Building Shared Memory For Ai Agents In Notion
- Source
- notion.com
- Author
- Notion Blog
- Date

Context
Notion’s Lore stores lessons from sessions in shared pages that later sessions can retrieve. Its engineering article describes keeping procedures, decisions and other reusable knowledge accessible to both people and agents, with expiration and maintenance tools to manage stale information.
Notion’s own evaluation used coding scenarios chosen because models could not reliably complete them without Lore. In 149 paired scenarios, agents with a memory collection seeded with relevant information passed 102, versus 82 without it. Memory helped in 27 scenarios and hurt in 7. Notion did not its messier team-created collection, which its own sampling suggests contains duplicates and low-value notes. The evidence supports relevant, retrievable memory as an aid on these hard tasks; it leaves the benefit of accumulating unmaintained session notes unresolved.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
- MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
- agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
articleAgent Memory Repo | Cognitioncognition.com
articleFilesystem-Based Memory for LLM Agents: Organization, Evolution, and SustainabilitySizhe Zhou, Sheldon Yu, Hui Wei, Junda Wu, Siru Ouyang, Yizhu Jiao, Shijia Pan, Julian McAuley, Yu Zhang, Tong Yu, Jiawei Han
articleA Skills Library For Every AgentNotion Blog
Checking sign-in…
Loading comments…


