xmem — Graph-based long-term memory for AI agents.

Most AI agents forget everything the moment a session ends. xmem fixes that. It's a hosted, graph-based memory layer that lets your agents store, connect and recall knowledge over time — not as flat text chunks, but as a real knowledge graph of entities, relationships and decisions.

How it works: Sign up, get an API key, install the plugin (OpenClaw plugin, Python SDK or REST API). Your agent instantly gains long-term memory — automatic context capture, multi-hop GraphRAG search, and entity/relationship tracking.

No infrastructure to run.

Why xmem:

True graph memory — entities, edges & decisions, not just vector chunks
Drop-in setup — plugin + API key, works in minutes
Multi-hop GraphRAG recall for deeper, connected context
Multi-tenant & production-ready
Up to 8× cheaper than comparable memory services
Start free — 1,000 memories at no cost.

Scale up as your agents grow. https://xmem.space

See documentation under: https://xmem.space/docs