Google is adding a background task feature to its Gemini Mac app. Called “Computer Use,” it lets users describe a goal while Gemini clicks, types and navigates apps, menus and forms to carry out the work.
Tasks run in a separate window
Computer Use works across local files, desktop apps and websites. According to Google, the task runs in an isolated background window with a live preview in the chat. That setup is intended to leave the user’s screen and cursor available for other work while Gemini proceeds.
The feature changes how users can delegate computer work: they can explain an outcome rather than handle each step themselves. Gemini then moves through the required interfaces, including forms and menus, as it attempts to complete the request.
Users retain oversight
Google says users decide which apps Gemini can access. They can also watch each step live and pause a task at any time, keeping the process visible while the assistant works in the background.
Key actions require approval beforehand. That gives users a chance to review consequential steps before Gemini takes them, rather than allowing the assistant to operate without any checkpoints. The combination of app access controls, live viewing, pause controls and approvals defines the safeguards described for the feature.
Examples span creative work and paperwork
Google’s examples show the range of tasks Computer Use is designed to handle. It can batch-edit photos and 3D scenes, use information from local PDFs to fill out shipping forms, and build presentations.
These examples involve moving information or work across different parts of a computer. Photo and 3D editing are creative tasks; filling forms from PDFs connects local documents with a website or app; building a presentation involves assembling material into a new deliverable. In each case, Gemini is meant to carry out multiple interface steps from a user’s stated goal.
The feature’s usefulness will depend on what the user wants done and which apps they allow it to reach. Google’s description presents it as a way to hand off routine or multistep desktop work without giving up the ability to observe and intervene.
Availability and the broader push
Computer Use is rolling out in English in the US today, and Google says the Gemini Mac app is available to download. The announcement places desktop task execution within a wider effort to make AI assistants operate software on a user’s behalf.
The source also notes that OpenAI has recently focused on Computer Use, pulling well ahead with its Astra model. Google’s launch therefore arrives as companies work on assistants that can act through computer interfaces, rather than only respond in conversation.
For Mac users, the immediate change is that Gemini can attempt practical tasks in the background while they continue working. App selection, step-by-step visibility, pausing and advance approval give users ways to supervise that work as the feature rolls out.