Gemini Can Now Use Your Mac Apps to Finish Tasks

Google has brought its Gemini AI assistant's computer-use feature to the Mac, letting it click, type and navigate apps on a user's behalf. The company announced the rollout on 19 November 2026, extending an agentic capability that was already available on Windows PCs to Apple's desktop.
The feature allows Gemini to carry out multi-step tasks across local files, desktop applications and websites. A user describes what they want in plain language, and Gemini handles the clicking, typing, menus and forms needed to get it done.
How it works
Gemini works inside an isolated background window, with a live preview of what it is doing shown in the chat. The user's own screen and cursor stay free, so they can keep working on something else while the assistant runs its task.
Users can pause or stop Gemini at any point, and the assistant asks for approval before key actions. It can only use the apps and programs the user has given it access to, with permissions managed in the Gemini settings.
Google's examples show the kind of repetitive work the feature is aimed at. It can list a batch of items for sale on platforms such as eBay, going through photos stored in a local folder, researching prices and staging the listings. It can fill in multi-box shipping forms from details in local PDF documents, or turn a pencil sketch into a three-dimensional model.
Who can use it
The rollout is limited for now. It is available in English in the United States, and it requires the Gemini app for Mac plus a Google AI Plus, Pro or Ultra subscription. The Mac needs Apple silicon and macOS Sequoia 15.0 or later, and the account must be a personal Google account held by someone aged 18 or over.
Google has not said when the feature will reach other countries or Intel-based Macs. The company also warns that Gemini offers on-demand file backups before making changes, but a backup is not always possible, and some actions inside an app cannot be undone.
Google catches up in the agent race
Google arrives slightly late to this particular race. OpenAI, Anthropic and Meta have already added agentic features to their desktop applications, letting their AI systems operate software directly rather than simply answering questions in a chat window.
The Mac launch comes in the same week as a wider rollout of Tasks in Gemini, a feature that lets the assistant work on multi-step jobs in the background, such as signing into websites, making online purchases or booking appointments. Tasks is being offered free in the United States, Japan, Korea, Canada, Australia and India, while computer use on the Mac remains a paid feature.
Google has also recently begun admitting early testers to a revamped personal AI agent, announced in September, designed to help families stay organised across email, calendars, chats and tasks.
The direction of travel is clear across the industry. AI assistants are moving from answering questions to doing jobs: opening files, filling forms and operating the software people use every day. The question for users is how much control they are willing to hand over, and Google's answer, for now, is to keep the agent in a background window, visible in the chat, with approval required before it takes important steps.