Persistor
Persistent memory for AI agents.
Knowledge graph + vector search hybrid memory. PostgreSQL-native, encrypted at rest, open source.
curl -fsSL https://persistor.ai/install.sh | bashPersistent memory for AI agents.
Knowledge graph + vector search hybrid memory. PostgreSQL-native, encrypted at rest, open source.
curl -fsSL https://persistor.ai/install.sh | bashFeatures
A purpose-built memory layer that combines the structure of knowledge graphs with the flexibility of vector search.
Store entities and relationships with full graph traversal. Your agent understands context, not just similarity.
pgvector-powered semantic search across all memories. Find relevant context even when the words don't match.
Every piece of data encrypted at rest. Your agent's memories are private by default, not by afterthought.
No exotic databases. Built on PostgreSQL with pgvector — battle-tested, easy to operate, easy to back up.
Your data never leaves your machine — not even for embeddings. Ollama-powered vectors, zero cloud dependencies.
Isolate memories per agent, per user, per workspace. Built for platforms that serve many customers.
How It Works
Single Go binary + PostgreSQL. One command to install, one command to initialize. CLI and REST API included.
Write memories as entities, relationships, and embeddings. Persistor handles encryption, indexing, and graph construction automatically.
Query by semantic similarity, graph traversal, or both. Your agent gets the right context every time.
Open Source
Persistor's core is open source and always will be. Self-host it, fork it, extend it. The code is yours.
Need managed hosting, automatic backups, and zero ops? Persistor Cloud is coming soon.
View on GitHub