For developers building AI agents in India ๐ฎ๐ณ
ARD is a new open specification that makes it easier for agents to securely discover and trust tools, MCP servers, and other agents across the web, helping build a more interoperable agent ecosystem
Agents are part of a massive, interconnected ecosystem. But how do they find and trust each other across different platforms?
Today, weโre proud to announce the Agentic Resource Discovery (ARD), an open specification alongside industry partners (including Cisco, Databricks, GitHub, GoDaddy, Hugging Face, Microsoft, NVIDIA, Salesforce, ServiceNow, and Snowflake). ARD gives any agent a secure, decentralized way to discover and verify capabilities (like tools, skills, MCP servers, and other agents) anywhere on the web.
Read the full announcement and get started: https://t.co/1I9WsT0tyH
Android 17 is here.
This release is built around the shift from OS to intelligence system with new tools, libraries, and agent skills to help your apps meet the moment.
If you build on Android, this is the one to read.
Dive in โ https://t.co/fy8sppSLV9
Android 17 has arrived! ๐
We're bringing together next generation tools, libraries, and agent skills to help your apps meet this moment.
Dive deeper into Android 17 โ https://t.co/dGXQLOqAJs
Multimodal inputs are live in the Prompt API. Pass image assets locally and get structured JSON schemas back for accessibility, tagging, and layout automation without hitting an external server.
See what you can build: ๐
https://t.co/adfQW4bXwE
Markdown image syntax is notoriously hard. With multimodal input in the Prompt API, your browser can ""see"" images locally to return a structured JSON object with alt text and captions. No server latency, just private, reliable data to build your <figure> tags โ https://t.co/T8Bq3clLRZ
Speed matters when you're building with AI โก
Meet DiffusionGemma, an experimental open model that explores text diffusion, a new approach to text generation that's optimized for speed.
โก๏ธ Up to 4x faster token output on dedicated GPUs
โก๏ธ Only 3.8B active parameters during inference
โก๏ธ Great for code infilling, inline editing, and self-correcting workflows
โก๏ธ Apache 2.0 licensed and open for experimentation
If you're building AI agents, coding assistants, or developer tools, DiffusionGemma is worth exploring to prototype faster and unlock new UX patterns.
DiffusionGemma, our experimental open model released under an Apache 2.0 license, explores text diffusion, an exceptionally fast approach to text generation.
Hereโs how DiffusionGemma accelerates development:
+ Faster token output: By shifting the bottleneck from memory bandwidth to raw compute, the model generates up to 4x faster token output on dedicated GPUs
+ Accessible hardware footprint: Activates just 3.8B parameters during inference, fitting comfortably within 24GB-VRAM high-end consumer GPUs when quantized
+ Novel workflows: Parallel token generation enables self-correction, making it ideal for code infilling, in-line editing, and non-linear structures
DiffusionGemma prioritizes speed over raw quality and accelerates best on compute-bound hardware (like @NVIDIAAI GPUs). Standard @GoogleGemma 4 remains recommended for production quality and memory-bound devices.
Data work, now inside your IDE โ๏ธ
Data Agent Kit brings data engineering + science tools into VS Code and agentic dev tools like Claude Code, Codex, and CLI workflows.
ICYMI: We launched Data Agent Kit, a unified, open-source collection of data engineering and data science skills, tools, and plugins that integrate into the environments practitioners already useโVS Code, Claude Code, Codex, Antigravity CLI, and more โ https://t.co/kqGqvrHImc
Android is evolving from an operating system to an intelligence system ๐ค๐ฑ
To help developers build the next generation of AI-powered experiences, Google has introduced new tools for hybrid AI, agent development, and dynamic UI generation on Android:
๐ฅ Firebase AI Logic Hybrid Inference: intelligently route between on-device and cloud models
๐ฅ A2UI Jetpack Compose Renderer: enable agents to generate and render native Android UI
๐ฅ ADK for Android: build multi-agent workflows with orchestration, context handling, and session management
Whether you're building AI assistants, personalized experiences, or agentic apps, these new capabilities make it easier to bring powerful AI experiences directly into Android apps.
To help you build more advanced AI features and explore building in-app agents, weโve released new APIs, framework and guidance:
๐ฅ Firebase AI Logic Hybrid Inference
๐ฅ A2UI Jetpack Compose Renderer
๐ฅ ADK for Android
Read more about these updates โ
https://t.co/9Ida2wqaaD
AI is much bigger than just code completion๐
Discover how to integrate AI across your entire product lifecycle, from architecture to deployment. Move past simple generation and start building smarter end-to-end workflows๐ ๏ธ
Catch the full session details below! ๐
The journey with AI in development is just getting started. Beyond code generation lies an incredible opportunity to transform how we build.
Here is how it unlocks that potential:
1๏ธโฃ Integrate AI across the entire software & product lifecycle.
2๏ธโฃ Let our traditional processes evolve and grow into smarter workflows.
โจ Dive into the full, inspiring session here: https://t.co/zAJUXYmh2v
From prompt to an Android App in minutes๐คฏ
Watch @patloeber build and deploy an AI-powered Android app directly onto his phone using Google AI Studio.
๐ ๏ธ Prompt your UI & features
๐ฑ Preview in the browser
๐ Deploy via USB
What are you planning to build with Google AI Studio? ๐
Full-stack Dart just leveled up. ๐
Build smart, context-aware apps from a single codebase using the new GenUI SDK for Flutter and Genkit Dart.
๐ Real-time adaptive layout interfaces
๐ป Shared frontend/backend logic
Watch here: https://t.co/BcPBw4OgXU
Eliminate context switching and accelerate development with Dart for Firebase Functions ๐ฏ๐ฅ
Learn how to build a complete serverless application, using full-stack @dart_lang across client and cloud environments โ https://t.co/wWdjINKFvP
Fine-tuning custom AI models is becoming more accessible ๐
This community project shows how to fine-tune Gemma 4 12B locally, even on hardware with just 8GB VRAM. Great inspiration for developers building domain-specific AI.
Want to teach Gemma to master chess?
Check out this awesome community project showing how to fine-tune Gemma 4 12B on your own data, 100% locally!
Running text, images, and audio on just 8GB VRAM makes custom models more accessible than ever.
Ready to bring your XR ideas to life? ๐ ๏ธโจ
Weโre expanding the Android XR ecosystem with new laptop-to-headset prototyping tools and massive backing for developers. Start building today.
The Android XR ecosystem is expanding, and weโre rolling out new resources to help you scale your experiences:
๐ป New tools to design and prototype right from your laptop
๐น๏ธ Official support for @UnrealEngine & @godotengine
๐ The Android XR Developer Catalyst Program
Start building today โ https://t.co/dd2pXyzGBp
Cleaner styling workflows โจ
#Chrome150 Beta is here to streamline front-end workflows natively:
๐ Auto-scale text to fit containers
๐ Native relative color alpha
๐ Seamless smooth-scrolling
Chrome 150 is now in beta, bringing clean native platform features to streamline your styling and navigation workflows โ https://t.co/EUunzOYuXu
๐จ CSS text-fit: Scales text nodes to perfectly fit container widths
๐ Relative alpha: Derives translucent colors natively via the alpha() function
๐ Scroll Promises: Resolves when programmatic smooth-scrolls complete
Real-time global apps just got way easier to build. ๐โก
Gemini 3.5 Live Translate is officially in the Gemini Live API!
๐ Real-time translations
๐ Supports 70+ languages
๐ Live in Google AI Studio
Try it in Google AI Studio now
Global meetings just got easier. ๐
Gemini 3.5 Live Translate is now available in Gemini Live API! Developers can build experiences for international audiences in over 70 languages.
Watch it automatically recognize language switches on the fly:
Try it in Google AI Studio: https://t.co/fMi1wwoXI0
Get demo code: https://t.co/BY1O8awCvI
Building fast AI experiences just got easier โก
Gemma 4 MTP is now merged into llama.cpp, enabling a fast + lightweight setup when paired with Gemma 4 QAT.
Building super fast experiences with Gemma just got easier.
Gemma 4 MTP is now officially merged into llama.cpp. Developers can now pair MTP with Gemma 4 QAT for a fast, lightweight setup.
Gemini 3.5 Live Translate is live via the Gemini Live API. 70+ languages, auto-detection, native intonation preservation, noise filtering, all in a single stream. Agora, LiveKit, and Pipecat integrations ready out of the box. Try it in AI Studio today ๐
Our latest audio model, Gemini 3.5 Live Translate, takes real-time speech translation to the next level for developers by delivering low-latency translation across 70+ languages.
By processing speech as it streams in near real time, the model enables devs to build low-latency audio experiences with:
โ Multilingual input: Understands multiple languages in a single session without needing to adjust settings.
โ Auto-detection: Identifies the spoken language and begins translation instantly.
โ Native audio processing: Generates more natural-sounding speech that preserves speakers' intonation, pacing, and pitch.
โ Noise robustness: Filters out ambient noise for clearer conversation in loud environments.
How did a simple AI translation feature help increase sales?
Karrot built it in under 2 weeks using Gemini + Firebase AI Logic, leading to
๐ฌ 2.4ร more users starting chats with sellers
๐ Better experience for non-English users
Building AI features for Android? Give this a read๐
Building AI features doesn't require a dedicated backend. ๐ฒ
Karrot used @Firebase AI Logic and Gemini Flash Lite to build a seamless in-app translation feature in under 2 weeks, driving 2.4x more chats among non-English users. See how they did it โ https://t.co/9F6fkISwto
AI can fuel your technical curiosityโก
If you are working on your first project @HiThereJoe shows how AI tools can help you by:
๐ Building quickly
๐ Instant explanations for complex logic
๐ Find answers easily
What are you building this week? Get inspired here: https://t.co/aUBKispMk0
AI helps lower the barrier to entry for new developers getting started. ๐ฒ
@HiThereJoe shares how to use AI to build quickly, look up documentation, and explain things as you learn โ https://t.co/MqFItl39Sz
AI agents require fast iteration, but security shouldn't be an afterthought๐
Agent Sandbox on GKE provides:
โก Fast deployment
๐ก๏ธ Built-in security
โ๏ธ Simpler infrastructure management
So, you spend more time building agents, instead of worrying about infrastructure.
Stop wrestling with infrastructure to run your AI agents โก
Agent Sandbox on GKE provides the secure, scalable environment you need to deploy workloads fast. Equip your developers to build and ship with total confidence!
Read more โคต๏ธ https://t.co/2oziTTWJm3
What would you delegate to a specialized AI agent?
๐ย Research
๐ย Planning
๐ย Coding
๐Testing
๐Customer support
Multi-agent architectures allow different agents to focus on what they do best - working together to solve complex tasks more efficiently.
Learn how to build them in Google AI Studioย ๐ท https://t.co/J8naLIK8Hk
Why have one AI agent handle everything when you can build a team? ๐ค
Multi-agent architecture allows a main agent to spin up specialized sub-agents and handle complex tasks in parallel - making workflows faster and more efficient inside Google AI Studio.
Check out the full I/O session: https://t.co/J9YyLBd2r3 ๐งต๐
Spend less time debugging and more time building.
With Android Studio Quail.1, Gemini can help analyze crash reports, identify root causes, and suggest fixes directly in your workflow.
A useful productivity boost for Android developers shipping apps at scale.
Android Studio Quail.1 is stable and ready for you to use in production.
This release brings efficiency gains to your agentic workflows, enabling you to analyze and fix app crashes using Gemini. Learn more โ https://t.co/YdMOIDyaV1