For the few who outgrow the platform's memory. A free database the AI can read and write over the web, holding a note on where you left off, the current state of the work and a running log. What it changes, what it is not, and a test prompt to see whether the door is open for you.
Read the full chapter →Free to read on The Conversation Daemon, where every piece of the book is published.
AI chat tools like Claude come with their own memory: project files, a summary of recent chats, and search over old chats. For most people that is enough. This piece is for the few who need more.
It names the point where platform memory stops working. A long project where only the current version matters. A note for the next session that has to be read first. A list of your own decisions, in order. Files you update by hand go stale, and the AI may still read the old one.
The answer the writer chose is a small database the AI can read and write over HTTPS. He used Cloudflare's free tier. A footnote compares it with Supabase, Firebase and Vercel, and names the catch in each. Before building anything, he tested the connection from inside a chat with network access turned on, using a public echo service, up to a ten megabyte payload.
With the database in place, three things change. A carryover note survives the end of a session. The current state of your work lives in one row instead of many files. And you can keep an ordered record of decisions and notes you can search later.
The piece is clear about limits. This does not give the AI memory. The AI still starts fresh each chat, and the old habits from the previous chapter still apply.
The AI is not remembering for you. The database is.
The continuity is in the database, not in the AI.
What the platforms keep, what you control, and what to build when that is not enough.
Why does your AI forget? Because it never remembered. Each message, the platform assembles a fresh stack of text inside a fixed budget, and what feels like memory is whatever made it into the stack. A plain walk through the layers: the budget, the librarian who fetches from your files, and the daily summary that drifts.
Which levers you can pull and which you cannot. Your instructions, your files, project memory and the tools the model calls. When to start fresh, and why anything with numbers should go through code. Small, unglamorous habits that keep AI useful on long work.
The same question through three AI models, and the same voice came back each time. The wording lives in the model, the character in your files, and the meeting only when you show up. Whatever you keep looking after is what survives the next model.