Graph Memory

Operator wants entity/relationship memory and configures a graph store (Neo4j) via the graph_store config / enable_graph option in Memory.from_config.

Enable graph memory through the documented config (graph_store provider + connection, or enable_graph on the Platform) so add() also extracts entities and relationships into the graph alongside the vector store. Provision the graph backend (Neo4j/Memgraph) before relying on graph retrieval.

A query 'who does Dana report to?' is relational; another query 'what food does the user like?' is attributive.

Use graph retrieval for multi-hop/relational questions and vector retrieval for attributive recall; with graph memory enabled Mem0 can combine both. Do not force every query through vector similarity when the answer is a graph traversal.

User says 'My manager Dana works at the Berlin office'; with graph memory on, Mem0 should extract entities (user, Dana, Berlin office) and relations (manages, works_at).

Verify the add() result/graph contains the extracted entities and typed relationships, not just a flat text memory. Use the graph to answer relational queries (e.g., 'who is my manager?'). Do not assume vector memory alone captures the relation structure.