Thinking in Graphs: The Cognitive Foundations of Agentic AI
summary
Claims that Graph RAG plus the Model Context Protocol produce explainable, agentic AI, demonstrated with a Neo4j-backed knowledge graph feeding LLM context, FastMCP for secure tool integration, XA for web crawl, and OpenAI models. Built a real-time ESG monitoring pipeline (used across 24 countries) that extracts entities and sentiments, and a facility-management assistant that auto-creates Jira tickets, sends emails, and enforces norms; both systems run and return auditable, source-linked answers rather than hallucinations.
AI-written summary · about
chapters (14)
- 0:00Thinking in graphs and agentic AI
- 4:59Limitations of current large language models
- 8:09RAG systems constraints
- 11:21Knowledge graphs enhance AI
- 13:29Graphs vs traditional RAG
- 26:45From reactive to proactive agents
- 30:03Model Context Protocol benefits
- 37:10Graphs, MCP, and LLMs synergy
- 40:48ESG reporting agent overview
- 57:57ESG agent demo
- 1:00:05Facility management assistant
- 1:09:16MCP workflow in FM
- 1:15:23Agentic AI with graphs summary
- 1:18:38Q&A
related talks
24mAgentic GraphRAG: AI’s Logical Edge
36mAgentic GraphRAG: AI’s Logical Edge
08mHarness Graph Power With Agentic AI: An Insights Discovery Use Case
21mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
20mBeyond RAG: Building the AI Corporate Brain with Knowledge Graphs and Virtue Ethics
42mMaking OSS AI Real with Knowledge Graphs
12mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
23mAgentic GraphRAG with MAGMA Manifold Agency with Graph Models and Augmentation