RL teaches agents to use their own notes; training survives 23M tokens of memory edits
simran_s_arora · x · 2026-11-05
Engramo Lab shares new research on using RL to complement text-based memory: agents rely on notes, wikis, and skill files, and training them to use these notes better is cheap compared to building them — and a single training run on an older snapshot of the notes stays robust even after 23M tokens of edits.
- Memory systems evolve constantly; the trained policy remains surprisingly robust through millions of tokens of changes
- Training on how to use notes may matter as much as building the notes themselves
More from coding & agent
- 5 ways to connect AI agents across frameworks with the A2A protocol (26.4k stars) — goyalshaliniuk · 2026-11-05
- Ask your model what's wrong with your harness — an easy trick to catch silent bugs — insomniating · 2026-11-05
- "Haven't pushed in 6 months, it's all AI code and worth nothing" — the wildest take in the Claude Code repo-wipe thread — ryunuck · 2026-11-05
- Solo dev builds hyper-realistic browser jet simulator with Claude Opus 5.5, 25,000 players in two weeks — aCakeShip · 2026-11-05
- FrontierPhysics: 56 expert tasks benchmark AI agents on end-to-end physics research — Bingran You · 2026-11-05
- Free browser tool turns iPhone Duo screenshots into 3D fold animations up to 4K — yihui_indie · 2026-11-05