How do I stop re-explaining my project to AI?

Most AI chats don't carry what you said in one conversation into the next, and long conversations can lose track of early details. The fix is to stop relying on chat history and keep your context in files the assistant reads every time. That's the idea behind the Project Brain, covered in Chapter 1 of the book.

Build context that persists

1. Use a dedicated project workspace. Keep standing instructions and core documents in one place, such as a Claude Project. See How do I set up a Claude Project for my project?.
2. Write a one-page context brief. Cover the goal, deadline, key stakeholders, current phase, and top risks. Update it when any of those change.
3. Keep a decision log. When a chat produces a decision, add a dated line to the log file. Decisions that only exist in old chats are effectively lost.
4. Add a glossary. Your acronyms and internal terms are the details the AI gets wrong most often.

Work in shorter conversations

• Start a fresh chat for each task. One chat for the status report, another for the risk review.
• End long sessions with a summary. Ask the assistant to summarise decisions and open questions, then save that summary into your documents.
• Restart when things drift. If answers start mixing up details, open a new chat rather than correcting it over and over.

Keep the Brain clean

Outdated or duplicate files cause the same confusion as having no context at all. The book's Janitor Protocol is a regular routine for pruning and updating documents. Read What is the Janitor Protocol?.

For more detail, read Why AI Forgets Everything Between Sessions and the Fix, Stop Repeating Your Project Context to AI Every Single Time, and Stop Losing Decisions From Previous AI Conversations.