Nobody enjoys the boring half of computer work: the same form filled in forty times, a downloads folder that looks like a junk drawer, a slideshow that needs building from three different dashboards. On 19 November 2026, Google gave Mac owners a new way to hand that work off. The Gemini app for Mac can now use your computer for you — opening your apps, moving through your files and browsing the web in the background while you get on with something else.
What "computer use" actually means
Computer use is the industry's name for an AI that doesn't just answer questions but operates the machine. Instead of telling you how to do something, Gemini does it: it navigates your native Mac apps, clicks through websites and works across local files and the web in a single, multi-step job.
The detail that matters most is how it shows its work: you watch the agent operate in a live preview inside your Gemini chat, while your actual screen and cursor stay free. You can keep writing an email while Gemini quietly sorts files or fills in forms in the background, and you can pause it, stop it, or approve its actions before they happen.
What it can do on your Mac today
Google's examples are deliberately unglamorous — the feature targets the repetitive, fiddly jobs people put off:
- Filling out forms — the same information entered across many pages or documents, without you retyping it each time.
- Batch work across files — repeating one task over and over across a set of files, rather than opening each one yourself.
- Organising files — sorting documents into folders based on criteria you set, so a messy folder becomes a tidy one.
- Building slideshows — pulling material together from your files or from information in online dashboards and turning it into a presentation.
It is the same direction the whole AI industry has been moving in. OpenAI, Anthropic and Meta already offer agentic features in their desktop apps, and Google is, by its own timing, late to this particular party. The Mac launch also lands in the same week as the broader rollout of Gemini Tasks, the in-app mode that lets Gemini call businesses, book appointments and make purchases on your behalf — a related but separate feature that works inside the Gemini app rather than across your whole computer. Google has also begun admitting early testers to a revamped "CC" agent, announced in September, designed to help families stay organised across email, calendars, chats and tasks.
The controls: you stay in charge
An AI that can click around your computer raises an obvious question: what stops it doing something you didn't want? Google's answer is the approval step. You can pause or stop Gemini at any time, and you approve its actions before they happen. The live preview means you see what it is doing as it does it, rather than discovering the result afterwards.
That control is the feature's real test: a form filled correctly is a small delight, while a file moved to the wrong folder is the opposite. Anyone trying computer use for the first time should watch the first few jobs closely rather than letting it run unsupervised.
Google is late — and Apple is coming
The timing is awkward for Google in one specific way: the company whose computer this runs on is reportedly building the same thing. Apple is said to be developing its own AI agent that can control and automate tasks on a Mac, joining a crowded field that already includes Meta's Muse, Anthropic's Claude and OpenAI's offerings. If Apple ships a native agent baked into macOS, Google's version will have to earn its place rather than simply being first.
That competition is good news for users. Each company is now racing to make the agent that does the errand, not just the one that explains it. OpenAI's ChatGPT Wallet already lets its agents pay for things with a tokenised card, and Gemini's own Tasks mode covers calls and bookings. The Mac's desktop is simply the next place that race is being run.
Who gets it — and the catch
There are two limits to know before you go looking for the feature. First, it is initially available in English in the US only, and Google has not said when it will reach other countries or platforms. Second, it needs the Gemini app for Mac and a paid Google AI plan — Plus, Pro or Ultra. Paid AI plans are now the norm across the big assistants; ChatGPT Plus set that pattern, and Gemini's computer use follows it.
So for most readers outside the US, this is a preview rather than something to switch on today. The rollout will be gradual, and the free tier of Gemini does not include it.
What this means for everyday AI
A chatbot that writes a good email is useful. An assistant that fills in the form, files the document and builds the slideshow while you do something else is a different kind of product altogether. That is the shift Google is betting on with computer use on the Mac — and it is the same shift behind Tasks, ChatGPT Wallet and Apple's rumoured agent.
The technology is here, at least in the US and behind a paywall. The open question is trust: how many people will let an AI loose on their files and apps, even with a pause button and a live preview? Google's answer is to start with the boring jobs, where the upside is obvious and the risk is low.




