Google gives a radical facelift to the search engine, 25 years later

BarcelonaThe season of major technology conferences for application and service developers begins, setting the tone for what consumers will see in the coming months and years. Before Microsoft's Build (June 2nd) and Apple's WWDC (June 8th), Google kicked things off this week with its I/O, steeped in artificial intelligence (AI) and a total overhaul of the iconic web search engine.

even persistent workspaces that the user can return to.This fundamental change is structural. Until now, Google searches have been an intermediary that led us to the websites containing the information we were looking for. The list of 10 clickable organic results was both the product we users saw and the flow of visits that sustained the economy of a large part of the web. Now, the list of links loses even more weight: Gemini's responses occupy more and more space on the page and, where appropriate, take the form of animations, graphics, and interactive elements generated by AI in real-time, without the user having to leave Google, and even persistent workspaces that the user can return to.Starting this summer, the search engine will also incorporate agents capable of following the topics the user chooses and will be able to make service reservations. The functions of agents and the creation of personalized mini-applications (I made one with my mobile... while watching the presentation!) will arrive first for paying subscribers, but will eventually be free.

However, if the search engine directly answers questions without sending traffic to the websites that contain the original information, the model that has encouraged constant website updates will be fractured. Visits from Google to informational websites have already dropped since the introduction of AI summaries, forcing some publications to close; the new wave of changes will further accentuate this trend. If websites don't receive visits, why should we update them? And if we stop updating them, what content will the AIs feed on?

Gemini with models and agents

Gemini is no longer the name of a model but has become the designation for all of Google's AI services: models, applications, autonomous agents, and the infrastructure that connects them. The company has gone from building products to building an "intelligence system".

The core model presented at I/O is Gemini 3.5 Flash, which you can find in your Gemini applications from Tuesday evening. The Pro version will arrive in the coming weeks. The other notable launch is Gemini Omni, capable of processing (input) and generating (output) text, image, audio, and video, and doing so simultaneously. The initial public version is called Omni Flash and is already available in the Gemini app and for YouTube Shorts users.

Gemini Spark, meanwhile, is an agent that can act autonomously 24 hours a day. Unlike conventional virtual assistants, it doesn't wait for a question: it runs on Google's servers and continues working with your phone locked or computer turned off. In the public demonstration, it was seen managing email, detecting unwanted subscriptions in bank statements, summarizing school communications, and drafting follow-up emails. It works with Google's own services (Gmail, Docs, Slides, and Calendar) and also with third-party tools like Canva, OpenTable, or Instacart. Google AI Ultra subscribers in the US will be able to try Spark starting at the end of the month.

On the other hand, Gmail and Docs incorporate voice functions, Gmail Live and Docs Live, which allow dictating messages and texts, generating drafts from verbally expressed ideas, and interacting with Drive and Calendar content without leaving the application. Google Keep will be able to organize verbal monologues in the form of structured notes.

YouTube is launching Ask YouTube, a conversational search that returns relevant snippets from video content accompanied by a written description. Google Maps incorporates Ask Maps, which allows you to query the map in natural language about routes and surroundings, from finding restaurants by cuisine type to planning routes with specific constraints. Google considers this the most significant renovation of its maps in a decade.

Adoption figures

Google executives seasoned the news with staggering figures. Its AI platform now manages 3.2 trillion tokens per month, seven times more than a year ago. A token is a basic unit of information that generative AI works with: approximately a word or a syllable, depending on the context. The Gemini application has 900 million active monthly users, more than double than a year ago. The AI mode of the web search engine exceeds 1 billion monthly users, and AI-powered summaries (Overviews) reach 2.5 billion users. Google has 13 products with more than 1 billion users, and five of these exceed 3 billion.

Google took advantage of I/O to announce a new category of laptops: Googlebooks. The Chromebook brand was born in 2011 to designate personal computers focused on the web browser and cloud services. 15 years later, Google is replacing it with a platform that unifies ChromeOS and Android.

Current Chromebooks run on ChromeOS, a system derived from the Chrome browser that operates mainly connected to the internet. Android, on the other hand, is the system for mobile phones, with a much broader ecosystem of local applications. The new Googlebooks will be able to natively run both web applications and Android applications, with Gemini as the integration layer. The Windows key is replaced by a G key that directly opens the assistant.

The so-called magic cursor invokes Gemini and positions it over any element on the screen to obtain contextual information. With Cast My Apps, Android phone applications are displayed on the laptop screen without needing to install them. The first Googlebooks manufacturers will be Acer, Asus, Dell, HP, and Lenovo, with processors from Intel, Qualcomm, and MediaTek. The launch is scheduled for autumn, with no confirmed price.

Veteran readers will remember Google Glass glasses; they came out in 2013, but disappeared quickly: they were expensive, offered little, and the camera generated social rejection. Now that Meta has shown that connected glasses can have a market, Google is back. Ray-Ban Meta combines the design of an established brand, integrated audio, a camera, and artificial intelligence assistance, and has had a positive reception. Google's proposal is a collaboration with Samsung on the Android XR extended reality platform, with frames from Gentle Monster and Warby Parker. The first wave will be glasses with microphones, speakers, and a camera to interact with Gemini by voice, but without a screen. Among the various functions it will have are responses about the environment, voice navigation, real-time translation, and notification management. They are expected to go on sale this fall, but we don't know the price yet.