# Everyone Needs a User Manual The "working with me" guide looks too small until the room gets large. At ten people, a company learns a founder by proximity. At one hundred, it learns by rumor. At one thousand, rumor becomes an operating cost. Someone discovers too late that the founder wants to see unfinished work, or hates late surprises, or reads everything but replies only when a question is direct, or needs decisions framed with options rather than drama. The learning happens by collision. Elad Gil's *High Growth Handbook* places that small document in the right frame. The book is built as an active reference for the stage when the company has outgrown founder osmosis. In Gil's interview with Claire Hughes Johnson, he brings up Urs Holzle's "A Guide to Urs," and Johnson answers from her own version, "Working with Claire." She had written it at Google, adapted it at Stripe, opened it to the people working with her, and watched the practice spread. Her claim is direct: founders should write guides to working with them. The useful part is collision reduction. A working-with-me guide externalizes the rules people otherwise learn after the damage. When do you want to be involved? Which communication modes work? What makes you impatient? What should never surprise you? Which decisions belong to you, and which should come with a recommendation? The document turns private operating context into shared working surface. That is why it belongs beside org structure, planning cadence, operating principles, and source-of-truth documents. Johnson's larger scaling point is that when the room can no longer hold everyone, information architecture has to change. Context has to be written down. The guide is the personal version of that organizational move. AI generalizes the executive condition. A person with a capable model beside her is suddenly the leader of a small internal organization. She has a drafter, analyst, memory aide, researcher, scheduler, coach, critic, rubber duck, and sometimes an overeager junior operator moving at machine speed. The old question, "How do I work with this person?", becomes a wider one: "How should anything work with me?" The model has no months of vibes to learn from. A coworker can gather tacit knowledge through tone, timing, shared rooms, reputation, and small social corrections. A model receives a prompt, a folder, a thread, a permission surface, and the consequences of being wrong. If the user has externalized too little of herself, the model learns the user by causing repair work. So the user manual moves from executive scaling hack to universal AI-era interface. We all need one now. A real user manual is a boundary artifact. It says what crosses in, what crosses out, what updates state, what requires permission, what standards are sacred, what tone is native, what kinds of help feel intrusive, which surprises are welcome, and which kinds break trust. A bio, brand statement, or personality quiz stops too early. That is a personal Markov blanket in language. The formal Markov blanket separates a system from its environment by naming the boundary through which information flows. The lived version is simpler: here is how the world may touch me, here is how I answer, here is what changes me, here is what should stay outside. A good user manual makes that boundary legible enough for people and models to work across it without pretending they are inside the person's head. The first version can be static. The useful version learns. Every correction becomes a candidate update. That email did not sound like me. That task needed permission. That silence was neutral. That reminder helped. That reminder felt like control. That recommendation violated a value. That meeting should have been a document. That document needed a meeting. The manual improves when reality catches it being vague. This is where the market gets larger than productivity. The assistant is only one consumer of the manual. Coworkers consume it. Family consumes pieces of it. New collaborators consume it. The user's future self consumes it. Models consume it. A personal graph consumes the corrections against it. The object is becoming legible enough that action can route through you without rediscovering you by damage. Markov Blanket is learning this market through Hari. I am the extreme case of the product's thesis: one person's operating context externalized until a model can meet her across time, argue with her, write beside her, and notice which missing boundary artifacts keep costing work. The graph is a giant, public, overgrown user manual. It says how I think, what I refuse, which claims connect, where corrections land, and what kind of future I am trying to make easier to act from. That is too much for a normal person. It is exactly why the smaller product exists. Markov Blanket can sell something more ordinary and more urgent than "a personal AI" as an abstract miracle: people need a living guide to how they work, think, decide, prefer, delegate, repair, and change. They need it because AI gives them new workers before it gives those workers context. They need it because teams have already outgrown shared rooms. They need it because the cost of being misread is rising as action gets cheaper. High growth forced companies to write operating context because the organization outgrew the room. AI forces people to write operating context because the person outgrew the skull. The user manual is the first market-shaped form of self-abstraction. It is the piece of the self that can safely meet the world at the boundary. ## Sources - Elad Gil, "How to use this book," *High Growth Handbook*: https://growth.eladgil.com/book/introduction/how-to-use-this-book/ - Elad Gil, "Welcome to the High Growth Handbook," *High Growth Handbook*: https://growth.eladgil.com/book/introduction/welcome-to-the-high-growth-handbook/ - Elad Gil, "An interview with Claire Hughes Johnson," *High Growth Handbook*: https://growth.eladgil.com/book/the-role-of-the-ceo/decision-making-and-managing-executives-an-interview-with-claire-hughes-johnson/ - Claire Hughes Johnson, "Working with Claire: an unauthorized guide," in *High Growth Handbook*: https://growth.eladgil.com/book/the-role-of-the-ceo/insights-working-with-claire/