Human-in-the-loop knowledge base for AI agents
I stopped letting my agents write to my knowledge base. They can read it; they can leave notes next to it; but the library itself is mine. That split exists because every agent I run loves to create, and none of them want to update. For months I gave my agents a shared Notion workspace. The idea was simple: they work, they write down what they learned, the next agent reads it. What I got was a…
The article details a knowledge management system for AI agents that separates sources and notes. Sources are permanent documents chosen by the user, while notes are temporary and untrusted. The system employs a 30-day TTL (Time To Live) for notes, automatically deleting them after this period. Notes are triaged weekly by the user who assigns them a verdict: Keep, Add to draft, or Remove.
The final selected notes are added to the source library after a human review. AI agents retrieve knowledge by querying the system, which returns verbatim passages with citations and a support score. The system emphasizes transparency, control, and least privilege, with separate profiles for different agents. The author argues that this approach reduces the risk of outdated or incorrect information being propagated through the AI agents.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
