Google has rolled out a "Computer Use" feature for the macOS version of its Gemini app, enabling the tool to carry out multi-step tasks across desktop applications, local files, and web pages.
Tasks run inside a separate background window, leaving the user's main screen and mouse cursor unaffected, while a live preview is shown simultaneously in the chat interface so users can pause or stop the process at any time.
Users can choose which applications Gemini is permitted to use, and the model requests approval before performing critical operations.
Among the scenarios Google lists as suitable are batch-editing photos, extracting weight, dimensions, and addresses from local PDFs to fill out shipping quotes, compiling metrics and screenshots from dashboards across different regions into a PowerPoint presentation, and analyzing photos of clothing in a local folder before listing the items on a second-hand platform.
The Computer Use feature arrives ahead of the official release of macOS 27.3, and the new version will alert users to the risks of granting restricted data access, clarifying the scope of what an AI agent can view and do.
The system lets users confine access to specific applications and folders, configure authorization on a per-request basis, and provides a transparent indicator showing when one application is currently controlling another.
Applications already installed and granted Mac permissions will not need to be reauthorized after macOS 27.3 is released, nor will they be forced to display a new explanation screen.
Gemini for Mac can be downloaded from Google's official website, and the Computer Use feature is available to Google AI Plus, Pro, and Ultra subscribers in the United States, with subscription plans starting at $4.99 per month, and Apple silicon Macs required.
Comments