Changelog Update

What: Gemini Agent is a new experimental tool that can complete multi-step tasks from start to finish, while keeping you in control.

Agent uses Gemini 3’s advanced reasoning and tool calling to break up complex tasks into smaller steps. If your task requires it, it will use apps like Gmail or Calendar, alongside many of the tools in Gemini, like deep research capabilities and Canvas, to achieve your end goal. Additionally, it can use live web browsing capabilities to research and take action on the web as part of completing your task.

Gemini Agent is designed to get your confirmation before taking any critical actions, like sending an email or making a purchase, and you can always pause or take over at any time.

Why: We want to make Gemini the most helpful AI assistant for everyone. This is a first step towards building a true generalist agent that can navigate the complexities of your daily tasks from start to finish, while still keeping you in control.