Google Adds Computer Use to Gemini for Mac, Catching Up to Rivals

Googleplex, Google's headquarters in Mountain View, California
Photo: Wikimedia Commons / The Pancake of Heaven, CC BY-SA 4.0

Google has given Gemini computer use capabilities on macOS, letting the AI assistant complete tasks on a Mac by navigating native apps and browsing websites in the background. The feature started rolling out on November 19, 2026, in English to users in the United States, according to TechCrunch.

What the new feature actually does

The update lets Gemini work across a user’s local files, installed apps and the web to complete multistep tasks without the person driving each step themselves. While the agent works, a live preview of its progress appears inside the chat window, and the user’s own screen and cursor remain free for other work.

Google built in checkpoints so the AI does not act unsupervised: users can pause, stop or approve individual actions before Gemini carries them out. Suggested uses include filling out forms, repeating one task across a batch of files, sorting files into folders by criteria the user sets, and building slideshows from local files or data pulled from an online dashboard.

This kind of “computer use” design, where an AI model sees a screen or app environment and operates it the way a person would, by clicking, typing and navigating menus, has become the common shorthand across the industry for an agent that works on a device rather than only answering questions in a chat window. Google frames the Mac rollout as an extension of that same idea to everyday desktop chores rather than a narrow, single-purpose tool.

Source: TechCrunch.

Who gets it, and when

The rollout is limited for now. It requires the standalone Gemini app for Mac along with a paid Google AI Plus, Pro or Ultra subscription, and it is available only in English and only to users in the US at launch. Google has not given a timeline for expanding the feature to other languages, countries or platforms.

The launch follows Google’s wider rollout of Gemini “Tasks” the week before, around November 17, 2026, an in-app agent feature aimed at signing into websites, making purchases and booking appointments on a user’s behalf. Google has also been letting early testers into “CC,” a family-organization agent it first announced in September 2026 that coordinates email, calendars, chats and tasks across a household.

The Gemini app itself has been built up on the Mac in stages through the year rather than arriving all at once. Google first shipped a native Gemini app for macOS in April 2026, added system-wide voice dictation and screen-aware commands in the middle of the year, and introduced a background file-organizing agent called Gemini Spark, limited to Ultra subscribers, before this computer-use expansion widened what the assistant is allowed to do on a user’s machine.

Why this counts as Google catching up

OpenAI, Anthropic and Meta have each already shipped similar agentic features in their own desktop apps, and TechCrunch described Google as “a little late” to offering the same kind of computer-use capability on the Mac. The company’s AI assistant has instead built up its desktop presence in stages over 2026, starting with a native macOS app in April, adding natural-language voice capabilities in the middle of the year, and introducing a background file-organizing agent called Gemini Spark for Ultra subscribers before this computer-use expansion.

That staged approach puts Google’s desktop agent rollout a step behind competitors that moved to full computer-use agents earlier, even as Google has pushed hard on AI features inside the Gemini app itself. The computer-use rollout comes just a day after Google introduced its Gemini 4 Neon model, a mid-tier model the company pitched on coding and enterprise performance, underlining how much of Google’s current AI push is arriving in rapid, overlapping releases rather than single large launches.

What is still missing

Google has not published pricing details specific to the computer-use feature itself beyond the existing Plus, Pro and Ultra subscription tiers, and it has not said which of those tiers unlocks which depth of capability. There is also no confirmed timeline yet for Windows, for languages beyond English, or for markets outside the US.

TechCrunch’s report does not include a statement from a named Google spokesperson on the record, and no independent benchmark or user-testing data comparing this feature to OpenAI’s, Anthropic’s or Meta’s equivalents has been published yet.

Frequently asked questions

Do I need to pay for this feature? Yes. It requires a Google AI Plus, Pro or Ultra subscription along with the Gemini app for Mac; a free Gemini account does not get computer use.

Is it available outside the US yet? No. At launch it works only in English for users in the United States, and Google has not announced an expansion timeline.

Can Gemini act on my Mac without my approval? Google says users can pause, stop or approve each action, and a live preview shows what the agent is doing while it works.

Whether Google expands the feature beyond English and the US, and how it performs against the agentic tools already running on OpenAI’s, Anthropic’s and Meta’s desktop apps, will be the next things to watch.

Source: TechCrunch.