Google I/O '25 Developer Keynote - American Sign Language

Key Concepts

Gemini 2.5 Pro, Gemini 2.5 Flash, Stitch, AI Studio, Live API, URL Context, Function Calling, MCP (Model Component Protocol), Google GenAI SDK, Androidify, Gen AI APIs, Gemini Nano, Material 3 Expressive, R8, Baseline Profiles, Compose Adaptive Layouts library, Android XR, Jetpack Compose, Firebase Studio, Gemma, MedGemma, Unsloth, Google Colab, SignGemma, DolphinGemma, Chrome DevTools, Baseline, scroll-driven animations, interest invoker API, anchor positioning, popover APIs, scroll button pseudo-elements, scroll marker pseudo, target current pseudo-class, AI APIs in Chrome, multimodal AI APIs.

Stitch: Blending Code and Design

Google Labs is experimenting with a new project called Stitch, which aims to blend code and design using Gemini 2.5 Flash. The goal is to enable developers to go from a prompt to an interface to code in a short amount of time.

• Process:
  1. The user provides a prompt describing the desired app (e.g., "make me an app for discovering California").
  2. Stitch generates a design based on the prompt.
  3. The user can iterate on the design by modifying parameters like dark mode, lime green color, and corner radius.
  4. Stitch uses Gemini 2.5 Pro or Flash to remix and change the design based on user input.
  5. The user can grab the markup (code) for the design and copy it to their IDE or Figma for further editing.
• Key Feature: The designs generated by Stitch are not static screenshots but actual designs that can be edited and used in development.
• Call to Action: Developers are encouraged to try Stitch at Labs.Google/Stitch and provide feedback.

Gemini API Updates in AI Studio

Logan Kilpatrick presents updates to the Gemini API within AI Studio, focusing on making it easier to build AI voice agents and other applications.

• Live API Enhancements:
  • New 2.5 Flash native audio model.
  • Proactive audio controls to ignore stray sounds.
  • Native support for 24 languages.
  • Improved session context management.
  • Improvements to function calling and search counting.
  • URL Context: A new tool that allows Gemini models to access and pull context from web pages using a link. Up to 20 links are supported at a time.
    • Example: Passing a link to the Developer docs for function calling and having Gemini provide a summary.
• Code Editor Integration:
  • Gemini 2.5 Pro is integrated into AI Studio's native Code Editor.
  • Optimized with the SDK for generating web applications that use the Gemini API.
  • Example: Generating an AI-powered adventure game using Gemini and Imagen.
  • The code editor is designed to be multi-turn and iterative, allowing users to refine their ideas with prompts.
• MCP (Model Component Protocol) Support:
  • The Google GenAI SDK now natively supports MCP definitions.
  • Facilitates building agentic apps with open-source tools.
  • Example: A new app using Google Maps and MCP.

Building a Keynote Companion with Gemini and MCP

Paige Bailey demonstrates how to remix and compose the Maps app to build a new application: a talking head keynote companion called KC.

• Functionality:
  • KC listens to the keynote, responds, and dynamically updates its UI based on what it hears.
  • KC counts the number of times "Gemini" is mentioned.
  • The Gemini Live API supports a sliding context window for long-running sessions.
  • KC can display Shoreline Amphitheater on a map and find nearby coffee houses with Wi-Fi.
  • KC can provide directions to a selected coffee house.
• Asynchronous Execution:
  • Asynchronous execution is enabled for seamless dialogue within a conversation.
  • The getFunFact function uses Gemini 2.5 Pro and search grounding to display a fun fact.
  • The behavior.NON_BLOCKING call enables asynchronous execution.
• Structured Outputs:
  • Improved structured outputs with function calling, ensuring the model conforms to a specific JSON return format.
• Deployment:
  • The app can be deployed via Cloud Run directly from AI Studio.
  • The app can be run and viewed with VS Code.

Building Excellent Android Apps with AI

Diana Wong and Florina Muntenescu discuss how to build excellent apps powered by AI on Android.

• Androidify:
  • A sample app that uses selfies and image generation to create an Android bot.
  • Uses Gemini models running in the Cloud via Firebase.
  • The app gets a description of the person in the photo using Gemini's multimodal capabilities.
  • It then generates an Android robot based on the image description using the Imagen 3 model.
  • The Androidify sample app is available on GitHub.
• On-Device AI:
  • Gen AI APIs powered by Gemini Nano offer APIs for common tasks like summarize, rewrite, and image description.
• Material 3 Expressive:
  • An update to the Material 3 design system that brings delight and playfulness to apps.
  • Includes a new shape library and expressive APIs.
• Live Updates:
  • A new feature in Android 16 that allows you to show time-sensitive updates for navigation, deliveries, or rideshares.
• Performance Optimization:
  • Enable R8 and Baseline profiles for improved performance.
  • Example: Reddit's app improved so much they got a full star rating increase within two months with R8 and Baseline profiles.
• Adaptive Layouts:
  • API changes in Android 16 to no longer react to orientation, resizability, and aspect-ratio restrictions.
  • New features in the Compose Adaptive Layouts library, like pane expansion.
  • Example: Canva, who invested in large screens, found that cross-screen users are twice as likely to use Canva every single week.
• Cross-Device Compatibility:
  • Android apps are being brought to more devices, including cars and XR.
  • Android XR is the extended reality platform built together with Samsung.
  • A Developer Preview 2 for Android XR SDK is launching, with new material XR components, updated emulator support in Android Studio, and spatial video support for Play Store listings.
• Jetpack Compose:
  • 60% of the top one thousand apps take advantage of the development speed Compose offers.
  • The latest stable release brings features like autofill, text auto-size, and visibility tracking.
  • CameraX and Media3 Compose libraries are being released.
  • The Jetpack Compose navigation library has been rebuilt from the ground up.
• Gemini and Android Studio:
  • Natural language can now be used to perform actions and make assertions with Gemini in Android Studio for end-to-end tests.
  • A new AI agent is coming soon in Android Studio to help with version updates.
• Gemini Code Assist:
  • Subscribing to Gemini Code Assist gives access to Gemini and Android Studio for businesses, designed to meet privacy, security, and management needs.

Building a More Powerful Web with Chrome

Una Kravets and Addy Osmani introduce new features in Chrome that help developers build better UI, debug sites more easily, and create AI features more quickly and cost-effectively.

• CSS-Based Carousels:
  • New CSS primitives in Chrome 135 make building carousels and other types of off-screen UI dramatically easier.
  • Uses the carousel class, scroll button pseudo-elements (::scroll-button), and scroll marker pseudo (::scroll-marker).
  • Example: Pinterest switched to using new CSS APIs for carousels, cutting down around 2,000 lines of JavaScript into just 200 lines of more performant, browser-native CSS, improving product pin load times by 15%.
• Interest Invoker API:
  • An experimental API that, when used with the existing anchor positioning and popover APIs, helps build accessible, complex, layered UI elements without JavaScript.
  • Uses interesttarget to trigger popovers on hover or focus.
• Baseline:
  • Shows feature availability across all major browsers.
  • Available in IDEs, linters, and analytics tools.
  • ESLint can be configured to warn about anything that doesn't match the targeted baseline version for HTML and CSS files.
• AI in Chrome DevTools:
  • AI assistance is baked into the panel, allowing developers to use natural language to ask Gemini questions.
  • AI assistance can create and apply fixes directly from DevTools.
  • The redesigned performance panel highlights layout shift culprits and provides AI-powered suggestions for fixing them.
• Gemini Nano in Chrome:
  • Seven AI APIs are being rolled out across various stages of availability, backed by Gemini Nano and Google Translate.
  • Data never leaves the device, making it suitable for schools, governments, and enterprises with strict compliance and data privacy rules.
  • Example: Deloitte is experimenting with an integration of Chrome's built-in AI APIs into their engineering platform to improve onboarding and navigation.
• Multimodal AI APIs:
  • New multimodal capabilities from Gemini Nano allow users to interact with Gemini using audio and image input.
  • Example: An AI usher that can extract information from a photo of a ticket and highlight the seat section in the app.
  • A hybrid solution with Gemini and Firebase works everywhere, on-device or in the Cloud.

Firebase Studio Updates

David East presents updates to Firebase Studio, a Cloud-based AI Workspace where you can create a fully functional app with a single prompt.

• Figma Integration:
  • You can bring your Figma designs to life in Firebase Studio with help from Builder I/O.
  • The Builder I/O plugin translates all the component code and opens up a window with Firebase Studio.
• AI-Powered Code Generation:
  • Gemini in Firebase Studio can generate code based on prompts.
  • Example: Building a single-product detail page using the existing component system and sample data.
  • Gemini breaks changes into multiple steps, making it easier to review each change.
• Backend Provisioning:
  • Firebase Studio will detect when your app needs a backend and provision it for you if your prompt includes a database or authentication.
  • Firebase Studio will set up the configuration for the backend services and generate code to authenticate users and save data to a database.

Gemma and MedGemma: Open Models for AI

Gus Martins announces Gemma 3n, a model that can run on as little as two gigabytes of RAM, and MedGemma, a collection of open models for multi-modal medical text and image understanding.

• Gemma 3n:
  • Shares the same architecture as Gemini Nano.
  • Engineered for incredible performance.
  • Much faster and leaner on mobile hardware compared to Gemma 3.
  • Added audio understanding, making it truly multimodal.
  • Available in preview on Google AI Studio and with Google AI Edge.
  • Coming to open-source tools like Hugging Face, Ollama, and Unsloth.
• MedGemma:
  • A collection of open models for multi-modal medical text and image understanding.
  • Works across a range of medical, image, and text applications.
• Fine-Tuning Gemma:
  • Unsloth is a library for fine-tuning LLMs like Gemma.
  • The new AI-first Colab is an agentic-first experience that transforms coding into a dynamic conversation.
  • Example: Fine-tuning Gemma to create a personalized translator for an emoji language.
• Gemmaverse:
  • Tens of thousands of model variants, tools, and libraries created by the developer community.
  • Gemma has been downloaded over 150 million times.
  • Close to 70,000 Gemma variants have been created.
• SignGemma:
  • A new family of models trained to translate sign language to spoken language texts.
  • Best at American Sign Language in English.
• DolphinGemma:
  • The world's first large language model for Dolphins.
  • Fine-tuned on data from decades of field research to help scientists better understand patterns in how dolphins communicate.

Conclusion

The presentation highlights Google's commitment to providing developers with powerful tools and APIs to build innovative AI-powered applications across various platforms, including Android, the web, and more. Key takeaways include the advancements in Gemini models, the ease of use of AI Studio and Firebase Studio, and the flexibility of Gemma open models. The emphasis on developer productivity, cross-platform compatibility, and responsible AI development is evident throughout the presentation.