Architecting Enterprise Agents Ecosystem: Standardizing APIs with MCP and Orch…
summary
Enterprise moves from laptop MCP to Kubernetes-deployed MCP middleware built with FastAPI + FastMCP and LangGraph, delivering a BU-level agentified service in weeks; local MCP is abandoned. Cross-BU routing uses A2A + vector search with agent cards in a central vector index, an Orchestrator that semantically discovers agents, Kafka for dispatch and Postgres for DAG resumability. Governance enforces isolated MCP datastores, no destructive endpoints, device-bound identities, LLM judges and LangFuse traces for continuous quality monitoring.
AI-written summary · about
chapters (7)
- 0:00Enterprise agent ecosystem intro
- 3:44Middleware: FastMCP & LangGraph
- 10:47Governance in agent systems
- 14:18Scaling MCP for agent comms
- 17:27Agent quality challenges
- 21:15Observability for agent performance
- 25:07Summary & Q&A
related talks
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
46mMCP and A2A Revolutionizing Interconnection and Collaboration of AI Agents
25mAgents + Data Streaming for Real-Time Enterprise Intelligence
18mThe "Agentic Organisation": Governing Multi-Agent Developer Teams with Nutanix Agent Gateway and MCP
32mFrom Monolith to AI Agent Modernizing Java Systems with MCP
40mBuilding Agents with Agent Bricks and MCP
18mScaling AI Agents With gRPC: Intuit's Enterprise MCP Service Marketplace
01h 23mA2A & MCP Workshop: Automating Business Processes with LLMs