Intelligent Data Platform : Powered by Semantic Layer, MCP & Agents
summary
Claims that a unified semantic layer plus MCP tools dramatically reduces agent hallucination by providing executable metrics, glossary and KG-backed definitions. Demo builds a Docker Compose Iceberg lakehouse (MinIO + Trino) and two MCP servers: a lakehouse MCP and a semantic-layer MCP using DBT metric flow and Ontop; two agents run (data-only vs data+semantic). Data-only agent invents thresholds and uses total_price; semantic agent uses canonical metric and KG rule, returning 10 premium customers versus the data-only agent’s 100+ and correct revenue.
AI-written summary · about
chapters (10)
- 0:00Intelligent data platforms
- 1:29Self-service analytics evolution
- 3:37Agent-to-data access challenges
- 7:10Unified semantic layer
- 13:30Semantic layers critical now
- 16:41Model Context Protocol
- 18:05Lakehouse architecture
- 21:35Demo architecture
- 23:20Agent behaviors comparison
- 30:06Key takeaways
related talks
25mAgents + Data Streaming for Real-Time Enterprise Intelligence
28mthe semantic layer your AI needs
28mAI Agents in Action: Harnessing MCP and OpenSearch for Modern Use Cases
49mSolving Context Bloat: Semantic Tool Routing in Multi-Server MCP Environments
12mMCP and the Dawn of Agentic Intelligence
30mComposable AI: Building Next-Gen AI Agents with MCP
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
46mMCP for LLM Agents, Infrastructure APIs & Data Services