An AI model's memory is limited. A developer must save important information to a separate file. This file helps the developer and the AI model continue work effectively.
By Pawan Nayar · from the LLOS AI Learning record
An AI model's memory is limited.
A developer must save important information to a separate file. This file helps the developer and the AI model continue work effectively.
When an AI model's session ends, all past conversations are gone. Decisions made, problems found, and solutions tried are lost. The next session will not know the previous work's details. A developer should update a documentation file regularly. This file should explain the current state of the project. It should not just list actions taken. Update the file every few hours or at important points. These points include before a session ends or after a big decision. Write updates when the information is still clear in mind. A good update explains why things are a certain way. It tells what works and what does not. It also states what to do next and what to avoid. If information in the file is no longer correct, rewrite it. Do not just add new notes below old ones. The file should always show the current truth. This practice is like writing comments in code. Comments explain why code is written a certain way. The documentation file explains why project decisions were made.
A developer works with an AI model for five hours. The developer fixes a complex bug. The developer then ends the session without updating the project's documentation file. The next day, the developer starts a new session with the AI model. The AI model has no memory of the bug or its solution. The developer must explain everything again, wasting time.
Resolved from this lesson's own record — real definitions, not invented for this page.
More from the same record, pulled by shared topic — not written new for this page.
An AI model must always read a specific set of files. These files include 'mistakes_and_lessons.md'.
An AI model can generate a large glossary by using a multi-pass data pipeline. Each pass focuses on different content.
Recording a lesson and enforcing a rule are different actions. Recording saves information about a past event.
An AI model must read `mistakes_and_lessons.md` first. This file contains rules learned from past problems.
An AI model's memory is limited. A developer must save important information to a separate file. This file helps the developer and the AI model continue work effectively.