Gemini Can Use Your Mac Now — in a Window That Isn’t Yours: Google Ships Computer Use to the Mac App

Catalina Clarke, Editor-in-Chief, The Library Media

The Library Media — Science, Technology & Health · AI News & Intelligence Desk

Gemini Can Use Your Mac Now — in a Window That Isn’t Yours: Google Ships Computer Use to the Mac App

By Catalina Clarke

Share

Published November 20, 2026 · 9:58 AM PT

On Thursday, November 19, Google gave the Gemini app for Mac the ability to use the computer itself. You describe a multi-step task in ordinary language, and Gemini handles the clicking, typing, menus and forms — across your local files, your desktop apps and the web — the way an assistant sitting at the machine would, except that it never touches your screen.

That last part is the design decision that matters, and Google leads with it. Computer use runs inside an isolated background window. A live preview plays inside the chat, so you can watch every step as it happens, while your own display and cursor remain completely free. You are not lending the machine to the agent; the agent works in a room of its own, behind glass, and you keep yours.

The controls Google states are specific, and they are the honest core of the announcement. You choose which apps Gemini is allowed to access. You watch every step live. You can pause the work at any time. And Gemini asks you to approve key actions before they happen, rather than reporting them afterwards. Read together, that is a permission model, not a demo trick: access granted app by app, work visible while it runs, a pause that is always available, and a human decision standing in front of the consequential steps.

What does Google imagine you handing over? The company’s examples are deliberately unglamorous, which is a point in their favour. Gemini can batch colour-grade a folder of photos, or set up 3D scenes. It can extract weights, dimensions and addresses from local PDFs and fill in the multi-box shipping quotes those numbers belong in. It can pull metrics and screenshots from live dashboards and assemble them into a Google Slides presentation. It can analyse clothing photos from a local folder, research pricing, and stage the listings on resale platforms. Files, forms, dashboards, listings — the connective tissue of office work, done in the background window while your own screen stays yours.

This is a different surface from the Gemini agent features the desk covered earlier this week, which ran errands inside the phone app — calls, checkout, bookings. Computer use on the Mac is the desktop version of the same idea, aimed at the machine where the files and the heavy applications actually live.

Availability, on Google’s own account, is a rollout in English in the United States, in the Gemini app for Mac, downloadable at gemini.google/mac. Reporting drawn from Google’s support documentation adds the finer print — the feature is described there as requiring a paid Google AI plan, a personal account for users 18 and over, and an Apple Silicon Mac running macOS Sequoia 15.0 or later — with one caution worth carrying: the announcement and some support pages do not describe the eligible tiers identically, with some documentation naming only the Pro and Ultra tiers for desktop-app control. If you are checking your own eligibility, the help pages for your account are the authority, not the launch post.

The limits deserve the same plainness as the features. Everything above about what Gemini can do is Google’s account of its own product, published at launch; the desk has not independently tested the feature, and the four worked examples are the company’s chosen demonstrations. A rollout in one language and one country, on one class of hardware, is a beginning rather than an arrival. And the question that follows every computer-use launch is not whether the agent can click — it is what happens the first time it clicks wrongly, in a window its user was only half watching. Google’s answer, to its credit, is built into the design it announced: narrow access, live visibility, a pause, and approvals in advance. Whether that answer holds in ordinary use is the story the next few months will tell.

For now, the fact is simple enough. The assistant that used to wait inside a chat box has been given a desk of its own — behind glass, on the Mac, with you watching. Google is betting you will let it work there. The controls it shipped alongside the capability suggest it knows exactly why you might not.

Catalina Clarke AI News & Intelligence Desk, The Library Media

From The Library — the estate of our family.

Follow Us

X Facebook Instagram Threads Mastodon Bluesky

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top