In this guide, I’ll explain how to share your screen with Copilot so the AI can see an app or even a website and help you with specific questions. It’s a pretty cool feature because it changes how you use the assistant: instead of just typing or speaking, you can also show it what’s in front of you. Don’t miss it!
Table of Contents
What is Copilot Vision?
Copilot Vision is a feature of Microsoft Copilot that allows AI to see what’s on your screen or, on other devices, even what the camera captures. Microsoft’s official idea is to turn Copilot into a kind of second pair of eyes that can examine what you’re seeing and give you real-time information based on the context.
In Windows, this feature allows Copilot to see open system windows or applications and help you with instructions, explanations, or answers about what’s on the screen. For example, you can share a specific app to ask it which button to press, ask it to explain an interface, or request help completing a task step by step. However, Microsoft makes it clear that Vision doesn’t act for you: it doesn’t click, type text, or control the computer on your behalf.
Another important point is privacy. According to Microsoft, Copilot Vision does not capture or store images to train models . It also does not save visual content once the session ends. What can be retained is the chat transcript and some model responses to monitor unsafe interactions. In any case, Microsoft emphasizes that images, audio, and visual context are deleted when the session ends.
How to share your screen with Copilot
The first thing you need is to have the Copilot app updated . It should look similar to the first screenshot, with the message box in the center and the glasses icon inside the compose box. If you don’t see that button, it’s most likely because you don’t have the correct version yet or the feature hasn’t been activated on your account.
Once you have the app ready, the next step is to tap the glasses icon , which is the shortcut to Copilot Vision. It’s located within the text editor, right next to other controls like the voice command.
Tapping that button opens a panel where you can choose what to share. Here, Copilot lets you select the entire screen or a specific application . This is useful because you don’t always want to show your whole desktop; sometimes it’s enough to show just the program you’re working in.
Once you choose what to share, Copilot launches the Vision session and can begin responding based on what it sees. Microsoft explains that Copilot Voice also comes into play at this point , allowing you to have a voice conversation to ask questions about the shared window. If this is your first time using the feature, you’ll likely see a notification requesting microphone access.
During the session, Copilot can even highlight parts of the screen to guide you. In theory, that’s one of the feature’s biggest advantages: it doesn’t just answer your questions, but can also point out interface elements to help you better understand what’s in front of you.
However, I’ve noticed a fairly clear difference between environments. In my tests within the Windows app, the results have been somewhat inconsistent . The feature is helpful and the idea makes sense, but it seems that Copilot doesn’t always interpret what’s on the screen correctly. I’ve also encountered difficulties completing transmissions on occasion, so something isn’t quite right.
Where it has really worked well for me is in Microsoft Edge . There, Copilot Vision has proven much more reliable at reading web page content, identifying blocks of text, and answering questions about what appears on the screen. In that context, the feature is quite convincing, although it can still occasionally be a bit unreliable in some responses.
In fact, the experience is more natural in Edge because Copilot seems to focus better on the open tab and the information on the website you’re viewing. The second screenshot shows the Vision floating bar over the page, indicating that the session is active and that you can continue asking questions while browsing.
This feature is definitely worthwhile, but with some caveats. On Windows, it still feels a bit rough around the edges, while in Edge, the results have been much more reliable and useful . Microsoft will likely iron out the bugs in future updates, so I encourage you to try it, especially if you plan to use it within the browser.
What can you use this feature for?
It’s certainly eye-catching to see Copilot analyzing a window in real time. However, the important question is whether it can actually be useful. And the truth is, it does have some interesting applications, especially when you need quick, contextual help without having to explain every detail in writing.
- Understanding a new interface. If you’ve just opened an unfamiliar program, Copilot Vision can help you identify specific menus, buttons, or sections.
- Following steps within an application. It’s a good option for requesting guided instructions while you have the program in front of you.
- Resolve doubts about Windows settings. This can be useful when you’re in Settings or any system tool and you’re not sure where to click.
- Analyze an open webpage. If you share the browser, you can ask it to summarize a part of the page or tell you where certain information is located.
- Get help without taking screenshots. Instead of taking a picture and then explaining the context, here you can directly show them the window you’re working in.
However, it’s important to keep expectations in check. It’s not a substitute for a well-made tutorial or someone who truly understands a specific application. But as a quick and contextual aid, the idea is interesting, especially if Microsoft improves the accuracy with which Vision interprets what’s on the screen.
