Google is bringing a new ‘Agent Mode’ to its Gemini app, CEO Sundar Pichai announced at the Google I/O 2025 event. Agent Mode in Gemini app is a new experiment coming soon to subscribers that lets one delegate complex planning and tasks to Gemini to get stuff done, the company said. During the I/O 2025 Keynote, Google announced the integration of agentic AI capabilities into Chrome, Search, and Gemini, enabling autonomous task handling like planning and scheduling, a shift from traditional AI's reactive nature.
Agentic AI, as seen in the A2A protocol launched in April 2025, allows interoperability between AI agents, aiming to streamline complex user tasks across platforms, a response to growing demands for scalable multi-agent systems.
How Agent mode in Gemini app will work
With the new Agent Mode in the Gemini app, users can give the AI a task, and it will carry it out on its own. Detailing how the new mode will work, Google explained:
“Say you’re looking for an apartment. Instead of you filtering through real-estate apps daily, Gemini can find listings that fit your criteria, schedule tours and add them to your calendar, and create side-by-side comparisons”.
The tech giant is also improving its AI assistant tool, Project Mariner that helps users by searching the web and handling online tasks. The AI tool can now manage up to 10 tasks at the same time, Pichai said.
He also introduced a new feature called “Teach and Repeat,” which allows users to show the AI how to do a task once, and it will remember how to do similar tasks in the future.
Project Mariner will be rolled out to more users this summer.