Join us at FabCon Atlanta from March 16 - 20, 2026, for the ultimate Fabric, Power BI, AI and SQL community-led event. Save $200 with code FABCOMM.
Register now!Join the Fabric FabCon Global Hackathon—running virtually through Nov 3. Open to all skill levels. $10,000 in prizes! Register now.
Hi there!
I'm trying to use Copilot for Power BI (standalone version) with a star-schema modeled dataset, following best practices in dimensional modeling. I've prepared the data specifically for AI consumption by simplifying the schema and creating a Markdown-based AI instruction that outlines the tables, measures, and relevant business context. However, when I ask Copilot questions about the data, some responses reference incorrect metrics or misinterpret variables.
Could someone explain a reliable way to structure AI instructions to improve accuracy? Is there a method to enrich the semantic model so that synonyms are better recognized and mapped—beyond what’s currently defined in the Markdown instructions?
Thanks!
Solved! Go to Solution.
Hello! In addition to star schema, please ensure you are following the rest of the guidance in this article: Optimize your semantic model for Copilot in Power BI - Power BI | Microsoft Learn
Prepare your data for AI - Power BI | Microsoft Learn
Proud to be a Super User! | |
Hi @davidpc ,
Thanks for reaching out to the Microsoft fabric community forum.
I would also take a moment to thank @audreygerred and @ibarrau , for actively participating in the community forum and for the solutions you’ve been sharing in the community forum. Your contributions make a real difference.
If I misunderstand your needs or you still have problems on it, please feel free to let us know.
Best Regards,
Community Support Team
Hello! In addition to star schema, please ensure you are following the rest of the guidance in this article: Optimize your semantic model for Copilot in Power BI - Power BI | Microsoft Learn
Prepare your data for AI - Power BI | Microsoft Learn
Proud to be a Super User! | |
Hi. I think the purpuse of copilot is assisting you when developing. If you want an IA to ask questions about that, then the resource would be Fabric Data Agents. The agents let you configure different type of instructions and even query examples to a lakehouse.
Theoretical difference: https://learn.microsoft.com/en-us/fabric/data-science/concept-data-agent#difference-between-a-fabric...
I hope that make sense
Happy to help!