Google brings Gemini automations to Android, lets AI order rides and food

Google Gemini

• Google adds AI automation to Gemini on some Android phones
• Gemini can now handle food orders and ride bookings
• Automation and scam detection limited to US and Korea at launch

Did our AI summary help?

Google is deepening Gemini’s role inside Android, introducing a new automation feature that lets users hand over everyday tasks such as booking a ride or ordering food to its AI assistant.

Announced on Wednesday, the update brings AI-powered multi-step task handling to the Gemini app on select devices. In beta, the feature allows users to issue a command and have Gemini carry out the process across supported apps. At launch, those apps are limited to food delivery, grocery and rideshare services.

The rollout is narrow to begin with. Automations are available only on the Pixel 10, Pixel 10 Pro and the Samsung Galaxy S26 series. Geographically, the feature is initially restricted to users in the US and Korea.

How Gemini automations work

Google says the goal is simple: let users offload parts of their to-do list to Gemini. In practice, though, the capabilities remain tightly scoped.

Automations must be triggered by an explicit command from the phone’s owner. Once activated, users can monitor progress in real time and cancel the task if something looks off. The actions run inside a secure virtual window on the device, with access limited to specific supported apps rather than the phone’s full data set.

The safeguards reflect a broader reality. AI agents capable of handling multi-step actions can save time, but they can also misinterpret instructions or take unintended actions. By limiting access and requiring direct user approval, Google appears keen to avoid early missteps.

Part of a wider AI automation push

Google’s move aligns with a growing industry push to turn AI assistants into task-completing agents rather than just chat interfaces.

ChatGPT allows users to schedule tasks or deploy an agent to handle actions such as navigating calendars or generating presentations. Anthropic has introduced tools that extend its Claude AI into file and workflow management. Even newer projects like OpenClaw have gained attention for managing everyday digital chores, from emails to flight check-ins.

Gemini’s automation feature signals that Google wants Android to be a frontline platform for this shift, embedding AI agents directly into the operating system experience rather than keeping them confined to a chat window.

Scam detection and Circle to Search updates

Alongside automations, Google is expanding AI-powered scam detection.

On-device Gemini models will now detect scam texts in the US, Canada and the UK on Pixel 10 series devices, with support for the Galaxy S26 series coming soon. Scam detection for phone calls is also arriving on the Galaxy S26 series in the US. Previously, the feature was available on Pixel phones in markets including the US, Australia, Canada, India, Ireland and the UK.

Google is also updating Circle to Search, enabling users to identify all items visible on their phone’s screen at once, rather than focusing on a single object.