guidesMay 28, 2026· 5 min read

Give LangGraph agents long-term memory with Crowkis

Durable, per-user memory for LangGraph agents that survives restarts and consolidates contradictions, self-hosted, zero egress.

LangGraph agents forget the moment a run ends, so every session relearns the user and re-pays for context. Crowkis gives them memory that lasts.

Recall known facts before the model call, store what you learned after. Memory is scoped to (agent, user), ranked by relevance blended with recency, and consolidating, a new fact that contradicts an old one retires it.

LangGraph memory node

mem.remember("prefers email over phone")
mem.recall("how should I contact them?")   # semantic recall

In plain words: Storage isn't memory. Memory is knowing which of the things you stored is still true.

It runs on bundled local models, so you can give LangGraph agents memory without shipping conversations to anyone. It's one self-hosted binary, Redis-compatible, free to run.