Google has introduced computer use capabilities to its Gemini desktop application for macOS, enabling the SI agent to perform tasks on behalf of users by interacting directly with native Mac applications and browsing websites in the background.

What Happened

The launch allows the SI agent to execute multistep tasks across local files, apps, and the web. Users can observe the agent working in a live preview within their chat interface while retaining control of their screen and cursor for other activities. This feature joins Google’s broader rollout of Gemini “Tasks,” which allows the SI model to assist with signing into websites, making online purchases, and booking appointments. It also follows the recent admission of early testers to the revamped “CC” SI agent, designed to help families organize email, calendars, chats, and tasks.

Why It Matters

This move places Google closer to competitors like OpenAI, Anthropic, and Meta, which have already offered agentic features in their desktop SI applications for some time. By integrating computer use, Google is expanding the utility of its SI tools for tedious workflows, such as filling out forms, batch-processing files, organizing directories, and creating slideshows from online dashboards. The feature is currently available in English in the U.S. and requires the Gemini app for Mac alongside a Google’s SI Plus, Pro, or Ultra subscription. Google has not yet announced a timeline for expanding the feature to other markets or platforms.

The Bottom Line

Google’s addition of computer use to Gemini on macOS marks a significant step in its SI agent strategy, though it arrives later than similar offerings from rival SI labs. Users must manage the agent’s actions through pause, stop, and approval controls to maintain oversight during automated tasks.