Originally written in French. Translated by AI — the meaning has been preserved, not the prose.
The book is out. It starts from a simple idea: a piece of work has a history — a goal, sources, choices, disagreements — and an AI only becomes useful if that history is held somewhere.
Today, what you do with it disappears along with the conversation. The subject, the documents, what you had decided and why. Next time round, you start over. The cost isn't the retyping: it's that the reasons never come back — the option you had ruled out, the objection left open, why you decided the way you did.
I ended up giving it that place. One space per subject, holding the goal, the documents and where they came from, the decisions, the disagreements, and where the work stands. The AI draws on it in every session: it knows where an idea came from, recalls a choice, brings back an objection I had raised three weeks earlier. The work resumes where it stopped. And six months on, it's still useful, for another subject.
That is what I call a Context. The book tells how it came about, what it changes, and how to build one.
What you'll find in it
The system comes out of a mundane problem: five hundred emails from a homeowners' association and no way left to answer "where do things stand on the roof?". Then it gets to work — on a call for tenders, on an article that starts from a vague idea, on a product question that takes months to mature. And by the end, you know how to build one.
Who it's for
To the people who spend their days turning information into decisions and documents: product managers, consultants, writers, researchers. The ones who already work with an AI and sense that something gets lost between two conversations.
It isn't a method to apply or a tool to install: it's the story of how something got built, with the mistakes that gave it its shape.
It was written that way
To write this book, I didn't open a word processor: I opened a Context. Everything stayed in it — the sources, the chapters, the writing decisions, the cut passages and the reason for cutting them.
When I came back to a chapter after three weeks away, I didn't reread it: I asked where I stood and what we had decided to leave out. The translations came out of the same place, without my having to go back over the text by hand.
The book is therefore the proof of what it describes.
Get it here
It's on this page.