Google announced on November 17, 2026 that Tasks in Gemini, the personal AI agent inside the Gemini app, is rolling out for free in six countries, according to Unite.AI and Android Police. Free access covers users 18 and older in the United States, Canada, Australia, India, Japan and South Korea, and Google says the rollout reaches hundreds of millions of people. (Android Police (opens in a new tab); Unite.AI (opens in a new tab))

The feature was previously called Spark and was limited to paying Pro and Ultra subscribers, Android Police reported. Google introduced Gemini Spark on May 19, 2026 as a cloud based agent running on Gemini 3.5. Tasks now runs on Gemini 4 Hydrogen, the fast model Google released on November 12, 2026 for agent work. (Android Police (opens in a new tab); Unite.AI (opens in a new tab); Data Studios (opens in a new tab))

Users reach the agent by switching the control on the Gemini home screen from Chat to Tasks, on the web or in the mobile app. The agent can keep working in the background and notify the user when results are ready, according to Unite.AI. (Android Police (opens in a new tab); Unite.AI (opens in a new tab))

Calls, logins and payments in the US

A set of advanced features starts in the US only. Gemini can call local businesses, sign in to websites through Google Password Manager without the user sharing passwords, compare products and check out with Google Pay. Android Police reported that the agent can book hotel, restaurant and service appointments, check Google Calendar for a free time and add the booking to the calendar. (Android Police (opens in a new tab); Unite.AI (opens in a new tab))

Google says Gemini tells businesses it is an AI when it calls, and businesses can opt out of these calls. Calling is not available in Indiana, Louisiana, Minnesota, Montana and Nebraska, according to Unite.AI and Data Studios. (Unite.AI (opens in a new tab); Data Studios (opens in a new tab))

Purchases use the Agent Payments Protocol with tokens from Visa and Mastercard on supported cards, and users can set a hard spending cap in advance. Data Studios reported that Gemini prepares the checkout, sends an itemized summary and buys nothing without one tap approval, using a single use virtual card from Google Pay. Receipts go to Google Wallet. American Express support is listed as coming soon. (Unite.AI (opens in a new tab); Data Studios (opens in a new tab))

US users can also assign work by forwarding an email to gemini@google.com or by texting Gemini after verifying a US phone number on an eligible carrier. Data Studios said the payment, text and email features require personal Google accounts and English. Gemini connects natively with Gmail, Calendar, Drive, Docs, Sheets, Slides, YouTube and Maps. (Android Police (opens in a new tab); Unite.AI (opens in a new tab); Data Studios (opens in a new tab))

Security and limits

Google says each user's tasks run in a dedicated, hardware isolated virtual machine on Google Cloud's Agent Compute Engine, with prompt injection filters and Safe Browsing limiting what the agent can open. Gemini asks for confirmation before purchases and other critical steps, and third party connections are off by default, Unite.AI reported. Data Studios noted that Google has not published independent test results for these protections. (Unite.AI (opens in a new tab); Data Studios (opens in a new tab))

Google also cautions that tasks can fail or miss their goal, and that on some websites Gemini hands the screen back to the user for the final click. Data Studios pointed out that Google has not stated usage limits for free users, including for tasks, calls or purchases. (Unite.AI (opens in a new tab); Data Studios (opens in a new tab))

Moving the agent from a paid beta to a free feature turns AI agents from a demo for subscribers into something ordinary phone users will try. The calling and payment tools are the parts that touch real money and real businesses, which is why Google is starting them in one country and wrapping them in caps and confirmations.