Age of AI Agents: Google Announces Agent Mode for Gemini, Chrome, and Search at I/O 2025

A day after Microsoft underscored the need for agentic AI for the next phase of generative artificial intelligence, Google announced its version of AI agents. A new Agent Mode will soon come to Google's leading services — Gemini, Chrome, and Search — allowing automation that involves looking up the internet for a particular query with required filters on a preferred website and performing tasks like booking a ticket on your behalf.

During the I/O keynote, Google CEO Sundar Pichai demonstrated how the Agent Mode will work. Pichai asked Agent Mode to look for an apartment in the nearby area. The Gemini foundational model-powered AI agent opened Zillow — a house-searching service in the US, browsed the website to set filters, and then used MCP to book a showing of the house for him — all without any human intervention.

“This is a new and emerging era,” he said. “The best way to bring research into reality is to make it really useful.”

Agentic AI has been a holy grail for the industry, with every leading company investing heavily to automate how we interact with our computers. While accessing still requires us to perform manual tasks, such as typing or speaking, agentic AI can learn about a user and perform basic functions with a simple command. A bot uses AI to crawl websites, documents, and files to take action on your behalf across any device, making things like booking a ticket or adding items to a cart entirely automatic.

Pichai said Agent Mode is currently experimental, meaning it may make mistakes, so you know when to rely on it. Since the Agent Mode will be available in Gemini, Chrome, and Search, it will allow a wide range of functions. Pichai did not elaborate exactly on how many chores Agent Mode could do. The new mode will be available only to subscribers, Pichai said.

Read more: Vivo V50e Review: A Feature-Loaded Phone That Should’ve Cost Less