Google is rolling out a feature that lets Gemini carry out multistep jobs on a Mac, moving among desktop apps, local files and websites. The work runs in a separate window, with a live preview in the chat. Users choose which apps Gemini can access and can pause the task or approve key actions as it goes.
Artificial Intelligence··Midday
Gemini moves among files, apps and websites
Google started rolling out computer use in its Gemini app for Mac in the United States on 19 November. A person can describe a job in ordinary language and have Gemini carry it out across their files, applications and websites. For work that takes several steps, Gemini can click buttons, navigate menus and enter text in forms.[1], [2]
Work continues in a separate window
Gemini works in a background window. The conversation shows a live view of its progress while the main screen and pointer stay available for whatever the user is doing. Google says users can control which applications the assistant is allowed to open. They can pause work and approve key actions before Gemini carries them out.[1]
Gemini can draw on dashboards to build slides
Google illustrates the feature with a shipping quote. Gemini would take package measurements, weights and addresses from local PDF documents and use them to fill out the request. In another example, it would collect figures and screenshots from dashboards for different regions and assemble a presentation in Google Slides, the company’s tool for making slide decks.[1]