Should long-running AI agents keep permanent memory?
Most RAG systems fetch needed knowledge, use it for an answer, and then drop it. That is simple and often sensible. For long-running agents or tools, some repeatedly useful knowledge might be worth keeping as .
Saving would create confusion, so the system would need a way to decide what is ant enough to remember. The main risk is that may store stale knowledge or bad , which can make later answers worse.
Key points
- Typical RAG retrieves information for one answer and does not keep it afterward.
- Long-running agents may benefit from saving selected knowledge as .
- Saving too much can create stale knowledge and bad .
- Useful needs clear save, refresh, and delete rules.
- Well-managed can reduce repeated and token use.