Source: blog.google/innovation-and-ai/products/gemini-app/tasks-in-gemini-free-new-features
Google says Gemini can now handle multi-step jobs such as booking appointments, analyzing utility bills and buying groceries, using a user's connected Gmail, Calendar and payment methods. In the examples Josh Woodward gives, Gemini calls several businesses at once, reads past billing statements and adds the booked time to Google Calendar. The update runs on a model Google calls Gemini 4 Hydrogen.
Purchases use the Agent Payments Protocol (AP2), with American Express (coming soon), Mastercard and Visa providing digital tokens on supported cards. Users can set a spending cap, and Gemini is designed to ask for confirmation before clicking Buy. Receipts go to Google Wallet, which is expanding to iOS. Gemini connects with more than 60 apps today, and developers can apply to the Connected Apps Platform.
Each task runs in a hardware-isolated virtual machine on Google Cloud's Agent Compute Engine, dedicated to one account. Tasks start from the Chat-to-Task toggle in the app, by texting Gemini on Android or iOS, or by forwarding emails to gemini@google.com. Calling is not available in Indiana, Louisiana, Minnesota, Montana or Nebraska, and texting requires a US phone number with an eligible carrier. Google says some websites still stop before the final step and return control to the user.
Why it matters if you build agents
Developers building agents get a large-scale example of a sandboxed agent that spends money through payment tokens, enforces spending caps and pauses for confirmation before a purchase. Google also says it is working with major websites and security providers on standards for agent access, so automated flows may still be interrupted by sites that are not yet adapted. Test any workflow that relies on calling or purchasing against the regional and carrier limits Google lists.
Get stories like this every weekday
One short email with the day's agent tooling news.