Google I/O 2025 Focuses On AI And XR With New Features, Updates

Google I/O 2025, held this week, was heavily focused on artificial intelligence (AI) and extended reality (XR), with numerous anounced updates and new features and their integrations across various platforms and services.

A slew of updates to Gemini, were announced. Gemini Live combines input from your phone’s camera, voice commands, and an agent-like ability to search the web, make phone calls, and collate information for you. It’s an extension of the experiments demonstrated last year with Project Astra, where Google’s machine intelligence engine can describe what it sees through a connected camera, remember key facts about your environment, and do hands-free tasks that you ask it to do by chatting with it in a natural way.

Gemini is making its way into Google’s suite of productivity apps like a new feature in Gmail called Personalized Smart Replies. It uses AI to absorb your personal writing style and preferred syntax—culled from all of your notes, emails, docs, and spreadsheets—and use it to generate long replies to emails that match your personal voice.

“With personal smart replies, I can be a better friend,” Alphabet CEO Sundar Pichai said onstage during his keynote, as he explained how he uses the feature to draft emails to questions from friends that he’d otherwise skip because he’s too busy being a CEO.

Following are the highlights of the announcements made during Google I/O this week.

AI Mode for Search: Google announced that AI Mode is now rolling out to all US users. This new feature, which operates in a separate tab, is designed to handle more complex queries and can generate custom charts and graphs. It will also support follow-up questions and use a custom version of Gemini 2.5.

AI Ultra Plan: Google introduced a new, expensive AI Ultra plan priced at $250 per month. This plan includes early access to the latest tools, unlimited use of costly features like Deep Research, 30TB of storage across Google Photos, Drive, and Gmail, and YouTube Premium.

Gemini 2.5: The latest version of Google's AI model, Gemini 2.5, was announced with improved functionality, enhanced security, and better cost efficiency. It includes a new enhanced reasoning mode called Deep Think, which can turn a grid of photos into a 3D sphere and add narration. The model also supports text-to-speech in 24 languages and can better understand conversational flow.

Project Astra: Google provided more details on Project Astra, its vision for a universal AI assistant. A demo showed Astra performing tasks like researching bike parts, calling local shops, and diving into emails to find bike specifications. This assistant aims to handle everyday tasks more efficiently.

Android XR: Updates were announced for Android XR, Google's operating system for extended reality devices. The company teased more details about its prototype smart glasses and confirmed Gemini support.

Veo 3: The latest version of Google's video generation tool, Veo 3, can now generate videos with sound, marking the end of the "silent era" in AI video generation. The company also highlighted improvements in generating photorealistic images with Imagen 4.

Google AI Studio: Google introduced Google AI Studio, a platform for evaluating models and building with the Gemini API. It includes a native code editor integrated with Gemini 2.5 Pro, enabling faster prototyping and the generation of web apps from text, image, or video prompts.

Stitch: A new AI-powered tool called Stitch was announced for generating user interface designs and corresponding frontend code for web applications. Users can iterate on designs conversationally, adjust themes, and export creations to CSS/HTML or Figma.

Jules: The async code agent, Jules, is now in public beta. Jules works directly with GitHub repositories to assist developers in coding tasks.

AI in Android Studio: Gemini in Android Studio was introduced as an AI-powered coding companion to enhance developer productivity. It supports various stages of the development lifecycle.

ML Kit GenAI APIs: New ML Kit GenAI APIs using Gemini Nano were announced for common on-device tasks. An AI sample app called Androidify was showcased, allowing users to create an Android robot of themselves using a selfie.

Deep Search and Search Live: Deep Search is coming to AI Mode, similar to Gemini’s Deep Research. Search Live will use Project Astra capabilities to offer a Gemini Live-like experience.

Personalized Smart Replies: Google will offer personalized smart replies in Gmail, which better match the user's writing style by analyzing emails and Drive documents.