Microsoft Copilot in Power BI: natural language questions to your model
João Barros
27 de August de 2024
2 min read
Microsoft Copilot in Power BI integrates generative AI directly into the service, letting you build reports, write DAX and generate narrative summaries just by describing what you want in natural language.
Requirements
- Power BI Premium license (P1+) or Fabric capacity F64+.
- The tenant admin must enable Copilot in Admin Portal → Tenant Settings.
- Semantic model published in a workspace with Copilot capacity.
Create visuals with natural language
In Power BI Desktop or Service, open the Copilot pane (ribbon button) and write:
User: "Show the monthly sales evolution by product category
for the last 12 months, with a trend line"
Copilot → automatically generates:
- Visual: Line and Clustered Column Chart
- X axis: date hierarchy (Month-Year)
- Legend: Product Category
- Line: Trend (DAX: LINESTX)
- Filter: last 12 months (DATESINPERIOD)
Automatic smart narrative
User: "Create an executive summary of 2024 sales with
the top 3 insights and comparison with 2023"
Copilot generates a dynamic paragraph with:
- Total 2024 sales vs 2023
- Growth % calculated in DAX
- Automatic outlier identification
- Conditional formatting by performance
Generate DAX measures
User: "Create a DAX measure to calculate year-over-year
sales growth as a percentage"
Copilot suggests:
YoY Growth % =
VAR CurrentSales = [Total Sales]
VAR PriorSales = CALCULATE([Total Sales], SAMEPERIODLASTYEAR('Dates'[Date]))
RETURN DIVIDE(CurrentSales - PriorSales, PriorSales)
Tips for better results
- Use descriptive field names in the model — Copilot relies on them to infer intent.
- Mark the date table as a Date Table so Copilot understands temporality.
- Be specific: "sales in € for the last 12 months" is better than "sales evolution".
Conclusion
Copilot in Power BI democratizes report creation, letting business users build analyses without knowledge of DAX or modeling. For analysts, it speeds up prototyping and writing complex measures.