Building Agents with Agent Bricks and MCP
summary
Agent Bricks claims to let teams build production agents quickly by wiring a vector index of protocols plus PDFs and regulatory docs into a no‑code knowledge assistant that kicks off in ~20 minutes and improves when a human adds a one‑line guideline (e.g. define “ongoing clinical trial”). MLflow logs traces, runs LM judges and records metrics like evidence groundedness and rational sufficiency while the workspace swaps models (GBD5, Sonnet, Gemini) and compares outputs (GBD5 mini vs Claude). MCP exposes tool endpoints via MCP endpoints install via marketplace and supports a multi‑agent supervisor combining specialized agents, all governed in Unity Catalog.
AI-written summary · about
chapters (7)
- 0:00Building agents with Agent Bricks
- 3:32Production-ready agent challenges
- 7:25Agent Bricks purpose
- 14:45Knowledge assistant demo
- 20:05Multi-agent supervisors and MCP
- 27:05MLflow for GenAI deployment
- 35:12MLflow judge builder Q&A
related talks
29mBuilding Interoperable Agents: When A2A meets MCP
21mBuilding Agents (the hard parts!)
40mHow To Tame Your Agents?
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
40mAgents in Action: Advancing Open Source AI Missions
01h 26mBuilding Agents with Amazon Nova Act and MCP
01h 28mAI Coding Agents and how to code them
28mThe Put That Sh*t On Agent: Building Multi-Agent AI Systems That Actually Talk to Each Other*