GraphRAG
GraphRAG is retrieval-augmented generation that retrieves from a knowledge graph — or a graph plus a vector index — instead of vector similarity alone. It grounds the model in explicit entities and relationships, so answers respect real facts and connections, not just topical similarity.
Also known as: graph RAG, graph-based RAG
Standard retrieval-augmented generation finds passages whose meaning is closest to the question using a vector database. That’s great for “find relevant text” and weak for “respect the actual relationships between things.” GraphRAG retrieves from a knowledge graph — entities and the explicit links between them — so the model gets structured facts (who owns what, what depends on what) rather than fuzzy nearest-neighbors.
In practice the strongest setups combine both: vectors cast a wide net to find the relevant neighborhood, and the graph constrains and enriches it with exact entities and relationships. That pairing is one of the more effective levers against hallucination when a domain has real structure — compliance, supply chains, org data — because the answer is anchored to facts, not similarity. The cost is building and maintaining the graph, so it’s worth it when relationships are the point, not when you just need topical recall over text.
Go deeper
- Vector database or knowledge graph — which should you use for AI retrieval? AI, decoded · Vector Database vs. Knowledge Graph for AI Retrieval
- What is RAG, and why do AI systems use it? AI, decoded · What Is RAG (Retrieval-Augmented Generation)
- What is agentic RAG, and how is it different from regular RAG? AI, decoded · Agentic RAG vs. Traditional RAG
