](https://www.lennybot.com/) | [Lenny’s Podcast](https://www.lennysnewsletter.com/podcast) |** **[How I AI](https://www.youtube.com/@howiaipodcast)** **| [Lenny’s Reads](https://www.lennysnewsletter.com/s/lennys-reads)** | **[AI/PM courses](https://maven.com/lenny) | [Public speaking course](https://ultraspeaking.com/lennyslist?via=lenny)*** *Annual subscribers get **19 premium products for free for one year**: [Lovable, Replit, Gamma, n8n, Bolt, Devin, Wispr Flow, Descript, Linear, PostHog, Superhuman, Granola, Warp, Perplexity, Raycast, Magic Patterns, Mobbin, ChatPRD + Stripe Atlas](https://www.lennysnewsletter.com/p/productpass) (while supplies last). **[Subscribe now](https://www.lennysnewsletter.com/subscribe?).*** Someone smarter than me once said, **“AI won’t replace you, but a person using AI better than you might.”** I believe this is exactly right. Right now, we all need to be building the skills that help us become that person using AI better. Lucky for us, Amir Klein is already that person and has written a guide for the rest of us. Though it’s targeted at product managers, the advice and workflows can be implemented by anyone in any function. Thank you, Amir, for giving us a glimpse into the future and the concrete steps to get there. *For more, follow Amir on [LinkedIn](https://www.linkedin.com/in/amir-klein-9b8444189/). You can also listen to this post in convenient podcast form on [Spotify](https://open.spotify.com/show/0IIunA06qMtrcQLfypTooj) / [Apple](https://podcasts.apple.com/us/podcast/lennys-reads/id1810314693) / [YouTube](https://www.youtube.com/@lennysreads).* ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/c33617f9-bda1-49fb-ad65-965d4cd30017_1456x970.png) The first month in my new role at monday.com, I was tasked with building our first AI agent. The goal was to create an AI co-pilot, something users could turn to for insights, explanations, or building complex workflows they wouldn’t know how to create on their own. To build that, I needed a ton of context—all the internal knowledge, decisions, assumptions, and scattered inputs that shape any product direction. And gathering all of that felt completely overwhelming. I was drowning. All that context lives everywhere: Slack channels, Notion pages, Monday boards, decks, Google Docs. Hundreds of tiny fragments I could never quite piece together. I kept running into mental blocks, forgetting what I knew from where, and getting stuck. Instead of trying to keep all of that context in my head like I always had, this time I wanted to try something new. I dumped everything I had into a ChatGPT Project, word-vomited all that was on my mind, and asked if it could help me get started. And boy, did it. Finally, I felt like I could smell a roadmap on the horizon, a direction was forming, and things began to click. Even better, I felt *somewhat* in control without being stressed about storing everything in my head. I could store it in the AI instead—a second brain. Instead of all that information overloading my own brain and pulling my attention in a hundred different directions, I could finally focus on the product work I love and need to get right to be successful: understanding the problem, shaping the vision, and building something meaningful. My good friend Tal taught us [how to think with AI](https://www.lennysnewsletter.com/p/build-your-personal-ai-copilot). I’m building on Tal’s post by showing what happens when AI becomes an extension of your mind—when it carries your context, grows alongside you, and ultimately amplifies what you’re capable of as a PM. ## **Context is important—but comes with a heavy mental load** No matter *what* we’re doing, we’re constantly trying to hold way too much information in our heads. I always imagine it like carrying a giant basket filled with random things like eggs, water bottles, watermelons, toy cars, a cactus (I hope you’re picturing a Dr. Seuss scene). And I’m on the go, so things are rocking all around the basket, and more things keep being added, and then an egg falls and cracks, one of the water bottles starts to spill over, a toy car keeps banging into one of the watermelons . . . basically anxiety in a metaphor. ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/5bcfe94d-b894-40b4-b540-f3d151ec6934_1452x1021.png) That’s what it feels like trying to hold all the context required to do product work. But the hardest part isn’t just *carrying* it; it’s that none of these pieces arrive neatly fitted together. Context comes in fragments: user feedback, metrics, market changes, internal constraints, past decisions, intuition. As PMs, our job is to assemble those pieces into a clear picture—shaping the problem, forming the hypothesis, and defining the solution space. When you can pull that together, you build products that solve real problems so well that customers change habits for them, pay for them, and genuinely feel the impact. But doing that synthesis in your head, and doing it over and over again, can literally feel impossible. That’s where AI comes in. When you feed in all of that context that you’ve been trying to juggle yourself, your ChatGPT Project becomes a second brain that can store the information and synthesize it for you. That means that it can know and retrieve the right piece of data for the right problem—like an instantaneous librarian—and even use what it knows to run analyses and generate recommendations, like an associate PM. It’s important to say: using a second brain doesn’t dull your role but actually sharpens it. Your reasoning, product sense, knowledge, and taste are still doing the real work; AI just amplifies them. You can’t outsource judgment or creativity. This isn’t AI thinking instead of you—it’s you thinking with more clarity because all the mental overhead is gone. You still make every decision; the second brain just clears the path from your insight to the output. ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/9c40ace8-e4df-4d0b-bd26-d88744f0404e_1456x970.png) #### **Step 1: Create its personality** If someone were to ask you, “Hey, do you want a really smart, eager, motivated, capable person by your side who knows what you know and is super-enthusiastic to tackle anything you want?” you’d probably say yes in a heartbeat. Well, you’re in luck, because that’s exactly what Projects can be. At monday.com, I had all of this information (my ever-growing basket) that I was in dire need to create a plan from. So I turned to ChatGPT, opened a Project, and got started. Cue the music. If you’re trying this yourself, Projects live in ChatGPT’s left sidebar under “New project.” ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/0e557423-fe2a-4492-adc3-0cb032bec4a3_1456x815.png) *(I’ll share how to do the same thing in Claude and Gemini later on.)* Once you’re inside, if you approach Projects like a second brain that you’re growing, you want to first make sure it thinks in the way you think. In other words, you need to build its personality. Each Project has an instructions section where you can first define this “personality” in plain language. ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/1ddada6e-be58-48ce-a679-be0826af983a_1456x648.png) A really awesome way of doing this is with the help of ChatGPT (of course). Open a new chat and describe what you want. For example: > `I’m a Monday PM working on AI agents. I’m building a ChatGPT Project to be my thought partner, something that’ll work with me on my initiatives, something that’ll know how to challenge me in all the right places, push back on areas that feel weak, and creatively think of alternatives with me. This Project’s “personality” has to be sharp, smart, fun, and not always agreeing with everything I come up with. It also needs to be a pro at product management—this includes product sense and product execution, with a strong sense for product taste and delight. Can you help me write the instructions for this project? :) Cheers!` It’ll output instructions for what you want: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/23039a59-4ae2-4571-bfdf-5e1ea9bbffa3_1456x725.png) Tweak what you want or just copy the whole thing as is into the instructions page, and your second brain is now eagerly waiting for you to feed it information! ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/79a93831-a93b-4a7d-8f4e-537ca24b0d2a_1456x685.png) #### **Step 2: Feed it information** Now comes the fun and legitimately relieving part: feeding it all that context that you are barely holding onto. Go to “Add files” and just dump it in. ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/a2b9e072-f56b-4419-aa5e-db1e205ac768_1456x592.png) I think the biggest “whoa” moment for me is realizing that *everything* is essentially text. The classics, like PRDs and docs, are a given, but decks, websites, Excel/CSV sheets, dashboards, and Slack channels all contain text too. They just need to be exported into PDFs and then you can upload them into the files section too. Once you see everything as text, you start to understand how much context you can actually give your second brain. Here’s what that looks like in practice: I’ll scroll through a massive Slack channel that’s gotten impossible to navigate. I’ll export it or, if that’s not possible, I’ll copy and paste the entire channel’s contents into a doc and save it as a PDF. Then I drop it into the Project. Now ChatGPT knows what has already been discussed, what decisions were made, and what issues keep resurfacing. When I need it to understand our product’s capabilities, instead of rambling on to it about how my product works, I go to our support or docs site, hit Command + P, and save the entire page as a PDF to drop in. That way, whenever I mention a feature, my second brain already knows how it works. I do the same with research data like transcripts, interviews, surveys, CSVs. Everything becomes fuel. Each file adds depth to the brain. In the case of the first initiative I led at monday.com, I started with a few decks that different colleagues of mine had made, downloaded PDFs of monday.com documentation pages explaining how specific things work, and added a bunch of CSV files containing Reddit threads of conversations thousands of people had about monday.com in relation to AI and our competitors. This was enough to get the ball rolling and start formulating a plan. The beauty of Projects is that you’ll start creating new artifacts from it. Whatever it is you’re creating—whether it be PRDs, overview docs, or strategy decks—once finished, you can take those, put them in a doc, download them, and feed them back into your second brain. Each new thread you open will be up to date with you on your work. It becomes a living thing. This is what my Project ended up containing after hundreds of threads: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/14317a45-0720-4493-be0f-4443c227e688_800x444.png) #### **Step 3: Let it cook** There’s no one specific thing to use Projects for. As your second brain becomes more and more knowledgeable about your work, you can lean on it for everything that you don’t want to do but needs to get done. For example: **1. Sign-up forms** I needed to create a sign-up form for users to get early access to the agent we were building. This is a classic case of something that seems pretty easy at first but ends up making you bang your head against a wall. How should I phrase what I want to ask? How do I make the output from the form clear and purposeful for me without making filling out the form exhausting for users? Your second brain can now swoop in to save the day as it holds all the context on your initiative. In this case, I asked: > `I’m sending out a form to users to sign up for a waitlist for our first agent. I want to put 2-3 questions on the form which gauges their expectation to ensure we’re aligned on what we’re building and to receive another level of verification around the pain point. These questions should be concise, and the user’s answer will be open-ended (free text).` My sign-up form looked like this: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/026eb721-00ac-4ab7-8220-306932bffc91_800x542.png) **2. Prototypes** Sometimes, when I’m deep in a conversation with the Project, we’ll hit a moment where the idea feels stuck in theory. We’re talking about flows, interactions, edge cases, but it’s all abstract. So I’ll say: > `“Let’s make this real. Write me a prompt for Lovable that builds this experience.”` I’ll drop in screenshots from Figma, highlight the areas I’m referring to, and describe how I want the behavior to work. > `“This panel should expand on hover. This tooltip should guide users through setup. Keep it lightweight, just enough to simulate the flow.”` The Project uses those images as a reference, the entire context it has from its lifespan, and it generates a Lovable prompt I can copy straight in. Ten minutes later, we’re looking at a working prototype. Here’s an example of a Lovable prototype that I was able to create literally within seconds, thanks to all the context my second brain held: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/b8b9b2eb-b964-447b-88b3-501a2f6cc9aa_800x835.png) And this is what we ended up shipping: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/9d4ac194-d2bc-4d5f-a0d0-7e5a934e32da_800x784.gif) It’s not just a “wow” moment—it’s a workflow. A prototype like this lets me gather my designer and engineer, show them the idea, and get feedback *before* anyone has to build. This workflow fits so natively into what I’m doing, and truly increases my productivity and the productivity of the whole team so tremendously, that it’s become one of my favorite things to do. **3. Tailored communications** When you zoom out, PM work is full of moments where you need to communicate the same initiative from a completely different angle. It happens daily. Sometimes hourly: - Marketing needs a short blurb for an announcement. - Leadership needs three crisp sentences for a QBR. - Enablement needs a deck that makes the value obvious to sales. - Your designer needs a user-testing script that speaks to the problem, not the roadmap. None of these tasks are hard because of the writing itself. They’re hard because you have to mentally reload the context, remember what matters to each audience, and then translate the same initiative in so many different ways. That reload cost is enormous! My second brain removes that step entirely. I just word-dump everything I want to be highlighted for the respective request (the overview, the goals, the nuances, what I believe is most important for which audience), all as unstructured and messy as you can imagine, and it returns tailored versions for each audience in seconds. I’m still the one making the calls, but the friction between *what I know* and *how I need to express it* disappears. Suddenly the tasks that used to eat half a day take minutes. Once you get into the habit of constantly adding more files to your second brain and opening more threads, you’ll become comfortable with your tailor-made companion by your side. Next, you’ll wonder how you did anything without it. Your second brain becomes such a powerful tool that sometimes you can even forget you aren’t speaking to a human being. You might end up having a moment like this: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/cdf76adf-ccf2-4e05-90cb-1a9d42b4ffb3_1456x380.png) Creepy? Maybe. But also comforting. ## **Context is the new interface** As PMs, context is the raw material we think with. It’s how we spot patterns others miss. It’s how we distinguish noise from signal. It’s how we turn a vague cluster of symptoms into a crisp problem statement that we rally around. When all that context lives only in your head, your output is limited by your working memory—the five or six ideas you can juggle at once. When the context sits outside your head, intact and retrievable, your thinking expands. You’re able to see the whole picture instead of just the part you can remember in the moment. *That’s* the breakthrough second brains give you: You don’t just remember more. You reason better. At monday.com, we saw the same pattern for months with our customers: they loved the *idea* of AI in our product, but they had no idea how to start. They didn’t know where to use it, how to set it up, or how to fit it into their workflow. Our team had user interviews, market research, feedback threads, product data about what was going on with our customers—I can feel my heart racing faster and faster just writing about it. We had a beautiful, massive, messy cloud of information, and before creating my second brain, I’d have spent days drowning and reassembling all of it in my head before I could form a clear hypothesis. Within minutes of my blabbering everything on my mind to Projects, it surfaced the key pain point for our users: they weren’t blocked by capability but by confidence. They needed scaffolding, not features. That insight alone unlocked the product direction we eventually shipped—and it was a huge success! The more I worked this way, the more obvious it became that your second brain amplifies you and your skill sets. Once it holds the same context you do—your information, but also your reasoning, your decisions, your instincts—it becomes an extension of how you think. AI isn’t replacing what I do. Rather, it’s expanding my capacity to do it well. ## **Appendix: Using Claude or Gemini as your second brain** Everything I described above works beautifully with ChatGPT Projects, but the principles don’t belong to any one tool. If you prefer Claude or Gemini, you can build the exact same second-brain setup there too. The workflow is almost identical: give it instructions, feed it context, and let it grow with you. In Gemini it’s called Gems, conveniently found on the left-hand panel: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/6ecf35e3-2652-4917-a5ef-0b341df33cb1_1456x826.png) When you open a new Gem, you put in the custom instructions: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/66fb4557-9965-4024-8d61-0c01064208a4_1456x826.png) In Claude it’s called Projects, also found on the left-hand panel: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/2a23f8c3-ca66-451e-9161-6ee366f07b46_1456x764.png) Click on “New project”: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/a943348e-6559-4ab6-85af-6d27fade6985_1456x301.png) Once in, name it and give it a description for future use: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/70e2a43d-3b41-42f8-8564-bad80f3c5dc4_1456x764.png) Like the other AIs, provide it with custom instructions from the right-hand Instructions component: ![Image from How to build your PM second brain with ChatGPT](https://substack-post-media.s3.amazonaws.com/public/images/29ab58cb-26cb-4195-86f4-d19f0f382583_1456x764.png) PM away! *Thanks, Amir!* *For more, follow Amir on [LinkedIn](https://www.linkedin.com/in/amir-klein-9b8444189/).* *Have a fulfilling and productive week 🙏* **If you’re finding this newsletter valuable, share it with a friend, and consider subscribing if you haven’t already. There are [group discounts](https://www.lennysnewsletter.com/subscribe?group=true), [gift options](https://www.lennysnewsletter.com/subscribe?gift=true), and [referral bonuses](https://www.lennysnewsletter.com/leaderboard) available.** Sincerely, Lenny 👋 ---