Building Agents on Databricks in 90 Minutes: Orchestrating Knowledge, Data, and Reasoning

Most enterprise AI questions don't fit neatly into one box. "Why did my bill go up?" lives in structured tables. "How do I turn off my hot spot?" lives in product docs and support tickets. Real assistants need both — and the orchestration to know when to use which.In this hands-on lab, you'll build a complete multi-agent customer support system for a fictional telco using Agent Bricks — no deep ML expertise required. You'll stand up a Knowledge Assistant over vector-indexed FAQs and historical tickets, query structured billing and customer data through a Genie space, and wire them together under a Supervisor Agent that intelligently routes every question to the right specialist. Along the way, you'll learn how Unity Catalog governance, vector search, and natural-language instructions combine to turn fragmented data into a single, unified, production-ready assistant — in under 90 minutes.