Google has officially released a computer-use feature for its Gemini desktop application on macOS. This new functionality allows the AI agent to perform tasks on behalf of users by navigating websites and native Mac applications in the background.

According to Google, the feature is designed to help users "get multi-step tasks done across your local files, desktop apps, and the web." The company suggests the tool could be used to speed up workflows, create presentations, list items for sale, and fill in forms. Other specific tasks the feature could assist in automating include organizing files into folders, processing multiple files, and creating slideshows using information from online dashboards or documents.

Users can monitor the AI's actions through a live preview window in the chat while they continue to use their computers. Google states that users have the ability to halt or pause the agent and must approve its actions before it carries them out. Additionally, users can decide which specific apps the Gemini AI is allowed to access.

The release comes as Google faces intensifying competition from rivals such as OpenAI, Anthropic, and Meta, all of whom have already released similar features. The rollout also precedes new restrictions from Apple regarding "full disk access" for apps, a move prompted by potential risks associated with AI agents having computer use capabilities. A preview of the settings for these permissions appeared with the macOS 27.3 beta this week.

Currently, the Gemini computer-use feature is available in English for users in the United States. To access the tool, users must have the Gemini Mac application and a Google AI Plus, Pro, or Ultra subscription. Google has not yet officially announced details regarding further expansion of the feature.