RSS Amplifier

AI Search Decoded · Feb 4, 2026

All Google I/O 2025 Announcements Directly Related to AI Search

0
Sign in to vote or save

Karine Abbou · AI Search Decoded

Karine Abbou June 01, 2025

What: A summary of all Google I/O announcements that directly impact marketing, content creators, and Search in general.

By whom: Me (with AI’s help to keep it not too long - I tend to write too much).

In brief: Over a hundred announcements during this Google I/O and a good quarter concern Search and content.

Other Link to help you: This LinkedIn post where I summarized, while following Google I/O Live, the main changes

👉 I don’t think anyone in the world right now is ready for what Google is bringing to the table with its New Era of Google Search.

During its Google I/O 2025 conference, Google introduced new AI features to make search easier and smarter. Here’s my condensed version of AI Search-related news:

Since the launch of AI Overviews last year, Google Search usage has profoundly changed. Users ask more questions, including more complex, longer, and multimodal questions. AI in Search makes it easier to ask any question and get a useful answer with links to the web.

AI Overviews is one of Search’s most successful launches in the last decade. AI Overviews has been extended to more than 1.5 billion users per month in over 200 countries and territories. Users using AI Overviews are more satisfied with their results and search more often. In the largest markets like the United States and India, AI Overviews drives more than a 10% increase in Google usage for query types that display AI Overviews. This growth accelerates over time. AI Overviews offers the industry’s fastest AI responses. It’s also an important driver of visual search growth in Google Lens.

Google is launching an all-new AI Mode for those who want an end-to-end AI Search experience. According to I/O announcements, this AI Mode began testing earlier this year in Labs and is now rolling out in the United States, with no Labs signup required. This is Google’s most powerful AI search, offering more advanced reasoning, multimodality, the ability to dive deeper with follow-up questions, and useful links to the web. A new tab for AI Mode will appear in Search and in the Google app search bar in the coming weeks. AI Mode is described as a total reinvention of Search.

Under the hood, AI Mode uses the “query fan-out” technique, which breaks down the question into sub-topics and launches a multitude of queries simultaneously on behalf of the user. This allows Search to dive deeper into the web than a traditional search, helping to discover even more of what the web has to offer and find hyper-relevant content.

This AI mode will be the first place where Gemini’s cutting-edge capabilities will be integrated. Starting the week following Google I/O (May 20), a customized version of Gemini 2.5, Google’s smartest model, will be integrated into Search for AI Mode and AI Overviews in the United States. Gemini 2.5 is described as the best-performing multimodal foundation model, and AI responses are of the quality and accuracy expected from Search.

New advanced capabilities are coming to AI Mode. According to I/O announcements, these features will launch first in Labs in the coming months to gather feedback from experienced users. AI Mode is described as a preview of what’s coming, and many features and capabilities will move from AI Mode directly to the main Search experience as Google receives feedback. I did a demo here.

For questions requiring an even deeper answer, Google is integrating advanced search capabilities into AI Mode with Deep Search. Deep Search uses the “query fan-out” technique, but taken to the next level, capable of launching dozens or even hundreds of searches. It can reason about disparate information and create a fully cited expert-level report in minutes, saving hours of research. Deep Search includes links to the web throughout the report.

According to announcements, Google is taking the next step in multimodality by integrating Project Astra’s live capabilities into Search. With Search Live, users can dialogue with Search in real-time about what they see, using their camera. Search becomes a learning partner that can see what the user sees, explaining difficult concepts and offering suggestions, as well as links to different resources. Search Live arrives in Labs this summer. Project Astra is the prototype from which new ideas are developed before being integrated into products like Gemini Live and new experiences in Search.

According to I/O announcements, Google is integrating Project Mariner’s agentic capabilities into AI Mode to help save time on tasks. Project Mariner is a research prototype exploring human-agent interaction, starting with browsers. It’s available to Google AI Ultra subscribers in the United States. Agents can perform up to ten different tasks at once. In Search, this enables managing tasks like ticket purchasing. For example, asking to find tickets for a game, AI Mode launches a “query fan-out” analyzing hundreds of ticket options with real-time prices and inventory, and handles the tedious work of filling out forms. AI Mode presents options matching the user’s criteria, and purchase can be completed on the preferred site, saving time while maintaining control. This will start with event tickets, restaurant reservations, and local appointments. Google is working with companies like Ticketmaster, StubHub, Resy, and Vagaro. Project Mariner capabilities are integrated into the Gemini API and more of its capabilities will be integrated into Google products throughout the year.

The new AI Mode shopping experience combines Gemini model capabilities with Google’s Shopping Graph. The Shopping Graph has over 50 billion product listings, constantly updated. This helps navigate to find inspiration, consider considerations, and refine products. Shopping queries launch a “query fan-out” to suggest options based on specific criteria. Dynamically generated results include a browsable display of personalized images and products. The dynamic panel feature on the right refreshes with relevant products and images.

If the user wants to see how an outfit would look on them, they can virtually try on billions of clothing listings by simply uploading a single image of themselves. This technology is the industry’s most advanced at scale and allows visualizing the appearance of billions of clothing products on a wide variety of people. Virtual try-on in AI Mode arrives in Labs starting today.

When the user has found the perfect item, they can ask the new agentic checkout feature to buy it on their behalf with Google Pay when the price is right. It’s always under the user’s direction and supervision. With one click, Search securely purchases the item with Google Pay. Agentic checkout arrives in the coming months. Shopping features arrive in AI Mode in the United States in the coming months.

For an even more personalized experience, AI Mode will soon offer personalized suggestions based on the user’s past searches. The user can also choose to connect other Google apps, starting with Gmail, to bring more personal context. For example, if the user searches “things to do in Nashville this weekend with friends, we’re big foodies who love music” before a trip, AI Mode can display restaurants with outdoor seating based on past restaurant reservations and searches. The user can get event suggestions during their stay, many of which are close to their accommodation, based on their flight and hotel confirmations. The user will see when AI Mode uses their personal context to help. This is always under their control, and they can choose to connect or disconnect this at any time. Personal Context arrives in AI Mode this summer. Google is working to bring Personal Context to Search, Docs, Gemini, and more.

When the user needs additional help processing numbers or visualizing data, AI Mode will analyze complex datasets and create charts highlighting them, entirely built for their query. For example, if the user wants to compare home-field advantage for two different baseball teams, Search will provide analysis and generate an interactive chart to answer their specific question, using Google’s real-time sports information. This feature will be available for sports and financial queries this summer.

AI Mode begins rolling out for everyone in the United States starting today. All new features presented at I/O will arrive for Labs users for AI Mode in the coming weeks and months. Users can activate the AI Mode experience in Labs to be the first to test them. Some features like virtual try-on are already available in Labs today. Search Live arrives in Labs this summer. No opening planned in France yet. In any case, no announcements on this subject - for now.

Google created a NotebookLLM to allow you to access all announcements made during Google I/O 2025 👉 Notebook LLM Google I/O.

🟣 I’m Karine. I help you adapt your content strategy to AI Search and GEO through a masterclass and a community focused on AI marketing. The next session of my GEO masterclass in english will be in May. Want to join?

Join the waitlist

No posts

Read the original on aisearchdecoded.substack.com

Comments

Nothing yet. Say the first thing.

    Sign in to join the conversation.