Standalone Copilot experience in Power BI (preview)

Copilot in Power BI is available as a full-screen, standalone Copilot experience that finds and answers questions about any report, semantic model, and Fabric data agent they have access to. This experience is different from the Copilot pane, which only answers questions about the report you currently have open.

Capabilities

The standalone Copilot experience in Power BI offers a range of powerful capabilities to help you interact with your data more effectively. Copilot provides a conversational interface that simplifies data exploration and analysis. Here are some key features available in this experience.

Search

You can ask Copilot to find reports, semantic models, apps, and data agents that you have access to. In response, you get a hyperlinked list of relevant items, with helpful details to make it easy for you to find the right thing. Learn more about searching for items

Summarize a report or topic

Copilot can provide summaries about an entire report or a specific topic within that report. You can use prompts like:

• "Summarize the relationship between weather and visits to Hawaii."
• "Summarize trends in sales last year."

Summaries in the standalone Copilot experience include both textual insights and relevant visuals within the narrative. Currently, summaries are only available for reports, not for semantic models. Learn more about summarizing a report or topic.

Ask questions about your data

Copilot can answer questions and create visuals using the measures and other data fields in your model. It can also create new data analysis expression (DAX) calculations if needed.

Unlike the Copilot pane within a report, the standalone Copilot experience doesn't reference report content when answering questions. Answers are always generated directly from the underlying model. Learn more about asking questions about your data.

Use a Fabric data agent

You can use Fabric data agents to get answers without leaving the standalone Copilot experience. Fabric data agents are custom-built experts in specific topics, trained by a developer within your organization.

Choose the right data

When you ask a question about your data, whether you're looking for a summary or have a specific query, the first step is selecting the appropriate data to answer it. In the standalone Copilot experience, you can choose the right data in two ways:

• Attach a report or semantic model: If you know exactly which report or semantic model you want to reference, you can attach it to your chat using the Add items for better results option.

  

  Then it appears in the Copilot conversation and is used to answer your questions.

  

• Let Copilot find the right data to answer your question: Copilot looks through the data you have access to, give you a few suggestions, and ask you to confirm which you'd like to use. It also presents some metadata about these suggestions, such as their workspace, if you opened them recently, and why they were suggested.

  

Prep data for AI

You must prepare your data for it to work well with Copilot. Preparing your data for AI creates the foundation for high-quality, grounded, and context-aware AI experiences. When data is unstructured or ambiguous, AI systems can struggle to interpret it correctly - leading to generic, inaccurate, or even misleading outputs.

If your data isn't prepared for Copilot, you get a warning that the answer might be low quality. You can choose to explore the data or view the answer anyway.

If you select the View answer button, the response appears in chat.

Once your data is marked as prepped for AI, responses appear in the chat without the warning.

Copilot requirements

Read about requirements to use the standalone Copilot experience.

Considerations and limitations

The standalone Copilot experience isn't yet available in the following regions: Spain Central, Qatar, India-West, and Mexico.