Google overhauls Search for the first time in 25 years, with Gemini 3.5 Flash taking on tasks

Google has unveiled the biggest update to its search engine in more than a quarter century. At its core is the new Gemini 3.5 Flash model, which turns Search from an information-retrieval tool into a personal AI assistant that can carry out complex real-world tasks.

Gemini 3.5 Flash: from model to search

Google has built Gemini 3.5 Flash directly into Google Search. As a result, Search can now not only find information but also complete complex, multi-step tasks on its own. Google optimized the model specifically for running AI agents and writing code, which lets it work through results quickly and make decisions without human help.

In practice, agents now see the screen the way a person does: they click links, read text, compare data across multiple sources, and carry out actions. This is no longer just a search engine that returns ten links. It's task delegation: the user describes what they need, and Search goes and does it.

What the new Search can do

• Find flights, automatically filtering by price, departure time, number of layovers, and airline ratings
• Compare prices for a specific product across several stores at once and highlight the best deal
• Fill out forms automatically and pull information from many sites for analysis
• Analyze product reviews, surfacing the key pros, cons, and trends in user opinion
• Coordinate complex bookings, accounting for schedules, availability, and cost

The first test of the feature is running in the US with a limited number of users. Before a wide release, Google is testing the agents for reliability and safety. It is especially important that they work correctly with payment systems and sensitive data.

What changes in the search ecosystem

This update overturns assumptions about what it means to be a search engine in the AI era. Google used to compete only with other traditional search engines such as Yandex, Bing, and DuckDuckGo. Now it competes directly with ChatGPT, Perplexity, and Gemini as standalone AI assistants, which promise to help users rather than just hand them links. Google chose integration over replacement.

For developers and SEO specialists, this raises the bar. Sites now need to work well not only for people but also for the AI agents that will "visit" them on users' behalf. Google used to talk only about Core Web Vitals and mobile-friendliness. Now site owners need to ask whether an AI can find the information it needs on a page on the first try, without 10 clicks and scrolling.

What this means in practice

Google is defending its search monopoly by building the AI revolution itself directly into Search. Rather than getting lost in the wave of new AI startups and alternative search engines, the company is turning its best-known product into a next-generation AI assistant.

For users, this means less time spent on routine online tasks. For Google, it means Search remains the central entry point for every interaction on the internet.