知Yizhi Daily
← Back to this edition

AI

Google brings computer operation capabilities to Gemini for macOS

Edition 2026-11-21 · 08:00 Beijing time

Google announced the launch of a computer use feature in the Gemini desktop app for macOS, enabling Gemini to operate local Mac apps, files, and websites in the background to complete multi-step tasks. Users can view a live preview of operations in chat and pause, stop, or pre-approve actions. The feature is initially available to U.S. English users and paid subscribers.

TechCrunch reported that Google unveiled the feature on Thursday. After a user issues a command, Gemini can carry out tasks across local files, desktop apps, and websites while displaying a live preview in the chat window; the user’s screen and cursor remain available for other work. As with similar AI agents, the system allows users to pause, stop, or approve actions before they occur.

Use cases listed by Google include filling out forms, batch-processing repetitive tasks, organizing folders according to rules, and creating slides based on information from files or online dashboards. The feature requires installing Gemini for Mac and subscribing to Google AI Plus, Pro, or Ultra. Google has not said when it will expand to other markets or platforms.

The report notes that Google is entering this field later than companies such as OpenAI, Anthropic, and Meta, which already offer agentic features in desktop apps. This week Google also expanded the rollout of Gemini “Tasks,” which can help users log into websites, shop online, or make appointments; its “CC” household AI agent, announced in September, only recently began admitting early testers.

Original sources

TechCrunch ↗
Back to this editionOpen interactive reader