For years, analytics has been passive: it surfaces insights, then leaves users to act on them. Fabric Data Agents, Fabric Apps, and App Functions change that. Users can ask a question, get an answer, and take action without leaving the application.

It starts with Data Agents. We'll look at what makes one reliable: choosing the right data sources, writing clear instructions and example queries, and validating answers. Then we'll organize several agents into a coordinated team that can handle questions no single agent can answer.

Fabric Apps is where users meet those agents. We'll embed the agent team directly in an app's interface, so conversation becomes part of the application instead of a separate chat window. We'll also cover the architecture that connects the app, the agents, and the data underneath.

App Functions close the loop. They let the app do something with an answer, whether that's starting a workflow or triggering the next action. We'll walk through orchestration patterns, including how agents suggest the right function at the right time, so users go from insight to action in one place.

You'll leave with a GitHub repository of repeatable labs, sample code, and reusable patterns for building your own agent-powered Fabric applications.