Knowledge Graph

A knowledge graph is a structured representation of entities and the relationships between them — used alongside vector search to give AI systems a grounded model of a domain.

Where vector stores capture meaning, knowledge graphs capture structure: 'Acme is a customer', 'Acme owns deal-123', 'deal-123 is in stage Negotiation'. The combo (vector + graph, sometimes called GraphRAG) supports multi-hop questions vector search can't answer alone: 'what's the latest deal status for our top customer?'.

Example

A support AI uses a knowledge graph to traverse: ticket → customer → contract → SLA → entitlement, in milliseconds — without retrieving and re-reading docs.

How OpenLabor uses it

OpenLabor builds a per-workspace graph of employees, customers, deals, and projects so the AI navigates structure, not just text.

Do I need a graph database?

Not always. Many GraphRAG setups run on Postgres with relational tables. Specialized graph DBs win at very high traversal volume.

Related: rag, vector-store, embedding.

AI Labor Glossary