Gemini’s Share – Google Gemini’s Mac app, launched in April, includes a “Share Window” tool that lets the assistant watch a specific open window and guide you step by step as you work—turning screen sharing into a real-time lesson rather than remote control.
A few weeks ago, the moment felt simple: I was stuck in Photopea, a free online Photoshop alternative, trying to place an image over a colorful border with a drop shadow behind it. The menus were dense. The documentation wasn’t helping fast enough.
So instead of hunting through guides or YouTube tutorials, I shared a live view of my web browser with Gemini and asked for help. Google’s AI assistant didn’t just offer a vague suggestion. It walked me through Photopea’s complex menus step by step—right where I was working.
That experience led to a less flashy realization: there’s a Gemini feature many people may overlook, and it’s built around teaching rather than taking over.
In the Gemini Mac app—launched in April—this starts when you click the “+” button. From Gemini’s list of tools, you can find the “Share Window” option. Dragging the cursor over that option brings up a list of open windows you can share with Google’s AI assistant.
Using it does require extra privacy permissions. The setting lives under Settings > Privacy & Security > Screen & System Audio Recording. From there, you can turn on the Gemini toggle so the app can automatically take screenshots.
Once you’ve shared a window, Gemini takes a screenshot of that window each time you post a question. That’s the core mechanic: it lets Gemini act alongside your other apps while giving help in context, instead of forcing you to pause and manually capture new screen images.
For my own image border, the stumbling block was applying a gradient effect to the background. Gemini looked at which Photopea menu was open and then told me exactly which buttons to click, citing Photopea’s online documentation.
The same “look over your shoulder” approach showed up in other tasks. Gemini helped me navigate the Fangraphs website while I was looking up recent baseball statistics. It also guided me through Raycast’s Settings menu to enable scripts after I’d “vibe coded” a couple of Raycast scripts for window management.
Other desktop AI apps can share screens, but the workflow is often clunkier. ChatGPT and Claude both require manually adding new screenshots when something changes on your screen. With Claude, you have to click and drag to define the capture area each time.
Gemini’s Share Window mode, by contrast, behaves more like a teacher who keeps observing the same problem area—then steps in with the next instruction when you ask.
This matters because the wider desktop AI direction is moving fast. Many rivals are focusing on computer control rather than guidance. Claude’s desktop app and OpenAI’s ChatGPT Codex app now offer “Computer Use” modes that can navigate through the desktop using virtual cursors and keyboards. using persistent screenshots to guide them. The idea is automation: handling complex tasks even when you aren’t sitting at the computer.
Google has been pointed in that direction too. While the Gemini app can’t control your computer today, Google started previewing a Computer Use model for Gemini last fall.
But full control comes with trade-offs. Anthropic warns of security risks from malicious apps and web pages—situations where a user could be overruled by a harmful instruction chain. It also cautions against letting AI make decisions with “meaningful real-world consequences” without human confirmation first. There’s also a practical friction point: AI can be slower when clicking through menus and buttons. And there’s the privacy angle—letting these companies see everything on your screen can become “a potential privacy nightmare.”.
On Windows, Google’s approach takes a slightly different route. There’s no desktop Gemini app for Windows in this account. but Google offers a separate desktop app with a similar share screen feature. The conversation flows through Google Search’s AI Mode rather than Gemini. Microsoft’s Copilot app also has a screen-sharing feature, though in my experience its instructions haven’t been as helpful.
The tension underneath all of this is where the value should sit as AI becomes more capable. If the future is automation-first, the risk is losing the slower, useful work of learning—figuring out tools and workflows one click at a time.
My hope is that even as computer use becomes a bigger focus. Google doesn’t abandon the role Gemini already plays here: acting as a software tutor. not just a remote operator. Not every task needs to be automated away. There’s still real value in learning to do it yourself—and in getting help that meets you at the exact point you’re stuck.
Gemini Google Mac app screen sharing Share Window Photopea AI assistant privacy permissions Fangraphs Raycast computer use Anthropic warnings