15 major announcements from the Google I/O 2025 conference

At Google’s annual I/O 2025 conference, the company made a pitch for artificial intelligence. Updates touched almost all of Google’s services, from search and browser to video, email, and smart glasses. Below is a brief overview of the key new features. Many features have already started rolling out, and some will arrive in the coming months. Google also introduced new devices, platforms, and subscriptions, making AI even more accessible and versatile.

AI Mode in Google Search launches in the U.S.

Google is opening up access to AI Mode, a new search mode featuring chatbot Gemini. Users will be able to ask tough questions, search for tickets, build graphs, and even discuss photos and videos in real time.

Project Starline has become a Google Beam

The 3D-effect video chat project has a new name – Google Beam. It’s being integrated into an HP device with a light-field display and six cameras. Duolingo, Salesforce and Deloitte are already preparing to deploy it in their offices.

Imagen and Veo got major updates

The Imagen 4 image generator now works better with text and offers more export formats, including square and widescreen. Visualization quality and detail have also been improved.

The second version of the Veo video model has gained tools for camera control and object removal. The third version of Veo can now synchronize video and sound, making media generation more natural.

New Flow video editor

Flow is an AI-driven video creation app. It uses the capabilities of Gemini, Imagen, and Veo to generate eight-second clips from text descriptions or images.

Users can combine scenes, add transitions, and assemble complete videos. Flow will be a handy tool for marketing, storytelling, and social media content.

Gemini 2.5 Pro with Deep Think mode

The updated version of Gemini has gotten an experimental Deep Think mode for complex problems. It analyzes hypotheses before answering, which is useful when working with math and programming.

The feature is only available to select testers. Google has also released a lighter Gemini 2.5 Flash model and improved it in AI Studio, making it available to a wider audience.

Xreal Project Aura and Android XR glasses

Google and Xreal are working on new smart glasses powered by Android XR. The device will get a wide viewing angle, built-in cameras, microphones, and Gemini AI support.

The device will have a wide viewing angle, built-in cameras, microphones, and Gemini AI support.

Samsung, Gentle Monster, and Warby Parker have also joined the project. This will expand design and feature choices for users interested in mixed reality devices.

Project Astra has become more responsive

The Project Astra AI assistant can now act without a command, reacting to what’s happening through the smartphone’s camera. It can, for example, spot an error in a task and suggest a fix.

The Project Astra AI assistant is now able to act without a command, reacting to what’s happening via the smartphone camera.

Google is making Astra more contextual and adaptive. This approach holds promise for automated assistance for learning, everyday life, and work tasks.

Gemini has arrived in Chrome

As of May 21, users of Google AI Pro and Ultra subscriptions will have access to the Gemini button in Chrome. It will allow you to refine and summarize information on pages, and even control navigation between tabs.

Two tabs are supported at launch, but Google promises expansion within a year. Integrating AI into the browser will make surfing faster and more productive.

A new AI Ultra subscription for $250

AI Ultra gives you access to Google’s most powerful models and increased limits in apps. It includes features from Gemini, Flow, NotebookLM, and the experimental tool Project Mariner.

Mariner can perform up to 10 tasks simultaneously, making AI useful in multitasking scenarios. The subscription is aimed at advanced users and businesses.

Mariner can run up to 10 tasks simultaneously, making AI useful in multi-tasking scenarios.

Search Live: search with a camera

Search Live allows you to communicate with search in real time by showing objects through the camera. This feature integrates Project Astra’s capabilities and makes search visual and conversational.

The user can ask a question with their voice, and the AI will answer by analyzing the image in the lens. This is especially useful for recognizing objects, text, and errors.

An AI can ask a question with a voice, and the AI will respond by analyzing the lens image.

Stitch: generating interfaces with AI

Stitch is Google’s new tool that creates user interfaces based on descriptions, sketches, and references. It speeds up development and makes it easier to visualize concepts.

The tool is already available through Google Labs. It can be used by both professionals and novice designers to quickly generate UI layouts.

Speech Translation in Google Meet

Google Meet has gotten a real-time speech translation feature. English and Spanish are supported at launch, and access is available to AI Pro and Ultra subscribers.

The feature comes in handy for international talks and online events. The translation is displayed on the screen with almost no lag, enhancing the quality of communication.

New smart replies in Gmail

Gmail’s updated “smart replies” use data from email and Google Drive. The system offers response templates that take into account the recipient’s context and communication style.

If the person you’re talking to is a boss, the AI will offer a more formal option. The innovation will be available in July in the web version and mobile apps of Gmail.

AI clothes fitting

Google is testing virtual clothing fitting from a photo. The AI analyzes figure, fabric type and fit to display a realistic image of the item on the user’s body.

AI analyzes the shape, fabric type and fit to display a realistic image of the item on the user’s body.

This feature will be part of AI Mode and will be available in the coming months. It will help avoid mistakes when buying clothes online.

Automatic Password Change in Chrome

If Chrome detects a compromised password, it can generate a new one and replace it on a supported site. The user will receive a confirmation prompt.

The feature launches later this year and is designed to improve security without extra effort. Support will appear gradually on major web platforms.

The 15 top announcements from the Google I/O 2025 conference was first published at ITZine.ru.