Google announced computer use for the Gemini app on Mac on November 19, 2026, giving its assistant the ability to navigate menus, type, fill forms and click through desktop apps and websites. Google says Gemini does this inside an isolated background window, leaving the user’s main screen and cursor free while a live preview appears in chat. (Google announcement)
The feature is rolling out in English in the United States to adults with personal Google accounts subscribed to Google AI Plus, Pro or Ultra. It requires an Apple-silicon Mac running macOS Sequoia 15.0 or newer. Meeting those requirements makes an account eligible, but rollout language means access may not appear immediately.
The consequential change is that Gemini can act across the software and files involved in a task. Its usefulness will depend on how well those actions work, how much supervision they require and whether the background window genuinely saves users from juggling another workflow.
Who Can Use Gemini Computer Use Now?
Google’s launch requirements are specific enough that having Gemini installed, or paying for a Google AI plan, is only part of the eligibility check.
| Requirement | Availability at launch |
|---|---|
| Application | Gemini app for Mac |
| Hardware | Apple-silicon Mac |
| Operating system | macOS Sequoia 15.0 or newer |
| Account | Personal Google account |
| Age | 18 or older |
| Subscription | Google AI Plus, Pro or Ultra |
| Language | English |
| Market | United States |
| Access timing | Rolling out, so availability can vary by account |
These restrictions come from Google’s announcement and its eligibility footnote. Intel Macs and free accounts fall outside the stated launch requirements. Work and school accounts are also outside the announced personal-account offering.
For someone checking access, I’d start with that table before troubleshooting the application. An unsupported account type or Intel processor is an eligibility issue; reinstalling the app would not change it.
Google directs eligible users to download the Mac app through its announcement. The post does not provide a detailed setup walkthrough, so there is no basis here for promising a particular menu location, permission-dialog sequence or activation switch.
How the Background Window Changes the Workflow
Gemini accepts a description of a task, then uses clicking, typing, menus and forms to work toward it. Google says the feature can span local files, native desktop applications and websites, putting several stages of a workflow within one assignment.
A shipping task illustrates the mechanism. The information might begin in local PDFs, while the destination is a website with separate fields for package weights, dimensions and addresses. Google lists extracting those details and filling multi-box shipping quotes as one of its suggested uses. (Google’s workflow examples)
The isolated window matters because interface automation can otherwise compete with the person using the computer. If an assistant needs the same cursor and foreground window, the user has to wait or risk interfering. Google says Gemini’s main-screen and cursor separation lets someone continue typing, designing or taking calls while the task proceeds.
The live preview provides visibility into that separate workspace. Users can watch Gemini’s progress from the conversation without surrendering their main working area.
My view is that this separation is the most practical part of the announcement. A repetitive task that takes several minutes can still be worth delegating if it leaves the computer usable. That benefit comes from keeping the work out of the way, even before considering any claim about how quickly Gemini completes it.
Background operation still involves supervision. The preview, pause control and approval checkpoints give the user ways to follow and interrupt the work.
App Permissions and Approvals Set the Boundaries
Google describes three controls over what Gemini can do:
- App access: Users choose which applications Gemini may access.
- Pause: Users can pause the task at any time.
- Action approvals: Gemini asks users to approve key actions before they happen.
The wording on approvals deserves attention. Google’s main announcement says users can “approve key actions before they happen.” That is narrower than a promise that every click, keystroke or intermediate step requires confirmation. The announcement does not enumerate which actions trigger approval. (Google’s control description)
That distinction affects how to use the feature. Permission to operate an app, visibility into its actions and approval of consequential steps are separate controls. An app permission determines where Gemini may work; the preview helps users observe it; an approval creates a checkpoint before a particular action.
For a cross-app workflow, the permissions are worth considering as a set. A task involving a PDF, a browser and a presentation application needs access across several contexts. I’d grant access according to the task at hand and keep the initial assignment narrow enough that its output is easy to check.
An Isolated Window Is Not a Complete Security Specification
Google describes an isolated background window, but its launch post does not explain the technical isolation boundary in enough detail to assess it as a security architecture. It also does not specify which processing happens locally, which data leaves the Mac or how every sensitive action is classified.
Independent security and reliability evaluations of this shipped capability were not available in the reviewed sources. The documented controls are therefore useful product details, while broader conclusions about security or task success would need more evidence.
The practical distinction is straightforward: keeping Gemini’s cursor away from yours solves an interaction problem. It does not tell you whether Gemini copied a number correctly, interpreted a document accurately or chose the right destination for a file.
Google’s Examples Show Where Review Still Matters
Google proposes several workflows that move information between files, applications and websites. They are vendor-supplied examples, not independently measured demonstrations of success.
Frequently Asked Questions
3 questions
1Does Gemini Computer Use Work on Intel Macs?
Intel Macs are outside the announced requirements. Google specifies an Apple-silicon Mac running macOS Sequoia 15.0 or newer.