Google Search gets its biggest redesign in 25 years

Google just gave its iconic single-line search box the biggest redesign in over a quarter-century.

At I/O 2026, the company announced that Gemini 3.5 Flash is now the default engine powering AI Mode across the globe. After just one year, AI Mode has crossed one billion monthly active users, with query volume doubling every quarter.

Instead of just returning blue links or text summaries, Google is turning Search into an always-on agentic engine that codes custom software and runs background tasks on demand.

A dynamic canvas and 24/7 background agents

The famous minimalist Search box is now fully dynamic. It expands as you type to help articulate complex thoughts, offering intent-predicting prompts that go well beyond classic autocomplete. You can also drag in text, images, raw files, videos, or entire Chrome tabs as inputs.

Conversations flow naturally out of AI Overviews on both desktop and mobile, letting you drill down into topics without losing context while dynamically refreshing source links.

Then come the Search agents. Operating 24/7 in the background, these autonomous bots continuously monitor the live web—blogs, social feeds, financial boards, and news—based on criteria you set. Looking for an apartment with hyper-specific requirements or waiting on a sneaker drop? The agent tracks changes around the clock and pings you with synthesized updates. This rolls out to Google AI Pro and Ultra subscribers this summer.

Google is also rolling out agentic booking across the U.S. this summer. Search can find niche local services—like a private karaoke room for six with late-night food—and even place phone calls on your behalf to book appointments for home repairs or pet care.

Generative UI and instant mini apps

The most radical shift is what Google calls agentic coding in Search, powered by Gemini 3.5 Flash and Google Antigravity.

Instead of static text, Search can write code on the fly to generate custom interactive UIs. Ask how an automatic watch movement works or dive into astrophysics, and Search will assemble live interactive simulations, sliders, and dynamic diagrams in real time. Generative UI arrives free for everyone this summer.

For longer-term projects like wedding planning or a fitness routine, Search can code persistent dashboards and trackers—essentially disposable mini apps tailored specifically to your data, pulling from live maps, reviews, and weather feeds. This mini-app creator will roll out in the coming months to Pro and Ultra subscribers in the U.S.

Meanwhile, Personal Intelligence—which securely connects Search with your Gmail and Google Photos (with Calendar coming soon)—is expanding for free across 98 languages in nearly 200 countries.

Why it matters

Google is fundamentally rewiring how people interact with the web. Moving from keywords to multimodal prompts is one thing, but converting search queries into functional, self-hosted web apps in milliseconds shifts Search from an information retriever to an execution runtime. For users, that means fewer third-party productivity apps and zero friction when managing ongoing projects.

For competitors like OpenAI, Perplexity, and vertical booking aggregators, this is a massive defensive wall. Google has one billion people already living inside AI Mode; giving them 24/7 autonomous monitoring and generative UI without leaving the homepage effectively cuts off downstream traffic.

We think turning the search box into an on-demand code generator is brilliantly intuitive, but it is not without massive risks. Running live code generation and persistent background agents at Google scale will strain inference infrastructure, and having an AI place phone calls or book services raises thorny liability questions the moment an agent gets an appointment wrong.