The Knowledge Graph Mullet: Trimming GraphRAG Complexity
summary
Claims that a property graph front with RDF triple storage back simplifies GraphRAG by keeping a property-model query surface while using triples for scalable storage. Shows Dgraph (DQL, posting lists, vector and geospatial indexes), the Dgraph MCP server with readonly and full endpoints, and Hypermode + modus agent orchestration using WebAssembly. Demonstrations build a news knowledge graph, an e‑commerce dataset, run vector→graph traversals, and auto-generate and save content to GitHub/Notion via MCP — it works with iterative fixes.
AI-written summary · about
chapters (19)
- 0:00Knowledge graph mullet
- 1:05Property graphs vs RDF triples
- 2:51Knowledge graph model
- 4:40Dgraph hybrid graph
- 7:51Querying with DQL/GraphQL
- 8:59News knowledge graph
- 11:07Graph RAG entry points
- 12:53DQL query examples
- 15:44Dgraph latest features
- 16:51Model Context Protocol
- 18:20Dgraph MCP with Cloud
- 20:55Graph schema/data
- 21:58Visualizing graph data
- 22:42Product recommendations
- 24:07Agent orchestration
- 26:20Domain-specific agents
- 27:00Hypermode agents demo
- 31:53Modus code ejection
- 32:15Closing resources
related talks
53mPractical GraphRAG - Making LLMs smarter with Knowledge Graphs
26mPractical GraphRAG Making LLMs smarter with Knowledge Graphs
30mAgentic RAG with Knowledge Graphs Explained
27mAdaptive Knowledge Graphs for Dynamic RAG Systems
27mLies, Damn Lies, and AIs
15mMultiple Knowledge Graphs for Sophisticated RAG
36mAgentic GraphRAG: AI’s Logical Edge
27mWhen Vectors Aren’t Enough: Unlocking AI’s Hidden Connections with Graphs