Fabric data agent has been available for a while, yet many implementations fall short of expectations. Not because of the technology, but because of weak design, unclear instructions, or poorly prepared data models.

In this session, you’ll learn how to build high‑quality, trustworthy agents by applying proven design patterns around Agent Instructions, Data Source Instructions, and Verified Answers. We’ll explore how to structure semantic models for AI, route questions intelligently across Lakehouse, Warehouse, Eventhouse, and semantic models, and avoid the most common pitfalls that lead to inconsistent results.

You’ll also see how to operationalize your agents using the Data Agent SDK for automated creation, updates, and evaluations, including golden‑record testing and diagnostics. After this session, you’ll have a clear blueprint for designing reliable, secure, and well‑adopted Fabric Data Agents—from setup to scalable success.