Google I/O '25 Keynote Highlights

Introduction

• Speaker: Josh Woodward
• Discussion on updates to Gemini 2.5 Pro, Flash, and the integration with the coding agent Jules.
• Emphasis on innovation and real-world application through Google Labs.
• Goal of the session: Showcase live demos within 59 minutes.

Gemini 2.5 Flash and Stitch

• Product: Gemini 2.5 Flash
  • Fast and cost-effective.
  • Used for prototyping.
• Project Stitch:
  • Experimental project in Google Labs.
  • Transforms design prompts into code and interfaces quickly.
  • Allows design customization (dark mode, color changes).
  • Integrated with tools like Figma for further edits.

AI Studio and Gemini API

• Speaker: Logan Kilpatrick
• AI Studio helps explore the potential of building with Gemini.
• New 2.5 Flash Model Features:
  • Native audio model with 24 language support.
  • URL Context tool for accessing web page data.
• Code Generation:
  • Tightly integrated with SDK for quick application generation.
  • Supports iterative and multi-turn code refinement.

MCP and Google Maps Integration

• Speaker: Paige Bailey
• MCP support in Google Gen AI SDK for agentic apps.
• Example of app integration with Google Maps.

Android and AI Advancements

• Speakers: Diana Wong and Florina Muntenescu
• Building with AI on Android: Enhancing user experience and productivity.
• New App - Androidify:
  • Uses AI for image and text processing.
  • Integrates cloud models via Firebase.

Material Design and App Performance

• Material 3 Expressive:
  • New shape library for UI design.
• Live Updates and R8 Performance:
  • Enhancements in app responsiveness and Play Store ratings.
• Adaptive Layouts for Cross-Device Functionality:
  • Optimizing for foldables, tablets, Chromebooks.

Gemini in Android Studio

• AI tools for enhancing development productivity.
• Jetpack Compose Updates:
  • Autofill, text autosize, visibility tracking.
• Refactoring and Testing Automation:
  • Natural language test scripting.

Web Development with Chrome

• Speakers: Una Kravets and Adi Osmani
• New Chrome features for building robust cross-browser applications.
• CSS Primitives for UI Elements:
  • Simplified carousel and hover card creation.
• AI Integration in Chrome DevTools:
  • Using natural language for debugging.

Firebase Studio Enhancements

• Speaker: David East
• Full-stack app creation with AI assistance.
• Integration with Figma and Builder.io:
  • Transitioning designs into functional code.
• Backend Setup in App Blueprint:
  • Automated backend provisioning and configuration.

Gemma and Open AI Models

• Speaker: Gus Martins
• Introduction of Gemma 3n:
  • Runs on minimal hardware.
  • Supports multimodal understanding.
• MedGemma for Healthcare Applications:
  • Open models for medical image and text analysis.

Future Directions

• SignGemma for Sign Language:
  • Models trained to translate sign language into text.
• DolphinGemma for Marine Communication:
  • Collaboration with researchers to understand dolphin communication.

Conclusion

• Recap of innovations and encouragement to explore new tools.
• Invitation to join local Google Developer Programs.
• Anticipation of future developments in software and AI.