Copilot Vision.

Copilot Vision is a visual-context capability in Microsoft Copilot. In experiences that support it, a user shares selected screen content or camera input, then asks Copilot questions about what it can see. Depending on the experience and account, Copilot may also draw on other information available to that user.

The name describes a capability, not one uniform configuration across every Copilot app, device, or account. The ways to start a session, supported inputs, data grounding, and administrative controls may differ. Availability and behavior can also vary by account type, subscription, region, configuration, and service update.

Without visual context, a person may need to describe an interface or paste relevant text into a chat. Vision can make that exchange more direct: the user can ask about a chart, a setting, or a section of a document while it is on screen.

The shared view is still only a view. It may not include underlying data, hidden content, or information outside the selected screen or camera input. If a detail is too small, obscured, or ambiguous, Copilot may misread it. A fluent explanation should not be treated as proof that the interpretation is correct.

1. Open a Copilot experience and start a supported Vision interaction.
2. Choose the screen, app, page, or camera view to share, where that option is available.
3. Ask a question about the visible content. Clarify the task if the response misses the point.
4. Stop sharing or end the session when the visual context is no longer needed.

The exact controls and interaction style depend on the app and device. In some experiences, Vision is used during a voice conversation; do not assume every version works the same way.

• Copilot Chat is a conversational experience. Vision adds visual context in supported versions; it does not by itself determine which work data the account can access.
• Microsoft 365 Copilot can provide a work-oriented Copilot experience integrated with Microsoft 365 applications and, where configured and permitted, work data. Copilot Chat and Microsoft 365 Copilot can differ in grounding, application integration, permissions, and governance. Licensing and capabilities vary.
• Copilot Actions refers to action-taking experiences that may attempt to perform steps. Vision is primarily about interpreting shared visual content, not operating the computer for the user.
• Windows Recall is a separate Windows capability with its own purpose and controls. It should not be treated as another name for Vision.

An operations lead opens a dashboard and shares the relevant view with Copilot. They ask what the trend appears to show and which part of the chart may need a closer look. Copilot can help explain visible labels or summarize a pattern, giving the lead a useful starting point for analysis.

The lead should still verify the underlying figures, reporting period, and business context before acting. If the dashboard includes customer or employee information, they should also consider whether sharing that view is appropriate under organizational policy.

A shared screen or camera view may reveal information beyond the item the user wants help with. Organizations should evaluate the specific Copilot experience rather than infer its data handling or controls from the Vision name alone.

• Share only the content needed for the task, and close or obscure unrelated sensitive material.
• Confirm whether the user is signed in with a personal or work account.
• Review the applicable privacy information, tenant settings, and internal rules for sensitive data.
• Establish who handles questions about account access, device configuration, Copilot behavior, and policy.
• Treat Copilot’s interpretation as assistance, not as approval or validation of a business decision.

Can Vision take actions on the user’s computer?
Vision is for discussing visual content. It is distinct from agent experiences designed to carry out computer tasks.

Can it use work data?
That depends on the Copilot experience, account, permissions, and configuration. Sharing a screen does not, on its own, determine what other information Copilot can use.

Is Vision available to every Copilot user?
Not necessarily. Access and capabilities can depend on the app, device, account type, subscription, region, and rollout.

Which license, admin center, or support team applies?
There is no universal answer based only on the term “Copilot Vision.” These depend on the specific Copilot experience and how the organization has deployed it. Confirm them in the relevant product and account configuration rather than assuming a particular license, administrative control, or support commitment.

Copilot Vision is most useful when a person needs help understanding something visible on screen or through a supported camera experience. It can reduce the effort of describing context, while leaving the user responsible for selecting what to share and judging whether the response is reliable.

For workplace use, distinguish visual assistance from work-data access and from automated action-taking. That separation helps set realistic expectations, apply the right controls, and keep consequential decisions with the people accountable for them.