GraphRAG
Retrieval-augmented generation over a knowledge graph rather than a flat vector index, using typed entities and explicit relationships so that retrieval can traverse connections rather than only match similarity.
The position
Vector similarity answers 'what looks like this question.' Finance mostly needs 'what is connected to this entity, and how' — a counterparty's exposure chain, an index's constituent history, a model's data lineage. Those are traversals, not similarity searches, and no amount of embedding quality converts one into the other. The graph is also auditable in a way an embedding space is not: you can show a regulator the path.
This section is a claim, not a definition. It is argued rather than asserted, and it is the part of this page you are invited to disagree with. Corrections and counter-arguments to [email protected] are published.