What A Time To Be Alive
Choose language: English

Gemini on the Mac can now operate apps and websites by itself

Google is letting Gemini on the Mac carry out tasks in programs, files and websites by itself. The assistant works in a separate window in the background and asks for approval before important steps. The feature is coming first to the United States, in English.

Title card with the words Gemini and Computer Use against a dark starry sky, a planet with a blue glowing rim in the bottom right and the Google logo in the top right.

, Google announced a new feature in the Gemini app for the Mac. Gemini can now carry out a task that takes many steps by itself, in programs on the computer, in files and on websites. The user describes what needs to be done, and Gemini clicks, types, chooses from menus and fills in forms. Google calls this computer use.

Gemini does this in a separate window in the background. The chat window shows live what the assistant is doing, while the user's own screen and mouse pointer stay free for other work. The user chooses which apps Gemini can access, can follow every step and can pause at any moment. The user has to approve important actions first.

What it can do

Google gives four examples. In a design program, Gemini can adjust the colours of a whole series of photos in one go. It can copy weights, dimensions and addresses from PDFs on the computer into a form to request a shipping price. It can collect figures and screenshots from overview screens showing business figures and turn them into a presentation in Google Slides, Google's presentation program. And it can look at photos of clothes in a folder, look up prices and prepare listings on a second-hand marketplace.

Who can use it

The feature is now rolling out in the United States, in English. It requires a Mac with Apple's own chip and macOS Sequoia 15 or newer, a personal Google account of someone aged 18 or over, and a subscription to Google AI Plus, Pro or Ultra. The Gemini app for the Mac itself can also be downloaded outside the United States.

Google had already released its own Gemini app for Mac and Windows, in which the assistant works in its own window. Since the end of June 2026, the Mac has also had Gemini Spark, the personal assistant for subscribers to Google AI Ultra. Spark works with folders that the user chooses, for example to sort PDFs or to draw up a budget from invoices on the computer. With computer use, Google goes one step further: Gemini now also operates the user's own programs.

What does this mean for the road to the singularity?

Assistants that operate a computer by themselves have been around for a while, but often only for developers, or they take over the screen so that the user cannot do anything else. What is new is that Google is putting it in an ordinary app, also for Plus subscribers, and that it runs in the background while the user keeps working. That makes it something people can leave running next to their own work, like a colleague who takes over the dull clicking.

If this approach becomes reliable, a program no longer needs a special connection to work with AI: anything a person can operate on a screen, the assistant can then operate too. For software makers, it then becomes important that their program works well for such assistants, or users will choose another one. For ordinary people, hours of retyping, sorting and filling in forms could disappear, even if they never learned to program. A mistake can have real consequences, though, such as a wrong address on a parcel. That is why choosing which apps the assistant may use, watching live and approving actions in advance remain important for now.

What is the technological singularity? Read the explanation →

Sources

  1. Google
  2. 9to5Google
  3. Google Gemini voor Mac