I asked a friend’s small business help me compare three internet providers for his shop. I spent 40 minutes clicking between tabs, cross-checking coverage maps, and hunting for phone numbers. Twenty minutes later an AI did the same job and handed me a clean spreadsheet with prices, availability, and direct lines. That gap between what a busy person can look up and what a machine can find is why I keep a personal assistant running in the background — and it is why Google’s free Gemini Tasks update matters more than the headline suggests.

Google called it a “beta” when it first appeared as Gemini Spark. The company has since renamed it, opened it to free users, and turned it from a novelty into a daily driver for hundreds of millions of people. If you have ever wished your AI tool would actually do things instead of just answering questions, this is the moment.
Why the free switch changes the game
For most of this year, Gemini Tasks lived behind a chatbot paywall. You could assign it multi-step projects — booking a reservation, comparing utility bills, planning a trip — but only if you were on a paid plan. That gate is now gone. The feature is rolling out free of charge to hundreds of millions of people in the United States, India, Canada, Australia, Japan, and South Korea, running on Google’s new Gemini 4 Hydrogen model.
The shift matters for a practical reason. Expensive AI is still AI you have to be paid to use. Free AI is the kind you actually let into your normal workflow instead of treating as a curiosity. Google’s own announcement of the free rollout on the Google blog covers the rollout and the new features in one place. And because the feature attaches directly to your Gmail, Calendar, Maps, and Password Manager, it only becomes useful once it sits inside the apps you already trust.
What Gemini Tasks can do for you
Switch the home-screen toggle from Chat to Tasks and the app stops returning answers and starts returning outcomes. Google lists several patterns:
- Scheduling and calendars. Assign it a travel itinerary, a day of appointments, or a recurring weekly briefing. It draws on your email and calendar instead of asking you to re-type dates.
- Shopping and price tracking. Ask it to watch for restocks or price drops on household staples, set a budget cap, and it will stage the checkout for your approval.
- Phone calls to businesses. One of the most tedious parts of errands — calling several shops to compare prices or book an appointment — is now something you can hand off. Gemini dials a few shops at once, identifies itself as an AI assistant, never shares your payment details over the line, and brings you a transcript of its plan before it calls.
- Secure sign-ins. At your request, it can log into a website you are already signed into and do a task there, then hand the screen back to you if the site needs a final human click.
- Payment via Google Pay. Purchases happen through the Agent Payments Protocol, which uses digital tokens on supported cards from American Express, Mastercard, and Visa. Your card numbers are never exposed directly to the shopping agent.
- Text and email instructions. Text Gemini on Android or iOS after a one-time phone verification, or forward emails to [email protected]. It can then sort out forwarded school emails, add dates to your calendar, or track insurance renewals, replying to you in a private thread.
TechCrunch’s coverage of the expansion details the calling and shopping features. Beyond the examples above, Tasks accepts what Google calls Skills — a way to define exactly how an agent behaves for a repeated need, so you stop re-prompting with the same instructions. You can also connect 50-plus third-party apps and services, from OpenTable to Instacart, and act inside them the same way you act inside your Google account. If you are in a niche market, the third-party integration list is the part you want to start with, because the built-in channels cover most everyday needs.
The underlying idea is that an assistant should run in the background, notify you when it is done, and pause whenever it needs a decision.
How to turn it on
One of the reasons adoption has been slow is that the feature hides behind a single toggle. Here is the full path:
- Open the Gemini app on Android or iOS (or visit
gemini.google.comon the web). - Tap the Chat / Task toggle at the top of the prompt box.
- Pick Tasks and describe what you want done, in plain language.
You can also reach the same capability outside the app. Set up Text with Gemini under Settings → Channels, or email it at [email protected]. The first text or call requires a one-time phone verification, which is how Google keeps the service from being abused by bots.
A practical daily workflow I would set up
Tasks feel useful only when they are anchored to a repeatable routine. Here is the setup I would build in the first week, in three steps:
- The daily briefing. Open Gemini in Tasks mode and ask it to compile a morning briefing from your email and calendar. Over a few days you will learn the phrasing that returns exactly what you want, and you can refine the prompt the same way you would refine a shortcut.
- The shopping watch. Tell it to watch a short list of household items you regularly buy, flag price drops, and stage any order under a dollar amount you set. The approval step is the whole point: the AI suggests, you approve.
- The bill review. Forward a month of utility bills to [email protected] with a one-line note asking it to calculate average daily usage. The agent draws on connected apps, pulls the numbers, and returns a short summary instead of a wall of text.
None of these require you to install anything or change your existing Google apps. The first time you run a task, it will ask for permission to read your email, calendar, or location. Those connections turn on by default and can be disabled at any time in the customise page under Gemini’s settings. If you are weighing whether to let an agent touch your calendar, Google’s own controls for Gemini history are a good starting point — my earlier guide to Gemini’s retention settings walks through what each toggle actually does.
Safety and limits
Google is upfront that these agentic features run inside a dedicated, hardware-isolated virtual machine tied to your account. The sandbox is designed to block prompt-injection tricks, and Google Safe Browsing restricts the web links the agent can open. Two habits keep the experience clean:
- Never paste passwords or payment details into a task prompt. Google’s support page is explicit about this, and it is the one rule that applies to every agentic tool, not just Gemini.
- Keep the approval chain on. Purchases and bookings pause for your tap. Turn off the “always approve” routine only for items you genuinely trust.
There are also regional and language limits. The free rollout covers the six countries above, and the rollout is largely in English. If you are outside that list, the feature is not available yet — but that is a rollout question, not a product decision. The underlying pattern, of an AI that executes rather than explains, is what the industry will keep building on.
For anyone who wants to understand the engine behind this shift and where it comes from, my step-by-step guide to building your own AI agent with Python covers the architecture and the local tooling. That path is different from the cloud-powered Tasks mode, but it rests on the same principle: get the machine to act, not just to talk.
For a tighter view of getting daily value from these tools, my practical rundown of five skills for using AI tools every day pairs well with this setup, since it explains how to frame requests so the assistant understands exactly what you mean.
In boxing, the opponent who keeps moving is usually the one who lands the clean shot. A personal assistant does the same thing for your day: it keeps the busywork moving while you stay on your feet. The practical payoff is not that AI is getting smarter overnight; it is that the tedious, repetitive part of your routine now has a worker that runs while you sleep.
Treat the first week as a series of small experiments, keep your approval dialogs on, and you will quickly see which of these tasks deserve to become permanent. I am keeping my Gemini Tasks toggle pinned to the top of my home screen — and that is a recommendation, not a feature.