Google is rolling out Tasks in Gemini at no charge to eligible adults in six countries, giving more people access to an AI assistant designed to work through multistep requests in the background. The expansion turns a feature previously associated with paid subscriptions into a free option for a much wider audience, although its most consequential new actions have narrower geographic limits.

The rollout began on November 17, 2026, in the United States, Canada, Australia, India, Japan and South Korea, according to Google. Tasks can be used for work such as preparing an itinerary, researching options or organizing information across connected services. In the United States, Google is also introducing ways to ask the assistant to call businesses, start tasks by message or email, and help with shopping.
There is an important distinction for readers outside those markets: a free rollout in six countries is not a worldwide launch. Pakistan is not among the six countries Google named for free access. Availability can change, and a reader's account, age, language and location can all affect what appears in the app.
What changed and who can use it?
Google's November 17 release notes say Tasks in Gemini is rolling out free of charge in the six listed countries using Gemini 4 Hydrogen. The company's productivity page specifies access for users aged 18 and over in those markets, with Pro or Ultra access in select other countries. The rollout is ongoing, so a supported country does not necessarily mean every eligible account has the feature immediately.
The feature sits within the Gemini app rather than being a separate promise of free API access. According to Google, eligible users can switch from Chat to Tasks and describe the work they want done. The assistant works in the background and can notify the user when results are ready. It can also be directed while a task is in progress.
The Verge reported that the feature was previously known as Spark and had been available to Google AI Pro and Ultra subscribers. The newer Tasks name appears directly in the Gemini interface. Names are useful for finding a product, but the important details are the account requirements and what actions are actually enabled.
This change does not establish unlimited use. Google describes the feature as free for eligible users, but readers should consult the current product and account information for any applicable limits. It also does not mean that a paid app, a purchase or a third-party service connected to a task becomes free.
What an AI task can do
A normal chatbot exchange often ends when the assistant returns an answer. Tasks are designed for requests with several steps, such as gathering information, checking a calendar and preparing an itinerary. Google also describes recurring routines, including daily briefings and repeated research. The user gives the instructions; the system performs the work in the background rather than requiring a separate prompt for each step.
That makes the feature attractive for work that is tedious but easy to describe. Someone planning a trip could ask for a shortlist organized around dates and preferences. A household could ask for help collecting events from messages into a calendar, provided the relevant integrations are available and connected. A researcher might request a recurring summary, then inspect the sources and conclusions before acting on them.
These are examples of possible workflows, not guarantees that every connected service supports every action in every region. An itinerary can contain outdated opening hours; an extracted date can be wrong; and an assistant can misunderstand a preference. Review the result and check original sources before booking travel, changing a schedule or sharing information with other people.
Chat, Tasks, Skills and schedules
Google describes Tasks as the part that performs a multistep job. Its productivity page describes Skills as reusable instructions and Schedules as time-based or conditional triggers. For example, a user could save a preference for concise meeting notes and apply it to a recurring briefing, if those controls are available for the account.
The distinction matters because automation has more moving parts than a single answer. A scheduled task may depend on a connected app, an active account and accurate permissions. If the connection changes or a source becomes unavailable, the user may need to revise the task. A recurring instruction should be checked periodically rather than assumed to have worked forever.
The new U.S. features have separate limits
Google's product announcement describes a broader set of actions, but the release notes attach U.S. rollout conditions to several of them. They should not be read as features that automatically arrive in all six free-access countries.
For eligible U.S. adults, Google says Tasks can call businesses to ask questions, compare quotes, check inventory or make bookings. Its release notes say users can review and approve a call plan before the assistant dials, then receive a summary and transcript. Google says the assistant identifies itself as AI and businesses can opt out. An account holder should still verify an important appointment directly if the outcome is unclear.
Google is also rolling out task instructions by text in the United States. Its release notes describe English-language access for adults with a personal Google account, a verified U.S. phone number and an eligible carrier. The same notes describe starting a task by forwarding an email, also initially for eligible U.S. users in English. These are separate from simply opening Gemini and selecting Tasks.
The email option illustrates why context matters. Forwarding a message can include other people's names, contact information or private documents. Before forwarding, consider whether everyone involved expects the material to be shared with a connected AI service. For workplace or sensitive information, follow the relevant organization's policies rather than assuming a personal account is approved.
Shopping, payment and confirmation
Google says an eligible U.S. user can set an item and maximum price, have Gemini prepare a checkout, and review an itemized summary. Its release notes say the purchase requires explicit approval and device confirmation before payment. The Google Pay feature is rolling out in English on the web, Android and iOS for U.S. adults with personal Google accounts.
The spending cap is useful only if it reflects the complete purchase the user intends to approve. Review the merchant, product, delivery address, total price, shipping and any recurring charges before confirming. Google describes the system as designed to pause at critical steps; users should treat the final review as a real decision, not a formality.
Google says it uses the Agent Payments Protocol and supported digital payment tokens for this workflow. It names Mastercard and Visa in its announcement and says American Express is coming later. Support can vary by card, merchant and rollout. Having a Google Pay account does not by itself prove that a particular transaction is eligible.
The company's examples also describe adding groceries to an Instacart order and handing the user to the service to complete it. That example should not be generalized into a promise that every shopping service can be ordered from automatically. Connected apps, their availability and their own terms determine what can be done.
How Google says it protects a task
Google says each task runs using a hardware-isolated virtual machine dedicated to the user's account, with a sandbox and additional defenses such as filters for malicious instructions and Google Safe Browsing. These are Google's descriptions of its design. They do not prove that every unsafe instruction will be caught or that an AI-generated result will always be correct.
A particular concern for any browsing agent is prompt injection: a page, document or message may contain instructions that try to redirect the assistant away from what the user requested. A sandbox can help isolate the execution environment, but the user still needs to review what the assistant is about to send, buy or change. Restricting connected apps to those needed for a task also reduces unnecessary exposure.
Google says users can adjust task settings, disconnect Google or third-party apps and delete chat history. Its productivity page says connected-app integrations are turned off by default and can be enabled in settings. Before connecting email or cloud storage, read the permission request and current privacy information; decide whether the benefit is worth granting access for that particular task.
The company acknowledges that tasks can fail or miss the mark, and says some websites may require the user to take over for the final step. That is a practical limit, not evidence that a particular website will always block the assistant. When a task involves money, health, legal documents or another high-impact decision, check the underlying information and use appropriate professional judgment.
How to try Tasks without overcomplicating it
If you are an eligible adult in a supported market and see the feature, open the Gemini app or web interface and look for the Chat-to-Tasks control described by Google. Start with a request that does not require personal records or an external action. A useful first test might be a list of public resources for a weekend activity, including links and opening hours to verify.
Give the assistant a clear outcome, constraints and a stopping point. Instead of asking it to manage an entire trip, specify the destination, dates, budget range and whether it should merely compare options. Ask it to present choices before making any reservation. That makes its output easier to inspect and keeps the important decisions visible.
- Check that your country, age and account are eligible and that Tasks appears in Gemini.
- Start with a low-stakes request based on public information.
- State the result you want, your constraints and which actions need approval.
- Connect only the apps needed for the request and review their permissions.
- Check sources, dates and totals before acting on the result.
For a recurring task, add a review date and decide what success looks like. If a briefing is meant to arrive each morning, confirm it actually arrives and contains useful links. If it begins making claims without supporting sources or repeatedly misses the requested time, update or stop it. Automation is helpful when it reliably saves effort, not when it quietly creates another inbox to check.
What this means for users in Pakistan and elsewhere
Google's announced free rollout names six countries and does not include Pakistan. That is a statement about the listed launch markets, not a permanent prediction. The company also describes Pro and Ultra access in select other countries; readers should check Google's current availability page and their own account before purchasing a plan specifically for this feature.
Do not assume that changing a profile setting or using another country's details creates legitimate access. Availability and eligibility are determined by Google's service rules and can depend on several factors. Readers outside the listed markets can still follow the rollout and evaluate the regular Gemini features that their accounts actually offer.
The expansion is notable because it brings an agent-style workflow to a free consumer tier, rather than requiring every reader to build an API integration. It also shows the limits of a broad product announcement: a free task mode, U.S.-only calling, text-based access and payment features are related, but they do not share identical eligibility requirements.
Questions to keep in mind as AI agents become common
The next useful measure is not simply how many actions an assistant can attempt. It is how often it completes the right task, how clearly it explains uncertainty and how easily a person can stop or correct it. For a recurring briefing, the test may be whether links and dates are accurate. For a phone call, it may be whether the recorded outcome matches what the business agreed to.
Google's launch provides a wider group of eligible users with a way to test those questions for themselves. Start small, keep permissions narrow and review important results. The convenience can be real, but a human still needs to decide what should be done with the answer.
Sources and reporting context
This article uses Google's November 17 product announcement, release notes and productivity page, together with independent reporting by The Verge. Features and eligibility are described as announced at publication time; rollout, terms and limits can change.


