
Google Gemini can now handle real-world tasks: calling businesses, making purchases & more
Google has announced a major expansion of its Gemini assistant, rolling out its task-completing Tasks mode free of charge to hundreds of millions of users across the U.S., Japan, Korea, Canada, Australia, and India via the company's new Gemini 4 Hydrogen model. The pitch, framed around saving users time and money, leans on what Google calls Personal Intelligence: because users' emails, calendars, photos, and payment cards already live in the Google ecosystem, Gemini can act on them directly.
Gemini can now make purchases through the Agent Payments Protocol (AP2), securing transactions with digital tokens in partnership with Mastercard, Visa, and soon American Express, while users keep earning their cards' points, cashback, and protections. Buyers can set a hard spending cap upfront, receipts and tracking flow into Google Wallet, and Google says the Universal Commerce Protocol (UCP) will arrive next for faster checkout.
Given an AI that can spend your money and call businesses on your behalf, Google devoted a section to safeguards. Each task runs in a hardware-isolated virtual machine dedicated to a single account within Google Cloud's Agent Compute Engine, regularly tested by the company's security red teams, with prompt-injection filters scanning web content and Safe Browsing limiting which links Gemini can open. The assistant is designed to seek confirmation before critical steps like clicking Buy and to ask clarifying questions when instructions are ambiguous, though users can mark trusted routine actions as “Always allow”. Users can also adjust task settings, disconnect linked apps, or delete chat history at any time. To get started, users toggle the Gemini app from “Chat” to “Task” or, in the U.S., simply forward emails to gemini@google.com.
No comments so far, maybe you want to be first?




