- Get link
- X
- Other Apps
Featured Post
By Mesum Mukhtar
LearnDailyAI
-
- Get link
- X
- Other Apps
Your AI assistant has learned a new trick: it can now use your computer for you. Google announced computer use in the Gemini app for Mac on 19 November 2026 — Gemini can click, type and fill in forms across your apps, files and websites, in its own background window. And in the same week, a job listing spotted by Bloomberg suggested Apple is building its own agent that does the same thing.
The race is no longer about which AI gives the smartest answers. It is about which AI can actually do the work, safely, while you watch. Here is what Gemini's launch does, what Apple's reported agent would do, and what it all means for you — including the honest news that most of our readers cannot use either one yet.
What Gemini computer use actually does
Until now, Gemini could tell you how to do something on your computer. Computer use lets it do the thing itself: you describe a task, add @Computer use to your prompt, and Gemini works through your Mac apps, your local files and websites on its own.
Google's announcement, written by Michael Friedman, Group Product Manager for the Gemini app, gives four examples of what it can do:
- Fill in shipping forms. Gemini can pull weights, dimensions and addresses from PDFs on your Mac and enter them into shipping quote forms — for multiple boxes, one after another.
- Batch colour-grade photos. Hand it a folder of photos and it can apply the same colour grading across all of them.
- Turn numbers into slides. Give it dashboard metrics and it can build a Google Slides deck from them.
- Turn a sketch into a 3D model. In Google's launch demo, Gemini takes a child's butterfly sketch, models it and rigs it for animation in Blender — doing the clicking itself.
The key detail is where the work happens. Gemini works inside an isolated background window, with a live preview right in your chat. Your main screen and your cursor stay free, so you can keep typing, designing or taking calls while Gemini works in the background.
Watch the official video: a sketch becomes a 3D model
Google's official demo (19 November 2026, 1:03) — watch Gemini take a child's butterfly sketch, model it and rig it in Blender, clicking and typing in a live background window while the user's screen stays free.
How it works, step by step
You do not need to learn a new app. The feature lives inside the Gemini app for Mac, which Google launched as a native macOS app on 15 April 2026. It is a free download from gemini.google/mac. Here is the flow:
- Open the Gemini app for Mac and start a chat, as normal.
- Add @Computer use to your prompt and describe the task in plain words — for example, "Fill in the shipping quotes for these three boxes using the details in the PDFs in my Downloads folder."
- Gemini opens its own background window and starts working through the apps, files and websites the task needs, clicking and typing as it goes.
- You watch the live preview in your chat. You can see every step as it happens.
- You stay in control. You choose which apps Gemini can reach, you can pause it at any time, and it asks you to approve key actions before it takes them.
That last point is the one Google stresses most. Computer use comes with four controls: you decide which apps Gemini can access, you can watch every step live, you can pause it whenever you want, and it asks for your approval before key actions.
Who gets it — and who does not
Here is the catch, and it is a big one for most of our readers. Computer use is rolling out now, but only for a narrow group — and Pakistan and India are not in it.
- Where: English, in the United States only, at launch. Google has given no date for other countries.
- Which Mac: an Apple Silicon Mac running macOS Sequoia 15.0 or later. Intel Macs are left out.
- Which plan: Google AI Plus, Pro or Ultra subscribers, on personal Google accounts, aged 18 and up. Free Gemini users do not get it.
- Which app: the Gemini app for Mac only. Google has said nothing about computer use coming to the Gemini app for Windows.
On price: Google AI Plus costs $4.99 a month and Pro costs $19.99 a month, with Ultra as the top tier above that — check Google's site for the current Ultra price, as it has changed. The rollout is also gradual, so not every eligible account will see the feature on day one.
The Apple surprise: a job listing gives it away
Apple has not announced an agent of its own. But Bloomberg spotted a job listing Apple posted this week for a senior engineering manager to lead a proactive, agentic AI assistant — and the listing describes exactly the kind of agent Gemini just launched. This is reported, not confirmed: the evidence is a job advert, and Apple has said nothing about the project itself.
The listing's pitch is hard to miss: "Imagine an AI agent that can use a computer the way a capable person does. It browses the web, works across apps, runs commands, writes and debugs code, and calls services, and it chooses the right tool for each step. It does all this privately and with the polish people expect from Apple."
The listing says Apple's agents will "plan and carry out multi-step tasks across apps, browsers, command lines, code, and APIs, and they can coordinate specialized sub-agents along the way." In plain words: an agent that moves between a browser, a document, a terminal and a code editor the way a person does.
How far along is it? Bloomberg reports Apple has eyed introducing the service as soon as the end of 2027 — but that the work remains in the early stages. The listing has since been pulled from Apple's careers site. Apple has not confirmed the project, and a job listing is not a launch date — treat the end of 2027 as a target, not a promise.
The timing is not an accident. Apple's new chief executive, John Ternus, has been pushing software chief Craig Federighi to move faster and take a bolder approach to AI — with an agent as the key effort. And the market noticed: shares of Meta and Anthropic slid in late trading after Bloomberg reported on Apple's plans.
Apple is also making it harder for everyone else
Here is the twist. While Apple builds its own agent, it is also making it harder for other companies' agents to take full control of a Mac. Bloomberg reports the upcoming macOS 27.3 will deploy warnings discouraging users from granting full access to third-party agents — and in the beta, turning on ChatGPT's Mac control already takes 14 distinct steps, up from four.
Apple's message is clear. An agent that can run code and act on your behalf needs "strong permission models, sandboxing, and transparency" — the listing's own words. Apple wants to be the company that sets those rules on its own computers, not the company that hands the keys to someone else's agent.
Gemini vs Apple's reported agent
| Gemini computer use | Apple's agent (reported) | |
|---|---|---|
| Status | Launched 19 Nov 2026 | Early stages; not confirmed by Apple |
| What it does | Clicks, types, fills forms across Mac apps, files and websites | Browses the web, works across apps, runs commands, writes and debugs code |
| Who gets it | US only, English, Apple Silicon Mac, macOS Sequoia 15.0+, Google AI Plus/Pro/Ultra, 18+ | Not known yet |
| Possible launch | Now (gradual rollout) | Eyed for end of 2027 (reported target, not a promise) |
| Controls | You choose the apps, watch live, pause, approve key actions | Listing stresses permission models, sandboxing, transparency |
Verdict: Gemini's computer use is real and shipping today — but only to paying subscribers in the US. Apple's agent is only a job listing so far, and its earliest reported target is the end of 2027. The real story is the direction of travel: every big AI company is racing to build agents that use your computer, and Apple — the company that owns the computer — has finally joined the race.
What it cannot do yet
First versions always have limits, and Google is upfront about some of them. Computer use does not work on Windows yet, and Google has not said when — or whether — it will. It does not work on Intel Macs at all.
Google has not said what the usage limits are, or whether the work runs on your Mac or in Google's cloud. And the examples are Google's own demos, not independent tests — nobody outside Google has yet checked how reliably computer use finishes real, messy tasks.
The limited US-only rollout is partly a safety choice: fewer users, more control, faster fixes. That is sensible, and it means the feature should be better by the time it reaches the rest of the world.
What this means for you
If you are in Pakistan, India or most of the world, you cannot use computer use yet. That is worth saying plainly, because most AI news is written as if everyone lives in California. But the feature still matters to you, for three reasons:
1. This is where AI is heading — from chatting to doing
For two years, AI assistants have mostly answered questions. Computer use is part of a bigger shift: agents that carry out multi-step work across your files, apps and the web.
Anthropic's Claude can already join Google Meet calls to take notes. Google's Gemini can now work through your Mac. Microsoft is building agents into Windows and Office.
The skill that will matter in the next few years is not just "how to prompt an AI" but "how to hand an AI a job and check its work." Learning that skill now — even by watching how these tools work — puts you ahead.
2. Repetitive work is the first thing to go
Look at Google's examples again: filling in forms from PDFs, grading a folder of photos, turning numbers into slides. None of these is creative work. All of them are repetitive, multi-step chores that eat hours.
3. The controls matter as much as the capability
An AI that can click and type on your computer can also reach your signed-in apps and see codes on your screen. Google's answer is user control: you choose the apps, you watch live, you pause, you approve key actions.
That is the right model. When computer use does arrive in your country, start with low-risk tasks and only grant access to apps you are comfortable with. Never let an AI do banking, passwords or payments unsupervised, no matter how good the demo looks.
Try it yourself
- Write one task prompt: pick a boring multi-step job you do every week and write it out as a full task — what the inputs are, where they live, and what the finished result should look like. That is the skill computer use rewards.
- Check your Mac: if you have a Mac, find out whether it uses Apple Silicon and which macOS version it runs. That tells you whether you will be eligible when computer use reaches your country.
Quick quiz
1. What do you add to a Gemini prompt to turn on computer use?
Answer: @Computer use — then describe the task in plain words, and Gemini works through your Mac apps, files and websites in its own background window.
2. Who can use Gemini computer use today?
Answer: Google AI Plus, Pro or Ultra subscribers in the United States, on an Apple Silicon Mac running macOS Sequoia 15.0 or later, on personal accounts aged 18 and up. It is not available in Pakistan or India yet, and Google has given no date.
3. Has Apple confirmed it is building its own computer-using agent?
Answer: No. The evidence is a job listing spotted by Bloomberg, which Apple has since pulled. Bloomberg reports Apple has eyed a launch as soon as the end of 2027, but the work is in the early stages.
Your turn
Would you let an AI click around on your computer — or does that cross a line for you? Tell us in the comments, share this post with a friend who lives on their Mac, and follow @learningdailyai on Instagram for the next chapter, explained simply, every day.
All logos and videos in this post are official Google/Gemini media, © Google, used here for editorial coverage.
Sources
- Automate Mac tasks with Gemini computer use — Google (The Keyword), 19 November 2026
- Apple may be developing its own AI agent that runs your Mac — TechCrunch, 19 November 2026
- Report: Apple job listings detail plans for autonomous AI agents — 9to5Mac, 19 November 2026
- Apple wants its own AI agent for the Mac, and a late 2027 launch is already in view — Digital Trends, 19 November 2026
- Claude Meetings: Anthropic's AI joins Google Meet calls — Pandromeda, 13 November 2026



Comments
Post a Comment