FAQ for apps in Microsoft Copilot Studio (preview)

Apps in Microsoft Copilot Studio are interactive web experiences that makers create by describing what they need in natural language. Copilot Studio generates a working first draft that the maker can review in a live preview, refine through conversation, test, publish, and share. Apps are currently in preview, and availability and supported capabilities can vary by tenant, environment, license, administrator settings, and organizational policies.

These frequently asked questions explain what apps in Copilot Studio can do, how you can use them, how performance is measured, and more.

What can apps in Copilot Studio do?

Makers can use Apps in Copilot Studio to:

  • Create an app from a natural-language description.
  • Generate screens, forms, fields, tables, views, filters, actions, and sample data.
  • Preview and interact with the app while it's being built.
  • Request changes to the app's layout, data, or behavior through conversation.
  • Use supported files, lists, tables, and data connections available in their environment.
  • Inspect generated files in a read-only code view.
  • Publish the latest version and share the app with permitted people or groups.

The capabilities available to a maker depend on the environment, connections, and policies configured by their organization.

What are the intended uses of Apps in Microsoft Copilot Studio?

Apps in Copilot Studio are intended to help business makers with a range of technical experience create task-focused productivity and business-process apps more quickly.

Common scenarios include tracking work or inventory, collecting information, managing requests, reviewing business data, and creating dashboards for an individual or team.

The generated app is intended to be a starting point that the maker reviews and refines. Organizations should continue to use their normal design, security, accessibility, compliance, testing, and release processes for complex, regulated, or business-critical scenarios.

How were apps in Microsoft Copilot Studio evaluated? What metrics are used to measure performance?

Microsoft evaluated apps in Copilot Studio through automated and manual testing. The evaluation criteria included whether generated apps:

  • Build successfully.
  • Address the maker's request.
  • Work as expected for common app scenarios.
  • Provide a useful and well-designed experience.

Microsoft also evaluates how the system responds to inappropriate, malicious, misleading, and adversarial requests. Findings from these evaluations help improve system instructions, safeguards, test coverage, and product behavior.

Evaluation continues as the product, supported models, and app-building capabilities evolve. User feedback also helps Microsoft identify inaccurate, unexpected, or harmful results.

What are the limitations of apps in Microsoft Copilot Studio? How can users minimize the impact of these limitations?

The current preview has the following limitations:

  • Generated layouts, fields, calculations, actions, sample data, or app behavior might be incomplete or incorrect.
  • Complex apps with many roles, conditional paths, integrations, or specialized rules can require more iteration.
  • The code view is read-only. Makers make changes by describing what they want Copilot to update.
  • Available data sources, connectors, and operations depend on the maker's environment and organizational policies.
  • Sharing an app might not automatically share its underlying data. App users might need separate permission to connected SharePoint content, Dataverse data, or other resources.
  • Results can vary across languages and cultural contexts.
  • Preview features, limits, connections, and deployment options can change before general availability.

To minimize the impact of these limitations, makers should provide clear requirements, build complex apps in smaller steps, and review the preview after each change. Before publishing or sharing, test the app's main scenarios, calculations, sample data, links, permissions, connections, accessibility, and error handling.

Microsoft implemented safety instructions and content filtering to reduce harmful or inappropriate outputs. Because AI-generated results can still be inaccurate or unexpected, makers should review and test each app before making it available to others.

What operational factors and settings allow for effective and responsible use of apps in Microsoft Copilot Studio?

The following practices support effective and responsible use:

  • Clearly describe the app's purpose, intended users, data, fields, actions, and business rules.
  • Break complex requirements into smaller changes and review the app after each change.
  • Test with representative accounts and data before using production data or sharing broadly.
  • Confirm that app users have the permissions and connections required for underlying data sources.
  • Use the Personal Developer Environment for creation, iteration, and testing, subject to the tenant and environment policies configured by administrators.
  • Review the generated app in the live preview and use Ask for follow-up changes to request corrections or improvements.
  • Publish only after the main scenarios work as expected.
  • Share the app only with people who should be able to open and use it.

Publishing and sharing are separate actions. Publishing makes the latest version available to the people the app has been shared with. Sharing controls who can open and use the app, subject to the organization's policies.

How do I provide feedback on apps in Microsoft Copilot Studio?

Select thumbs up or thumbs down on a Copilot response, and add comments to suggest improvements, report a problem, or describe an inaccurate, unexpected, or harmful result.

  • Last updated on