Michael Hunger is Neo4j’s Head of Product Innovation and Developer Strategy, developing practical ways for generative AI to retrieve connected, explainable information. His work on GraphRAG addresses a fundamental limitation of conventional retrieval: finding semantically similar passages does not necessarily uncover the relationships needed to answer complex questions.
Hunger has spent more than three decades in software development and approximately 15 years working with Neo4j. Earlier, he led Spring Data Neo4j, worked on cloud-hosting initiatives, and created an open-source CSV batch importer. His responsibilities expanded into developer relations, open-source ecosystem development, and product innovation, with projects spanning graph algorithms, GraphQL, Spark and Kafka integrations, APOC procedures, and database connectivity. A technical author, former InfoQ editor, and Java Champion, he is coauthoring GraphRAG: The Definitive Guide with Stephen Chin and Jesús Barrasa.
Building AI around connected knowledge
Connected context improves retrieval. Hunger’s work on graph-enhanced retrieval combines semantic search with relationships among documents, people, products, and other entities. Following those connections helps answer multi-hop questions and exposes the information supporting a generated response.
Knowledge graphs need deliberate construction. His approach starts with a lexical graph representing documents and their internal structure, adds domain-specific entities through schema-guided entity extraction, and enriches the result with graph algorithms and cross-document summaries. Existing customer records or product catalogs can anchor extraction and improve grounding. His AI Engineer technical walkthrough demonstrates how these layers become connected retrieval pipelines.
Developer tools make GraphRAG operational. Hunger has helped advance the LLM Knowledge Graph Builder and NeoConverse, tools for turning source materials into knowledge graphs and querying them conversationally. His GraphRAG ecosystem overview also covers integrations with established AI development frameworks. He contributed import functionality connecting Microsoft’s independently developed GraphRAG output to Neo4j.
Agents benefit from bounded graph operations. His work on Model Context Protocol integrations gives AI systems explicit tools for neighborhood lookup, shortest-path discovery, and domain-specific queries, making the context behind an answer more visible and controllable.
Build a retail graph, search product descriptions, personalize results through purchase paths, and use graph embeddings to supply recommendations for a seasonal email.
GraphRAG combines index search with explicit relationships, preserving document structure and enterprise entities so retrieval can return the context an answer needs.
▶ Watch ↗