Announced 19 Nov 2026 · Sources checked
What can Gemini do on a Mac?
Google's Gemini app for macOS can now take actions directly on the user's computer.
The company announced the computer use feature on 19 November 2026, letting the AI navigate through local files, desktop applications, and websites while the user continues working on other tasks.
According to TechCrunch, Google suggests the feature could handle repetitive tasks like filling out forms, organizing files into folders based on specific criteria, batch-processing the same action across multiple files, creating slide decks from live dashboard data, and staging listings for items on resale platforms.
The company also highlighted creative workflows such as batch color-grading photos in design software or setting up complex 3D scenes.
The rollout comes alongside a broader expansion of Gemini's agentic capabilities. Google only this week began widely rolling out Gemini Tasks, a feature that lets the AI handle web-based actions such as signing into sites, making purchases, or booking appointments.
The company also recently opened early testing for Project CC, a personal AI agent announced in September that manages email, calendars, chats, and tasks for families.
Together, these features signal Google's push to position Gemini as a cross-platform productivity agent that can act both inside the browser and on the desktop, rather than a simple chatbot confined to a text window.
The computer use feature is the most significant of these launches because it extends Gemini's reach beyond Google's own ecosystem. Unlike Tasks, which operates within the web browser, computer use works across any native macOS application the user permits.
That means the AI can interact with third-party design software, file management tools, and productivity suites that have no direct integration with Google's services.
It also means Gemini inherits all the complexity of navigating arbitrary desktop interfaces, where buttons, menus, and form fields are not tagged with machine-readable labels the way web pages are.
How does the computer use feature work?
The AI operates inside an isolated background window that shows a live preview of its actions directly in the chat interface, according to Google's blog post.
The user's main screen and cursor remain completely free, so they can keep working on other things while Gemini handles the clicking, typing, and menu navigation.
Michael Friedman, who authored the Google announcement, described the workflow: the user describes what they need, and Gemini handles the clicking, typing, menus, and forms automatically.
The live preview lets users watch every step, and they can pause at any time or approve key actions before they happen.
The feature is available through the Gemini app for Mac, downloadable from gemini.google/mac. It is initially rolling out in English for users in the United States.
Google has not announced a timeline for other languages or regions. The feature requires a Google AI Plus, Pro, or Ultra subscription.
Why is Google adding this now?
Google enters the desktop computer use space well behind its competitors. TechCrunch reported that OpenAI, Anthropic, and Meta have been offering similar agentic features in their desktop applications for some time.
OpenAI's ChatGPT desktop app gained computer use capabilities earlier this year, and Anthropic's Claude has been available on macOS with file and app interaction since 2025.
The timing is notable because Apple is simultaneously moving to restrict what AI agents can do on Mac.
A preview of new macOS restrictions for apps requesting full disk access rolled out with this week's macOS 27.3 beta, according to The Verge. These restrictions explicitly address the potential risks from AI agents with computer use capabilities.
Google's launch before those restrictions take effect gives the company a window to establish the feature on the platform.
What about privacy and safety?
Google built several safeguards into the computer use feature. Users choose which apps Gemini can access, rather than giving the AI blanket permissions.
The background window approach means actions are visible and auditable in real time. The company says users can pause any action, stop the AI at any point, and must approve critical actions before they execute.
The Verge noted that Google positions these controls as privacy and safety measures, distinguishing its approach from competitors who may request broader system access. The app-level permission model gives users granular control compared to a full-disk-access alternative.
TechCrunch added that as with other AI agents offering computer use, users can pause, stop, and approve Gemini's actions before they happen, giving them oversight over what the AI does on their machine.
The live preview window lets them follow along without giving up their own screen space.
How does it compare to similar tools?
Google joins a rapidly expanding category. Vercel Labs earlier open-sourced agent-desktop, a Rust CLI that gives any AI agent computer use through accessibility trees.
Cursor added computer control capabilities for its coding agent, including iOS Simulator testing. DeepSeek has macOS and Windows desktop builds of its Harness tool.
The market for desktop AI agents has grown from experimental demos to production features from every major AI company.
The late arrival means Google can observe what has worked and what has caused problems. Anthropic's computer-use explainer laid out the fundamental challenge: every click depends on a new screenshot, making the approach token-intensive and slower than API-native integrations.
Google's emphasis on user-permission controls and an isolated background window suggests it has designed around some of the failure modes seen in earlier implementations, but it has not disclosed technical benchmarks or latency figures for its approach.
Whether users actually want an AI moving files on their desktop remains an open question.
Uber reported in September that its AI agents booked just four rides in the entire third quarter, suggesting that demand for autonomous agent actions on behalf of users may take time to materialize even when the technology is ready.
For Google, the bet is that Gemini's integration with the broader Google ecosystem — Gmail, Drive, Calendar, and Workspace — gives it an advantage that standalone agent tools cannot match.
Common questions
Does Gemini on Mac computer use work with any app?
Users can decide which apps Gemini can access. The feature works within those allowed applications across local files, desktop apps, and the web.
Is Gemini on Mac computer use free?
No. Computer use requires the Gemini app for Mac and a Google AI Plus, Pro, or Ultra subscription.
Is computer use available outside the US?
The feature is rolling out now in English for users in the United States. Google has not said when it will expand to other markets.
Sources & further reading
How this story was made
Written by Kristian Kostov with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. Cover photography is illustrative and does not depict the specific announcement or product.
Our editorial standards





