Long-running agents accumulate bad memories — dev calls for an explicit memory lifecycle
Arc_bong · reddit · 2026-08-30
A developer on Reddit raises a real-world pain point: for agents running across many sessions, long-term memory works great at first, but after enough history accumulates the effect reverses —
- stale decisions get retrieved even after the underlying situation changed
- conflicting memories from different sessions look equally relevant
- the agent burns context on old, no-longer-useful information
- simply improving retrieval doesn't necessarily improve task outcomes
The core claim: memory systems need an explicit lifecycle, not an ever-growing retrieval store. The author asks the community what works in practice: separating semantic facts / episodic experiences / procedural instructions; decaying, expiring, or periodically consolidating memories; keeping provenance + timestamps so agents can judge trustworthiness; and — most interesting to him — evaluating memory by downstream task success rather than retrieval precision/recall. A memory can be retrieved "correctly" and still make the agent's next action worse. He also compares LangMem, Mem0, Letta, and Lyzr Control Plane, which make different assumptions about where memory should live in the agent stack.
Related event: Long-Running Agents Break as Stale Memories Accumulate(2 posts)→
More from coding & agent
- Agent named 'Ilya' reportedly scores 1.1pt higher on benchmarks — daniel_mac8 · 2026-10-16
- Gemini CLI fix: substring matching inlined unrequested binary files into model context — priyanshu1976 · 2026-10-16
- Building AI agents on real healthcare claims data: the domain work that makes them work — alex_verem · 2026-10-16
- OpenAI stealth-testing mysterious internal model 'Aeon' in Codex desktop app and Dots — cedric_chee · 2026-10-16
- Vibe-coded iPhone Duo sniping bot built with Codex in 17 minutes, 41 tests passing — xiaohu · 2026-10-16
- Long-term multiplayer agent memory tool tops BEAM memory benchmark — CurieuxExplorer · 2026-10-16