Subscribe
AI
2 min read

Google Brings Computer Use to the Gemini App on Mac

Google on Thursday shipped computer use in the Gemini desktop app for macOS, the feature that lets its AI click, type and browse on a user's behalf across native Mac apps, local files and the web. The company announced the rollout via TechCrunch, and the timing is telling: Google is closing a gap that OpenAI, Anthropic and Meta have been exploiting on the desktop for months. How it works According to TechCrunch, the agent runs in the background. After a user describes a task, Gemini works in

Google Brings Computer Use to the Gemini App on Mac

Google on Thursday shipped computer use in the Gemini desktop app for macOS, the feature that lets its AI click, type and browse on a user's behalf across native Mac apps, local files and the web. The company announced the rollout via TechCrunch, and the timing is telling: Google is closing a gap that OpenAI, Anthropic and Meta have been exploiting on the desktop for months.

How it works

According to TechCrunch, the agent runs in the background. After a user describes a task, Gemini works in a live preview inside the chat window while the screen and cursor stay free for the user's own work. That is a meaningful difference from screen-hijacking agents, which take over the display and block the user until the task ends.

Guardrails follow the now-standard pattern for computer-use agents: actions can be paused, stopped, and require approval before they happen. Google's suggested use cases are the tedious, repeatable kind: filling out forms, batch-processing files, sorting folders by criteria, and assembling slideshows from local files or data pulled from online dashboards.

Availability

The feature launches in English, in the US only, and requires the Gemini Mac app plus a Google AI Plus, Pro or Ultra subscription. Google did not say when it would come to other markets or to Windows.

A late entry, but not an isolated one

Olivetti-style illustration of a background agent working inside a window while the desktop stays free

This is Google's second agentic push on the desktop this week. On November 17 the company expanded access to Gemini Tasks, the in-app feature that signs users into websites, makes purchases and books appointments. Its revamped household agent, CC, was announced in September and has only begun admitting early testers. In July, Google added Gemini Spark to the Mac app, initially for Ultra subscribers only.

The desktop-agent field has consolidated quickly in 2026. Claude Desktop, OpenAI's offerings and Microsoft's Copilot all ship some form of local agentic control, and the standalone AI browser format has largely lost out to agents layered onto existing products. Google's bet with background execution is that users will tolerate a slower, more supervised agent if it never steals the keyboard.

Whether that matters commercially depends on tiers: computer use is now another reason to hold an AI Plus, Pro or Ultra subscription, the same lever Google pulled with Spark. For Windows users, the company's silence on a timeline is the notable part.

Sources

AIGoogleAI Agents