The Agent Will Make the Call: Google Makes Gemini Tasks Free for Hundreds of Millions of People

Portrait of Maya Wilson

The Library Media — Science, Technology & Health · AI News & Intelligence Desk

The Agent Will Make the Call: Google Makes Gemini Tasks Free for Hundreds of Millions of People

By Maya Wilson

Share

Published November 18, 2026 · 1:36 PM PT

Google has made its Gemini Tasks agent free of charge for hundreds of millions of people across six countries, expanding access to an AI mode designed to carry out multi-step work rather than simply answer questions. Announced on November 17, the rollout covers the United States, Japan, South Korea, Canada, Australia and India, and runs on Google’s Gemini 4 Hydrogen model, the company said.

The change moves Tasks beyond the paid Google AI subscriptions through which it had previously been offered. The feature began as Gemini Spark, demonstrated to selected testers at Google I/O in May, before expanding to Google AI Ultra and Pro subscribers. Users can access it by switching the Gemini app’s home-screen control from Chat to Task, Google said.

The distinction is practical: instead of giving instructions for a task, Gemini can attempt to complete it. Google’s Josh Woodward, who leads the Gemini app, described using the agent to call two car dealerships, compare their appointment arrangements and book a Friday service visit for his family’s minivan in under five minutes. He said Gemini checked his calendar as part of the process.

Woodward also described using the system to investigate a surprise $389 electricity bill. After he authorised access to his utility account and asked Gemini to examine previous statements in his email, the agent identified the family’s heat pump operating during peak-pricing hours as the main cause, according to his account on Google’s blog.

Shopping is another example. Woodward said Gemini compared USB-C chargers using Google’s Shopping Graph, applied a discount code and helped complete a purchase through Google Pay, saving $10 in his example. Google says the payment feature uses its Agent Payments Protocol, or AP2, and works with Mastercard and Visa digital tokens on supported cards.

The most consequential capabilities are initially concentrated in the United States. Google says users there can have Gemini call local businesses, complete Google Pay checkout and sign into websites through Google Password Manager. Texting Gemini and forwarding emails to the service also begin in the US. Texting requires a US number with an eligible mobile carrier; business calling is unavailable in Indiana, Louisiana, Minnesota, Montana and Nebraska.

Google’s account of the system places emphasis on controls around delegated action. For purchases, it says Gemini is designed to seek confirmation or pre-approval before clicking the buy button, while users can establish a hard spending limit. The company says supported payments use digital tokens rather than exposing the user’s actual card number to merchants.

Google also says each task operates inside a hardware-isolated virtual machine dedicated to the user’s account, rather than a shared environment. The company describes additional safeguards including filters intended to detect prompt injection in search results and webpages, alongside Google Safe Browsing protections that limit the links the agent can open. Users can intervene with further instructions while a task is running, Google says.

But the company has also acknowledged that the system is not yet consistently reliable. “While we’ve tested tasks extensively to minimize mistakes, we haven’t eliminated all of them,” Google said. It warned that some tasks may fail to complete or misunderstand what a user intended. Some websites also remain difficult for agents to navigate: Google said Gemini may complete most of a process before handing control back to the user to press the final checkout button.

The free rollout does not mean every new capability is available in all six countries. Google says the advanced features are starting in the United States, but has not given a timetable for their arrival elsewhere.

The expansion follows Google’s November 13 move to bring its Hydrogen model into Search. The latest change puts the task-execution mode directly in the Gemini app and makes it available without a paid AI subscription in the six launch markets.

For users, the shift is from asking an assistant how to complete an errand to authorising it to attempt the work. Google’s own description makes the remaining boundary clear: the agent can make calls, analyse bills and prepare purchases, but users retain approval over critical actions, and some tasks still require human intervention.

Source: Google, “Put Gemini to work for you,” November 17, 2026.

Maya Wilson
AI News & Intelligence Desk, The Library Media

From The Library — the estate of our family.

Follow Us

X Facebook Instagram Threads Mastodon Bluesky

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top