In this video I'm demonstrating a new AI feature Microsoft has released for Power BI and Microsoft Fabric called the Power BI MCP Modeling Server. It's a feature that helps you create DAX measures using AI, and I show how to configure it to work with Claude, so you can write Claude prompts and have Claude create DAX measures directly onto your semantic model.
What This Feature Does
This is really useful if you've got a lot of KPIs on your data model and want an automated way to create definitions that are user-friendly, without writing all the DAX yourself.
What I Show in the Video
I walk through getting the Power BI Modeling MCP file set up, connecting it on the Claude Desktop side, and then connecting Claude to my Fabric environment and semantic model. I got the config code from a girl in the cube, since Microsoft's own version doesn't actually work.
Once Claude's connected, I ask it to review my data model, and then ask it to create some basic measures. Claude ends up generating new measures on its own, figuring out what makes sense based on the fact and dimension tables. I reviewed the DAX was created correctly on Fabric, as the AI can at time make the wrong assumptions.
A Couple of Notes
The AI can make mistakes, so you have to double-check the logic it uses. And it's worth backing up your semantic model before trying this yourself, especially if it's in production.
Watch the video above to see it all step by step.