Microsoft tests Copilot Vision desktop share in Windows 11 Insider Build: Here’s what’s new

Microsoft is testing a new Copilot feature in the latest Windows 11 Insider build called Copilot Vision. It allows Copilot to view your entire desktop or specific app windows and respond in real time based on what’s visible. Earlier, it could only compare two apps side by side. Now it can work across your full screen to offer broader assistance.

Copilot can answer questions, suggest improvements, and guide you through tasks based on what’s on your screen. Microsoft says it may assist with things like editing your resume, offering feedback on creative projects, or helping you navigate a new game. The feature is similar to what Google’s Gemini Live and Apple Intelligence are beginning to offer.

What is Copilot Vision and how does it work?

Copilot Vision allows the assistance to access and interpret the visible content on your screen in real time. It helps you with what you are working on, reads that content and offers relevant suggestions, summaries or whatever you ask.

For example, suppose you have a PDF document open on your computer screen, you can ask Copilot to summarise it or tell main points of the content. If you are working in an Excel sheet, Copilot might help you spot errors, generate new formulas, help you in solving your issues with quick formulas, and compare values in a table without needing to describe the sheet. When browsing the web, it could highlight product differences or summarise a long review page. All of this happens through screen context, reducing the need to switch apps or copy and paste information.

Importantly, this feature uses screen capture permissions, and Microsoft says Copilot only accesses your screen after you give permission. You also have full control over what it sees and when it stops.

Real use cases for Copilot Vision

Now these will help you understand better how Copilot Vision helps you in real life.

• Document review: Copilot can pick out key terms or show what’s changed in a contract, so you don’t have to scroll through the whole thing.
• Spreadsheets: It can look at your sales data and give quick insights, fix errors, or turn numbers into charts right on your screen.
• E-commerce: Compare specs of products side by side directly from your screen without searching again.
• Email writing: If you have a report or slide deck open, Copilot could help draft a quick summary to send to a colleague.
• Troubleshooting: If you are stuck on a system setting or error screen, Copilot could explain the issue or offer a fix based on what's visible.

How to use Copilot Vision and Desktop Share in Windows 11?

The new Desktop Share feature makes it possible for Copilot to view your screen in real time and respond to you based on what is on the screen. To enable this feature open the Copilot app and click the glasses icon. From there, choose which part of your desktop to share. Once active, you can ask questions like “What does this error mean?” or “How do I create a table in Excel?” and Copilot will respond based on what it sees.

You can even stop sharing any time by selecting the "Stop" option in the same panel. You can also activate the Vision using your voice, thanks to its feature called Voice Conversation. You can verbally say your request and Copilot will respond using visual content on the screen. Copilot listens and responds based on what’s on your screen, without needing to type or click.

Microsoft has added a Highlights feature that shows helpful tips based on the apps, files or websites you are using without you having to ask.

Where Microsoft is headed with Copilot?

Copilot Vision stands out by removing the need to type long prompts. It sees what’s on your screen and helps right away, making it feel more like a built-in assistant than a chatbot. Unlike screen sharing or remote tools, it works locally and is just for you. 

The feature is still in testing with no set release date, but Microsoft is clearly moving toward assistants that respond to both what you type and what you see on screen.