Copilot in web modeling (Preview)

Microsoft Employee

3 months ago

If you haven’t already, check out Arun Ulag’s hero blog “Microsoft Build 2026: Building Agentic Apps with Microsoft Fabric and Microsoft Databases” for a complete look at all of our Microsoft Build announcements across our Fabric and database offerings.

______________________________________

Copilot in web modeling (Preview), rolling out this month in the Power BI service, is an AI-powered assistant helping you analyze and improve your semantic models using natural language so you can spend less time on manual edits

Model smarter with AI

Copilot works directly in the model authoring experience, allowing you to explore and refine your model conversationally. Whether you're cleaning up naming conventions or building out key metrics, Copilot can suggest improvements and apply updates on your behalf.

Capabilities with Copilot:

Analyze your model and identify areas for improvement, such as inconsistent naming or unclear structure
Make schema updates, including renaming tables and columns, creating relationships, and generating DAX measures

Built for safe, controlled changes

Copilot respects your existing Power BI permissions and Admins, Members, Contributors can use Copilot to make changes to the active model. At the start of each session, you’ll provide permission before any analysis or updates occur. A restore checkpoint is also created automatically to easily revert your model if needed, using semantic version history.

Get started

To try Copilot in web modeling:

Open a semantic model in Model view in the Power BI service
Switch to Editing mode
Select Copilot from the ribbon

From there, you can start with suggested prompts or ask your own questions to analyze and refine your model.

Figure: Copilot in web modeling running the prompt of “Improve table relationships” and specifying which relationships to create.

Coming soon

We’re continuing to expand Copilot’s capabilities to help you prepare models for richer AI experiences. Upcoming enhancements will help improve AI readiness by defining clear AI instructions and refining your data schema to support better natural language experiences.