Google is rolling out Tasks in Gemini at no cost in the United States, India, Canada, Australia, Japan and South Korea, bringing its background AI assistant to eligible users without a paid Google AI subscription. The November 17 expansion lets people hand off multi-step work, from researching travel to organizing email, while a separate set of features for calls, messaging and purchases starts in the U.S.
Tasks was previously known as Gemini Spark. Unlike an ordinary chat reply, a task can continue working in the background, use connected services or browse websites with permission, and return with a result. The tradeoff is that some jobs involve access to accounts or the ability to take actions, so it is worth checking exactly what each permission allows.
Who can use Gemini Tasks for free?
Google’s current Tasks help page lists free access in the six launch countries. Users must be at least 18, sign in with a personal Google account and have Gemini activity saving enabled. Work and school accounts are not currently supported. Tasks is available in the Gemini mobile app, the Mac app and the web app.
Outside those six countries, Google’s guidance lists paid access in supported locations, with exceptions including the United Kingdom, European Economic Area, Switzerland and Nigeria. Availability is rolling out, and some localized help pages still describe the older paid requirement; check the option in your own account before relying on it.
How to start a task
- Open the Gemini app or visit gemini.google.com and sign in to an eligible personal account.
- Choose Task beside the prompt box. If it is not visible, the rollout may not have reached your account.
- Describe the goal and any limits. For example, ask Gemini to compare train options for specific dates and return the prices and links, rather than asking it to book a ticket.
- Review the plan and any requests for access. You can provide more direction while the task is running and inspect the result before acting on it.
Tasks can also run on a schedule. Google’s guidance cautions against scheduling sensitive work: if an automated task runs while you are offline, you may not be there to stop an unintended action. Google lists a maximum of 15 tasks running at once; usage limits also depend on the account and the complexity of the work.
Choose permissions carefully
A task may ask to use a browser, interact with an app or share information. Google’s permissions guide distinguishes Allow once from Always allow: the latter can apply to future task threads as well. For connected Google products, an option to allow an action for one chat lasts for that thread. Start with one-time access when testing an unfamiliar workflow.
When a task uses local Chrome, it may reach sites where you are already signed in. Google says Chrome asks before the first connection and Gemini requests confirmation for each browsing task. You can stop browsing from the Gemini in Chrome side panel, although Gemini may try another permitted route or abandon the task. Local browsing also requires the computer and Chrome to remain awake.
Do not type passwords, payment details or other sensitive information into a task thread. Google advises taking control of the browser and entering those details on the relevant website when needed. Review connected apps and browser permissions in settings, and verify important results independently: permission prompts and other safeguards are not a guarantee against mistakes.
Purchases and calls start in the U.S.
Google’s release notes say its Google Pay integration is rolling out in English to eligible U.S. adults on web, Android and iOS. A user can set an item and a maximum price; Gemini can prepare a checkout and present an itemized summary. Google says explicit approval is required before money moves. A spending cap is useful, but it is still important to check the seller, total price and delivery details at the approval step.
Gemini can also call U.S. businesses to check availability or compare quotes. Google describes a call plan for the user to review before dialing, followed by a summary and transcript. The assistant identifies itself as AI and does not share payment or sensitive account details on the call, according to the release notes.
U.S. users are also receiving ways to send tasks by text or email. Texting requires a verified U.S. number and an eligible carrier. These additions are separate from the wider six-country free rollout, so a free Tasks account does not necessarily include every new capability.
What to check before relying on it
Start with a low-stakes research task and compare the output with the source sites. Check which accounts are connected, whether permission is limited to one action or saved for later, and whether a recurring schedule is still needed. If a task involves purchases, messages, bookings or private data, read the confirmation screen before proceeding.
For the underlying announcement and examples, see Google’s product announcement. TechsCurrent previously covered Spark’s browser access, the earlier version of this workflow.