Google announced at its annual I/O developer conference this week that it is replacing the core engine behind its search product and redesigning the search box itself, describing the interface change as the biggest upgrade to that feature in more than two decades.
The conference was held in Mountain View, California. According to a report by CNET, the announcements represent the most explicit statement yet of where Google is taking its flagship product, moving away from the familiar blue-link format it developed over 25 years and toward a conversational AI model.
Gemini 3.5 Flash, a reasoning-focused model built for complex tasks, will become the new default engine powering AI Mode globally. Alongside that change, Google is rolling out what it calls an intelligent search box that expands to accommodate longer queries, accepts photo and PDF uploads, auto-completes nuanced prompts, and can access contextual sources such as open browser tabs to support multi-step research.
Robby Stein, Google's vice president of product for Search, presented the updates as a continuation of a progression from AI Overviews to AI Mode and now to a unified AI search experience. He said a billion people use Google's AI Mode each month.
"This is a very exciting time for Search," Stein told reporters ahead of I/O. "People can ask really anything on their mind and people's curiosity is fairly endless."
AI Overviews, which currently generate automated summaries at the top of search results, will now transition directly into AI Mode when users want to continue a conversation. Rather than receiving a static AI-generated answer, users will be able to follow up within the same interface, turning a search into an ongoing exchange.
Google is also introducing interactive widgets generated by the Gemini system. These tools can simulate physics, visualize concepts, build calculators, or function as persistent mini-applications for tasks such as trip planning, health tracking, or moving logistics. Some of these widgets can draw on connected personal data when users opt in.
The company is building these features around what Stein described as the integration of frontier AI models with Google's live data systems, which span web pages, business listings, product information, images, and financial data. He said connecting the new Gemini model to those systems raises the overall quality of answers users receive.
The shift reflects pressure Google has faced from standalone AI chat products that handle complex, multi-step questions in ways traditional keyword search was not designed to support. The I/O announcements position Google's response as a merger of its existing data infrastructure with the conversational capabilities of its Gemini model family, rather than a separate product built alongside search.
No firm rollout date for all features was announced at the conference.
