Google announced that its AI helper, earlier called Spark, is now named Gemini Tasks and is free in six countries [1]. Free account holders in the United States, Japan, South Korea, Canada, Australia, and India can now hand off multi-step jobs without paying for a plan [2]. The tool runs on the Gemini 4 Hydrogen model to handle research, scheduling, and routine errands without asking users to babysit every single step [1].
Rollout of the Spark Rebrand
The release changes Google’s consumer rollout by turning an experimental tool into a standard feature. In October, Google integrated the system into the main Gemini app as a full-time assistant after testing it under the Spark codename [2]. People who want to test the Gemini Tasks feature will spot a switch near the text entry area [1]. On mobile phones, that toggle sits above the prompt box, while web users find it below the input field [2]. Flipping the control switches the app from Chat mode to Task mode, letting the system work through complex goals in the background without making the user wait for every single sentence to generate. It frees people from watching a progress bar [1].
Josh Woodward, VP of Product for AI at Google, said the goal is to save people both time and money by taking everyday chores off their hands. Before this rollout, running background tasks needed a paid monthly plan under the AI Pro or AI Ultra tier [1]. Free users in the six supported countries now get full access to the Hydrogen model [2].
The difference between chatting and handing off work is central to how the tool behaves. In standard Chat mode, users submit a prompt, read the response, and type new follow-up questions to move an idea forward step by step [1]. Task mode ends that back-and-forth by letting Gemini plan sub-steps, search web sources, and build finished results on its own [2]. It doesn’t force people to wait on each paragraph [1]. If a request needs ten minutes of research, the agent works quietly until done. Abner Li of 9to5Google reported that Google chose the United States, Japan, South Korea, Canada, Australia, and India because these markets provide a large base of English and localized language speakers [2].

Shopping and Phone Calls in the US
In the United States, Google is pairing Gemini Tasks with consumer services to handle practical real-world errands. The software can place phone calls to local shops using Google Duplex technology to check store stock, confirm hours, and save shoppers from waiting on hold. Jay Bonggolto of Android Central noted that this setup lets users hand off chores that normally demand calling businesses or searching retail apps [1].
Google shared a holiday travel example to show how this shopping routine works in practice. A traveler preparing a family trip back to Oklahoma needed a good USB-C charger. Instead of scrolling through dozens of accessory listings on retail websites, the person asked Gemini to find the best options. The agent checked listings, asked follow-up questions about wattage and color preferences, and found an active discount code that saved $10 at checkout before finishing the purchase through Google Pay. It turned a tedious search into a quick chore. The transaction went through without opening a separate store app [2].
Another utility example showed Gemini analyzing household bills from customer portals and inbox records. A homeowner hit with a surprise $389 electric bill gave the agent permission to check historical statements through Google Password Manager. After reading portal records along with billing PDFs and emails stored in Gmail, the agent found the exact reason behind the surge. The household heat pump turned on around 4 p.m. every day, right in the middle of peak pricing hours. The owners said they never would have caught that timing pattern digging through the files on their own. The agent solved the mystery in minutes [2].
Managing Files with Gemini Tasks
Beyond web browsing tasks, Google expanded how Gemini Tasks interacts with its office tools. The update allows Gemini to draft emails, write documents in Google Docs, edit tables in Google Sheets, and add multiple events into Google Calendar without toggling between tabs. That coordination links text drafting with schedule updates [1].
People do not need to open the Gemini app every time an errand pops up. They can text instructions through Rich Communication Services (RCS) on Android, forward incoming receipts through Gmail, or send messages through Apple iMessage and Google Messages without touching the main interface. It works over text. That flexibility lets a user forward a flight confirmation or hotel reservation to Gemini and ask it to build a full schedule [1]. While Google earlier rolled out Gemini business tools for small firms to help office teams, this update targets ordinary daily routines.

Google also linked the agent to Google Password Manager, solving a common hurdle for web agents. Most web bots stall whenever a website asks for an account login. Gemini can use saved logins stored in Google’s encrypted vault to sign into web portals without showing user passwords to outside scripts [1].
Virtual Machine Defenses
Handling passwords, reading incoming mail, and making purchases introduces serious privacy risks [1]. Google built a sandboxed environment to keep user data safe. Every single web browsing action initiated by Gemini Tasks runs inside an independent, hardware-isolated virtual machine that keeps web code completely separated from the device operating system and local storage. It runs in isolation. By spinning up a temporary virtual instance for each job, the software stops rogue scripts from staying between sessions or reading personal documents stored on the phone [2].
The setup includes defenses against prompt injection attacks, where rogue sites try to hijack an agent by hiding sneaky commands inside web text [2]. The virtual machine pairs with Google Safe Browsing to block dangerous domains in real time [1]. That structured execution builds upon concepts seen in the Gemini decisions API for fast bounded choices, establishing boundaries for background operations.

Financial safety relies on human approval instead of full automation. Even though the agent can find goods and fill shopping carts, it can’t finish an order or charge a card on its own. The software pauses at the final checkout screen and asks the user to review the total price, shipping address, and item quantity before paying [1]. Google says this check keeps buyers in control of their money and prevents surprise charges. Users must approve any payment before money leaves their account [2].
Where Gemini Tasks Still Needs Help
Even with these background skills, Google acknowledges that Gemini Tasks is not ready to run without occasional human help. Some business websites use bot-detection tools that block automated browser sessions from filling forms. When the agent hits a security check or an unsupported checkout page, it stops and asks the user to step in and finish the job manually [1].
Communication clarity is another spot where the system asks for guidance. If a user prompt has missing details or vague terms, Gemini pauses to ask clarifying questions before moving forward. In the Oklahoma charger case, the model asked about desired wattage and colors before scanning the Google Shopping Graph. Asking questions early stops the model from guessing [2].
The rollout gives millions of people free access to background AI agents. Josh Woodward said user feedback from the United States, Japan, South Korea, Canada, Australia, and India will guide upcoming updates to the Hydrogen model [1]. Emma Roth of The Verge reported that this release moves Google closer to building a true personal assistant that can act on behalf of real people [3]. Full autonomy is still incomplete. Even so, making background task automation free across six major countries lets everyday users test tools that earlier required a paid subscription [1].
- ONLINE NEWS Bonggolto, J. (2026, November 18). Google just made its best Gemini agent free — here’s where you can get it. Android Central. [Article Link]
- ONLINE NEWS Li, A. (2026, November 17). You can now assign Gemini ‘tasks’ for free in six countries. 9to5Google. [Article Link]
- ONLINE NEWS Roth, E. (2026, November 18). Google rebrands its Gemini Spark AI agent as Tasks and makes it free in six countries. Techmeme. [Article Link]