Power BI semantic models — now talking back
Power BI does many things well. Letting non-technical users ask questions of a semantic model is not one of them. Power BI Q&A's natural-language layer never quite caught on, and getting answers usually still means knowing which page to open and which slicer to click.
AI Insights 365 changes the layer. Connect any Power BI Premium dataset via its XMLA endpoint and our agent will write DAX automatically.
What you get
- All your existing semantic model investment — measures, hierarchies, calculation groups — works as-is.
- Plain-English questions like "Total sales by region last quarter" turn into proper DAX queries.
- You see the DAX, the rows, and a chart.
- Your Power BI admins keep central control of the model; AI Insights is read-only.
Setup
You'll need:
- Power BI Premium (P/EM SKU) or PPU — XMLA endpoints are gated to these
- An Azure AD service principal with Build permission on the workspace dataset
- Tenant ID, Client ID, Client Secret
Full walkthrough in Connect a Power BI semantic model.
Why this matters
You don't have to migrate off Power BI to get conversational analytics. You don't have to replace your model team's work. You add a chat layer on top of the model that already exists — and the questions your business users actually ask, finally have a direct answer.