Blog: Agents Don't Need Memory Plugins, They Need Documentation
itsOmSarraf_ · x · 2026-10-05
Kevin Liao argues the entire agent memory plugin ecosystem solves the wrong problem: every product shares the same flawed RAG architecture — slicing conversations into vector snippets and injecting the top-5 similar ones per prompt, a lottery that never gives the agent real understanding of your project.
His thesis: agents don't need memory, they need documentation.
Reposter badlogicgames goes further: for code, neither memory nor docs are efficient — your codebase is all you need; keep it modular, modules small, plus a small maintained map of where things live.
More from coding & agent
- SpaceX engineer: AI agents now merge PRs on their own — over 1,000 a month — itsOmSarraf_ · 2026-10-05
- Opus 5.5 as daily driver, GLM for cheap prod: a practical model split — ivan_bezdomny · 2026-10-05
- Astra agent reverse-engineers macOS binaries to clone all 27 Liquid Glass effects in GPUI — steipete · 2026-10-05
- Jeff Dean's 1-hour AI engineering lecture: from LLM basics to graph-orchestrated agent swarms — glenbeer · 2026-10-05
- Fully local parkour sim vibe-coded with GLM 5.3 Flash on 2x DGX Sparks, recipe open-sourced — -dysangel- · 2026-10-05
- Someone mapped an entire company's workflows onto OpenAI Dots — a playbook solo builders can steal — Roger_M_Taylor · 2026-10-05