gensai.aiGraphRAG is a variant of RAG that relies on a knowledge graph, a network of entities (people, products, places) and the relationships between them, rather than on plain text passages. Popularised by Microsoft Research in 2024, the approach handles cross-cutting questions that require connecting several documents better.
In practice at Gensai
Gensai is exploring GraphRAG for highly interconnected document bases, where classic RAG struggles to link sources together.