Google I/O 2025 delivered a host of groundbreaking updates, making it clear that AI and cross-platform development are at the heart of Google’s vision for the future. From smarter APIs to new tools for building on Android, web, and Firebase, the keynote showcased how developers can leverage the latest advances from Google’s platforms and models, especially the powerful Gemini family.
Building with Gemini: AI for Every Developer
Google AI Studio now features Gemini 2.5 Pro, making it even easier to prototype and build with the Gemini API. Developers can generate web apps from text, images, or video prompts and explore ready-made starter apps.
New agentic capabilities, like URL Context, allow Gemini models to fetch and reason over live web content.
The Gemini SDKs now support Model Context Protocol (MCP) definitions, streamlining integration with open-source tools.
Gemini 2.5 Flash Native Audio introduces advanced voice capabilities in 24 languages, offering more natural conversational AI experiences.
The newly launched Stitch tool uses AI to generate user interface designs and front-end code, which can be exported to CSS/HTML or Figma.
Meanwhile, Jules, an asynchronous code agent now in public beta, integrates with GitHub to automate common coding tasks from upgrades and tests to coordinated edits across repositories.
Android: Smarter, More Adaptive Apps
Generative AI is transforming Android development. New ML Kit GenAI APIs powered by Gemini Nano enable intelligent, on-device features. Google demonstrated this with Androidify, an app that creates personalized Android avatars from selfies.
Support for adaptive app experiences now extends to cars and XR devices, while Material 3 Expressive helps developers design standout apps across over 500 million devices.
Gemini in Android Studio enhances productivity with AI-powered assistance throughout the development lifecycle. The keynote previewed agentic features like Journeys for end-to-end test automation and a Version Upgrade Agent to manage dependencies, helping developers build better, faster.
Web: Building Powerful, AI-Enhanced Experiences
Google is simplifying web development with new features in Chrome and popular web tools. Building interactive carousels now requires just a few lines of CSS and HTML, thanks to new primitives in Chrome 135.
The experimental Interest Invoker API allows developers to create dynamic UI elements, like tooltips and hover cards, without JavaScript.
Baseline feature availability is now integrated into VS Code, ESLint, and other tools, allowing developers to easily track compatibility across browsers.
AI in Chrome DevTools brings Gemini-powered suggestions and contextual debugging directly to the workflow.
While new Gemini Nano APIs like Summarizer, Language Detector, and Translator are rolling out to Chrome, reducing latency and enhancing privacy.
Firebase: Rapid Prototyping with AI
Firebase Studio now lets developers turn Figma designs into full-stack apps in minutes using Gemini 2.5. The platform can recommend and provision backend services like Firebase Auth and Cloud Firestore, streamlining the deployment process. Firebase AI Logic brings generative AI models into both client and server-side applications, with deeper integrations and improved observability.
Open Models: Flexibility and Innovation
Google’s open model family, Gemma, is expanding. Gemma 3n runs on as little as 2GB RAM, enabling advanced AI on portable devices.
MedGemma offers multimodal medical comprehension, supporting developers building healthcare solutions.
The new AI-first Colab experience, powered by Gemini 2.5 Flash, simplifies tasks like model fine-tuning and UI development.
Google also previewed SignGemma, a model for translating sign language, and DolphinGemma, the first large language model tailored to dolphin communication research.
Google I/O 2025: Key Takeaways for Developers in the Age of Gemini AI