Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
APPLIES TO: 
Power BI Desktop
Power BI service
In Power BI, Copilot helps you use the transformational power of generative AI to get the most from your data. It provides chat-based experiences where Copilot can help you with tasks from on-the-fly analysis for business users to data analysis expression (DAX) generation for advanced creators. The product also includes various AI-infused Copilot features.
This article provides an overview of the Copilot capabilities for Power BI.
Note
Copilot is enabled by default. Administrators can turn off Copilot in the Fabric admin portal if your organization isn't ready to use it. For details on turning Copilot on and off, see Enable Fabric Copilot for Power BI.
Some Copilot experiences are generally available, and others are in preview. The report agent Copilot pane available on the right side of reports is generally available. The Power BI agent available as a standalone, full-screen experience accessible from the Power BI left navigation is in preview. The Power BI app agent available as a full-screen screen experience accessible from an app's left navigation is in preview. Copilot experiences for Data Factory, Data Engineering, Data Science, Data Warehouse, and Real-Time Intelligence are in preview.
Sovereign clouds aren't supported. Copilot isn't yet supported for sovereign clouds due to GPU availability.
If you have questions about privacy or security with Copilot, see Privacy, security, and responsible use for Copilot in Power BI.
Requirements at a glance
Before exploring capabilities, confirm that your environment meets the minimum requirements:
| Requirement | Details |
|---|---|
| Capacity | Paid Fabric capacity (F2 or higher) or Power BI Premium (P1 or higher). Trial capacities and free SKUs aren't supported. |
| Admin setting | The Users can use Copilot and other features powered by Azure OpenAI setting must be on. Copilot is enabled by default, but your admin might turn it off. |
| Region | Your Fabric capacity must be in a supported region. |
| Sovereign clouds | Not supported. |
| Data preparation | For best results, model owners should prep their semantic models for AI before use. |
For experience-specific requirements (report pane, standalone Copilot, apps), see Copilot requirements.
Prompt limits
Copilot for Power BI supports up to 10,000 characters per prompt in all Copilot surfaces, including the standalone experience, the report Copilot pane, Copilot in apps, Power BI Mobile, and Power BI Embedded. This expanded limit allows you to provide longer, more detailed instructions and richer context in a single message.
Data preparation
You need to prepare data to work with Copilot. Model owners need to invest in prepping their data for AI to ensure Copilot understands the unique business context, prioritizes the right information, and delivers responses that are consistent, reliable, and aligned with your goals. Without this prep, Copilot can struggle to interpret data correctly - leading to generic, inaccurate, or even misleading outputs. Learn how to prep your data.
Capabilities
Copilot in Power BI offers a range of capabilities designed to enhance productivity and streamline workflows for business users, report authors, and data model owners. These features use AI to simplify complex tasks, provide actionable insights, and improve the overall data experience.
Copilot for business users: Chat with your data
For business users, chatting with your data can mean many things. It might involve finding content. It could include asking for ad-hoc analyses or insights. You can also quickly create and analyze visuals. You might even get summaries of entire reports or focus on specific topics. All these tasks are similar to what you might ask your analyst to do when you have follow-up questions about a report they made. When a question is related to data in the semantic model, Copilot uses the semantic model to answer the question. Otherwise, it answers from the large language model's (LLM) general knowledge. Right now, three chat-with-your-data experiences exist for business users: the standalone Copilot experience, the Copilot pane for reports, and Copilot in apps (scoped to curated app content and supports verified answers by app authors).
The standalone Copilot experience (preview): Find data and ask questions about any item
Copilot in Power BI is available as a standalone, full-screen experience that you can use to find and analyze any report, semantic model, and Fabric data agent you have access to. Unlike the Copilot pane, which only answers questions about the report you currently have open, the standalone Copilot finds and answers questions about any data you have access to. Learn more about the standalone Copilot.
The standalone Copilot can help with the following tasks:
- Search for artifacts.
- Summarize a report or topic.
- Answer questions about your data.
- Answer questions using a Fabric data agent.
Copilot in apps (preview): Ask questions and get summaries scoped to an app
You can access the app-scoped Copilot experience from an app's left navigation. Use it to search and ask questions based on the curated content included in that app - such as reports, dashboards, and related artifacts. Unlike the report-scoped Copilot pane (which is limited to the open report), app-scoped Copilot operates across the app's curated scope. It provides summaries of app content and answers to questions based on what users have access to within the app.
App-scoped Copilot supports verified answers prepared by app authors. When verified answers exist for common questions, Copilot can surface those author-provided responses to increase reliability and consistency. Learn more about Copilot in apps.
Considerations and limitations
When you purchase new capacity or scale up capacity, it can take up to 24 hours for Copilot to recognize the change and become available for use.
The standalone Copilot and Copilot in apps experiences aren't yet available in the following regions: India West, Indonesia Central, Korea South, Malaysia West, New Zealand North, Qatar Central, Taiwan North, Taiwan North West, UAE Central, France South, Germany North, Norway West. While prompts submitted in languages other than English might occasionally return relevant responses, multilingual use isn't officially supported at this time.
Responses in app-scoped Copilot are limited to the app's included content and the permissions users have within the app. When verified answers are available, the app might prioritize and return them for matching questions.
The Copilot pane: Ask questions about an open report
Business users can use the Copilot pane available on the right side of reports to ask a question about report content or summarize key insights in seconds. The report agent can help business users with the following tasks:
Business users can also include summaries when setting up subscriptions to reports.
Copilot for report authors: Build reports and enhance your models
Report authors can use Copilot in many ways, such as the following tasks:
- Create and edit a report
- Learn how to Write Copilot prompts for creating report pages
- Summarize the underlying semantic model
- Create a summary visual on the report itself
- Write DAX queries
- Add descriptions for your semantic model measures
Some of these capabilities are available through the Copilot pane, whereas others are accessible through inline Copilot experiences.
Clear the Copilot chat
The clear chat button removes the contents of the current conversation and any previous context set throughout the conversation. This action is essentially the same as starting a new chat. This action doesn't refresh the underlying data. Use clear chat when switching topics to avoid overloading Copilot with unrelated prior context. Clear the chat to ensure clear and focused responses.
If you ask the same exact prompt on an unchanged semantic model (same schema, data, and custom instructions) within a rolling 24-hour window, Copilot responds from the cache for efficiency. You see the same response in this case. The clear chat button doesn't affect this. If you aren't expecting the same response, try changing your prompt or refreshing the underlying model.
Copilot requirements
To ensure a seamless experience with Copilot in Power BI, make sure you meet specific requirements. These requirements vary depending on the type of Copilot experience you're using, such as the standalone Copilot or the Copilot pane within reports. Here's a breakdown of the general and specific requirements for using Copilot effectively.
Tip
Can't find Copilot? If Copilot isn't available in your Power BI environment, see Enable Fabric Copilot for Power BI for step-by-step instructions on how to enable it. Copilot is enabled by default, but your administrator might disable it or your organization might not meet all the requirements listed in this section.
General requirements
The following requirements pertain to all Copilot experiences in Power BI.
- Your administrator needs to enable Copilot in Microsoft Fabric.
- Your organization needs a paid Fabric capacity (F2 or higher) or Power BI Premium capacity (P1 or higher). A Power BI Pro or Premium Per User (PPU) license alone isn't sufficient—Copilot requires organizational capacity.
- Copilot in Microsoft Fabric isn't supported on trial SKUs or trial capacities. Only paid SKUs are supported.
- Your Fabric capacity needs to be in one of the regions listed in Fabric region availability. If it isn't, you can't use Copilot.
- If your tenant or capacity is outside the United States or France, Copilot is disabled by default unless your Fabric tenant admin enables the tenant setting called Data sent to Azure OpenAI can be processed outside your tenant's geographic region, compliance boundary, or national cloud instance in the Fabric admin portal.
- Due to GPU availability, Copilot doesn't yet support sovereign clouds.
Requirements for the standalone Copilot
The following requirements pertain to the full-screen standalone Copilot experience you access through the left navigation.
Enable tenant settings
- Enable the setting called Users can use Copilot and other features powered by Azure OpenAI at the tenant level. Enabling it at the capacity level (that is, if delegated) isn't sufficient.
- Enable the tenant setting called Users can access a standalone, cross-item Power BI Copilot experience (preview).
Access to a Copilot-supported capacity
To use the standalone Copilot experience, you must have access to a Copilot-supported capacity. Use a dedicated Fabric Copilot capacity (FCC). If your organization has an FCC, all Copilot usage across Power BI automatically bills to a single, centralized capacity. If your organization doesn't have an FCC, standalone Copilot automatically selects a workspace linked to a capacity that supports Copilot (that is, it's linked to an F2 capacity or higher, it's in a supported region, and it's enabled for Copilot) for usage tracking and billing. If you don't have access to an FCC or access to a workspace backed by a capacity that supports Copilot, you can't use the standalone Copilot experience.
Autoselection of a Copilot workspace
The service automatically selects a workspace from a partially randomized list of eligible workspaces, weighted toward workspaces with more available capacity. This approach helps balance usage and prevents overloading any single capacity. This selection persists across sessions as long as the workspace remains Copilot-enabled and has available capacity. If the workspace becomes disabled for Copilot or reaches its capacity limit, the service automatically reassigns a new one and notifies you. If an FCC becomes available, it always overrides the autoselected workspace. You receive a dismissible notification showing the chosen workspace and a direct option to change it.
Change your Copilot workspace
You can update your Copilot workspace anytime through More > Manage workspace in the standalone Copilot experience. Once you manually select a workspace, the service only overrides it if an FCC becomes available. If your chosen workspace becomes disabled or reaches its capacity limit, the service alerts you and provides options to select a new workspace manually or through autoselection.
Requirements for Copilot within reports
These requirements pertain to using Copilot within Power BI reports, including the Copilot pane experience.
- Power BI Desktop:
- You need write access to a workspace that's on a paid Fabric capacity or Power BI Premium in the Power BI service, where you plan to publish the report. Read more about using Copilot in Power BI Desktop.
- Power BI service:
- The report must be located in a workspace linked to either Premium Power BI (P1 and above) or a paid Fabric capacity. Learn how to check your license type.
- To generate summarized insights in reports, you need at least read access to the workspace, report, or app.
- To add a narrative visual to a report or generate Copilot reports, you need edit access to the workspace or report, or build access to the semantic model.
Requirements for app-scoped Copilot (preview)
These requirements pertain to using app-scoped Copilot in Power BI apps, the full-page Copilot experience accessible from app navigation.
- Meet the general requirements for Copilot, including the setting called Users can use Copilot and other features powered by Azure OpenAI must be enabled at the tenant level. It isn't sufficient to enable it at the capacity level (that is, if delegated).
- Like standalone Copilot, app-scoped Copilot uses an FCC or an autoselected workspace for usage tracking and billing. Or you can manually select a workspace.
- The setting for showing Copilot in the app navigation must be enabled. You manage this setting while publishing or updating an app under Advanced settings > Show Copilot in app navigation. Each app has its own setting.
Note
Copilot in apps is on by default and you can manage it per app.
Power BI Copilot compute usage
The Microsoft Fabric Capacity Metrics app provides visibility into compute consumed by Copilot in Power BI. Capacity admins can use the Item history page to view a 30-day breakdown of Copilot usage, measured in capacity units (CUs). By filtering on the Copilot in Fabric operation, you can isolate all compute usage generated by Copilot requests and break it down by Power BI experience (report, semantic model, or mixed) to understand where compute demand originates.
The system processes all Copilot CU consumption as background capacity operations. This design smooths demand and prevents sudden compute spikes, which helps ensure a more consistent experience across the capacity. Token consumption drives billing for Copilot in Fabric. Fabric bills separately for any downstream actions that Copilot triggers, such as DAX queries, data refreshes, or email subscriptions, through their standard Fabric pathways.
For more information, see Understand the metrics app item history page.