Google has updated its Gemini application for macOS to include a new “computer use” capability. This feature enables the AI assistant to execute multi-step tasks by interacting with local files, desktop applications, and web browsers, as explained by the company.
According to Google, users can leverage this new tool to accelerate workflows, automatically fill out forms, create presentations, and list items for sale. To address security and transparency concerns, the app allows users to specify which applications Gemini is permitted to access. Additionally, a dedicated background window provides a live preview of the AI’s actions, enabling users to monitor the process in real time within the chat interface.
The feature is currently being rolled out in English for users located in the United States. The timing of this release comes ahead of similar AI developments expected from Apple.
Is this just basic automation or does it actually understand the context behind the tasks better than macros?
Watching an AI click my apps in real-time is both impressive and slightly terrifying for my privacy.