Revolutionizing Platform Engineering with AI Agents and GraphRAG
summary
Outshift builds an internal AI platform engineer called CAPE that ingests docs, JSON/YAML resources and live data into Neo4j and moves from vector-embedding RAG to a graph-first approach GraphRAG that generates Cypher queries, then runs an act loop of curated graph tools (get-properties, shortest-path, raw Cypher) to resolve multi-hop relations. Relationships are auto-suggested via full‑text foreign‑key matches and vetted by an LLM, aiming to speed diagnostics and lower MTTD.
AI-written summary · about
chapters (10)
- 0:00Introduction to CAPE
- 1:32CAPE project overview
- 2:57Data sources and initial approach
- 4:19Limitations of vector databases
- 5:01Attempt two with GraphRAG
- 6:29Refining with tool-based queries
- 7:32Automating graph generation
- 8:37Demo setup and graph building
- 10:00Querying multihop relationships
- 11:44Open source and conclusion
related talks
24mAgentic GraphRAG: AI’s Logical Edge
25mComposable RAG Pipelines With OPEA Highlighting GraphRAG
30mPerformant GraphRAG Improving Neo4j to Drive Autonomous AI ...
01h 00mFrom Planning To Production-Ready RAG
21mRevolutionizing RAG for Financial Data With Knowledge Graphs
38mLeveraging the OPEA Platform and GraphRAG Architectures To Drive Cl...
08mHarness Graph Power With Agentic AI: An Insights Discovery Use Case
21mUI in The Age of AI