Movatterモバイル変換

This article answers frequently asked questions about Copilot for Microsoft Fabric and Power BI.

Note

Your administrator needs to enable the tenant switch before you start using Copilot. See the article Copilot tenant settings for details.
Your F or P capacity needs to be in one of the regions listed in this article,Fabric region availability.
If your tenant or capacity is outside the US or France, Copilot is disabled by default unless your Fabric tenant admin enables theData sent to Azure OpenAI can be processed outside your capacity's geographic region, compliance boundary, or national cloud instance tenant setting in the Fabric Admin portal.
Copilot in Microsoft Fabric isn't supported on trial SKUs. Only paid SKUs (F2 or higher, or P1 or higher) are supported. If you're using an Azure subscription with monthly credits (such as through a Visual Studio/MSDN benefit), you may not have access to the required SKUs unless billing is enabled and the appropriate capacity is provisioned.
For more information, see the articleOverview of Copilot in Fabric.

Power BI

Can Copilot be enabled for specific workspaces within a tenant?

Copilot is enabled at the tenant level and access can be restricted by security groups. If the workspace is tied to an F2 or P1 capacity, Copilot experience will be enabled.

When you're using Copilot, who has access to what data?

The data that Copilot can access depends on your role-level security and user-based permission on Fabric.

If you don't have permission to access specific data, then prompting Copilot for it won't retrieve the information.

Can Copilot prompts be saved for future reference?

Copilot prompts can't be saved for future reference. The only experience where it's possible to view your prompts is by using the chat-magic function in notebooks.

Does enabling Copilot and agreeing to the setting of "Data sent to Azure OpenAI can be processed outside your capacity's geographic region, compliance boundary, or national cloud instance" mean all my data is sent or processed outside my country/region?

Not exactly. While the prompt itself is sent to Azure OpenAI, it doesn't mean your data is sent or processed outside your country/region.

The prompt isn't used to train any models.

I loaded my semantic model, but it doesn't meet all the criteria listed in the data evaluation. What should I do?

The criteria listed inUpdate your data model to work well with Copilot for Power BI is important because it helps you get a better quality report. As long as you meet seven of the eight points, including Consistency, the quality of the reports generated should be good.

If your data doesn't meet that criteria, we recommend spending the time to bring it into compliance.

I was given a Copilot URL, but I can't see the Copilot button. Why is that?

First, check with your admin to see if they have enabled Copilot.

Next, when you select a Copilot-enabled URL, you have to initially load the semantic model. When you've completed loading the semantic model, then you see the Copilot button.

I selected the Copilot button, and it's stuck on Analyzing your semantic model.

Depending upon the size of the semantic model, Copilot might take a while to analyze it. If you've waited longer than 15 minutes and you haven't received any errors, chances are that there's an internal server error.

Try restarting Copilot by closing the pane and selecting the Copilot button again.

I loaded the semantic model and Copilot generated a summary, but I don't think that it's accurate.

This inaccuracy could be because your semantic model has missing values. Because AI is generating the summary, it can try to fill the holes and fabricate data. If you can remove the rows with missing values, this situation could be avoided.

I generated the report visuals, but the quality of the visuals concern me. I wouldn't choose them myself.

We're continuously looking to improve the quality of the Copilot-generated visuals. For now, we recommend that you make the change by using the Power BI visualization tool.

The accuracy of the narrative visual concerns me.

We're continuously working to improve the accuracy of the narrative visual results. We recommend using the custom prompts as an extra tool to try to tweak the summary to meet your needs.

I want to disable Copilot immediately as I'm concerned with the data storage you mentioned.

Contact your help desk to get support from your IT admin.

Why did Copilot respond with general knowledge data?

When a question is related to data in the semantic model, Copilot will use the semantic model to answer the question, otherwise it may answer from theLLM's general knowledge.

I want to suggest new features. How can I do that?

You can submit and vote on ideas for Microsoft Fabric on theIdeas page of the Fabric Community. Read more about giving feedback in theLearn about Microsoft Fabric feedback article.

Real-Time Intelligence

Does Copilot respond to multiple questions in a conversation?

No, Copilot doesn't answer follow-up questions. You need to ask one question at a time.

How can I improve the quality of the Copilot answer?

Provide any tips or relevant information in your question. For example, if you're asking about a specific column, provide the column name and the type of data it contains. If you want to use specific operators or functions, this will also help. The more information you provide, the better the Copilot answer will be.

What access level do I need on a KQL queryset to use Copilot?

You need read access to the KQL queryset to use Copilot. In order to insert and execute the Copilot-generated query in the KQL queryset, you need to have write access to that KQL queryset.

What database does the Copilot-generated query run against?

The Copilot-generated query runs against the database that the KQL queryset is connected to. If you want to change the database, you can do so in the KQL queryset.