Agentic GraphRAG and an explorer for your own stack

Neo4j’s graph-explorer-kit is presented as a Python-based Agentic GraphRAG toolkit built on neo4j-graphrag that helps developers use Document Intelligence knowledge graphs within applications through retrievers, agent tools, and an HTTP API. The author demonstrates the workflow using 240 released UFO-related documents, converting PDFs and scans into a provenance-linked graph containing entities, events, agencies, claims, evidence, redactions, and source passages; the resulting dataset includes more than 54,000 records and 58,000 relationships. A configurable ProjectSpec defines a domain’s schema, extraction rules, Cypher examples, and seed questions, while retrieval can combine vector search, full-text search, graph expansion, and text-to-Cypher queries to connect information spread across separate documents. The accompanying Graph Explorer visualizes streamed answers, relevant passages, extracted entities, neighboring graph context, schema information, and timelines, and can ingest additional documents with OCR support. The post notes practical limitations, including noisy scanned text, ambiguity in entity merging, and the fact that displayed related graph context is not necessarily evidence for every claim, emphasizing that the graph represents reports, observations, agency conclusions, and disagreements rather than asserting explanations for UFO events.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.