What Is GraphRAG? Knowledge Graphs vs Plain Vector Search — [AI Stack 17]

GraphRAG is a form of retrieval augmented generation that first organizes your documents into a knowledge graph of entities and the relationships between them, then retrieves over that connected structure, so the model can answer questions that require linking many facts rather than matching a single passage. Episode 17 of The AI Stack explains the two questions plain vector RAG can't answer — multi-hop and global summary questions — how a knowledge graph fixes both, and the real tools behind it, including Microsoft GraphRAG and the Neo4j graph database, plus the honest tradeoffs in cost and complexity. ▶ The AI Stack — full course in order: [playlist] ← Previous: [AI Stack 16] Chunking Strategy & Re-ranking → Next: [AI Stack 18] Agentic RAG — When Retrieval Becomes a Loop #GraphRAG #knowledgegraph #RAG