Microsoft Fabric Data Agents are now generally available and consumable over an API, which means you can put a governed, natural-language analytics experience directly inside your own application. The question is no longer whether you can do it, but how to do it cleanly.

This is a hands-on, demo-driven session where we build the whole thing live. We'll start by creating and publishing a Fabric Data Agent, covering the best practices that make an agent actually reliable: data source selection, instructions, and validation. From there we'll set up the app registration that lets an application authenticate to the agent securely.

Finally, we'll embed the agent into a Fabric application built with Rayfin, the open-source SDK for building apps that run on Fabric, ending with a working conversational analytics experience inside a real app.

You won't leave with just concepts. We'll share a GitHub repo with repeatable labs and example patterns you can use to accelerate embedding agents into your own applications, so you can go from this session to a working build of your own.