Databricks
summary
Databricks unveils agent bricks, a packaged multi‑agent stack that builds RAG pipelines, converts text to SQL, and auto‑benchmarks with MLflow 3. Demo shows a bookstore assistant that reads docs, queries inventory, and orders out‑of‑stock items. Claims: >90% of orgs stall at PoC; one client sees 8× faster pilot‑to‑prod and +40% accuracy; auto‑optimization yields 20–27× speed and 69% cost reduction. Production tooling ties into vector indexes, MCP servers, and content tools like 11 labs and Google AI Studio.
AI-written summary · about
chapters (8)
- 0:00AI agents introduction
- 3:13What is an AI agent
- 5:24Agent development building blocks
- 7:57Bookstore demo multi-agent system
- 10:49Creating agents with Agent Bricks
- 13:18Evaluating AI agents
- 14:23Optimizing agents with MLflow 3
- 17:11Key takeaways
related talks
37mOn Building Production-Quality AI Systems
49mBuild & deploy generative AI apps in days with Databricks Lakehouse AI
17mHow will AI transform your data platform?
18mDatabricks drives predictive maintenance, grid optimization with analytics
17mOn Bringing AI to Business Intelligence
23mBuilding AI-Driven Real-Time Lakehouse APIs
02h 52mKeynote Day 1 - Full
45mFrom hand-down maps to AI/BI: a journey toward Databricks