Tech • IA • Crypto
Google I/O 2026 Developer Keynote 5-Minute Recap
TL;DR
Google has introduced Antigravity 2.0, a new agent-focused platform and SDK that enables developers to rapidly build, fine-tune, and deploy AI-powered applications across cloud, desktop, and Android environments.
KEY POINTS
Launch of Antigravity SDK and Managed Agents
Google unveiled the Antigravity SDK, providing developers with a unified “agent harness” optimized for the Gemini API. The system allows users to create custom AI agents in a single API call by combining instructions, tools, and datasets. Prebuilt agents are also available for immediate use, lowering the barrier to entry for experimentation and deployment.
From Idea to App Without Local Setup
A major feature is the ability to build applications directly in AI Studio without installing software or configuring local environments. Developers can generate full Android applications simply by selecting a build option and prompting the system, significantly reducing setup time and complexity.
Agent-First Desktop Application
Antigravity 2.0 introduces a desktop environment designed around agent workflows. It enables tasks such as fine-tuning models, generating code, and orchestrating deployments through natural language prompts, reflecting a shift toward AI-driven development interfaces.
Rapid Fine-Tuning with Gemma Models
The platform supports fast customization of models like Gemma 4 using techniques such as LoRA fine-tuning. In demonstrations, developers generated training and evaluation code, launched training jobs, monitored loss metrics, and deployed models within minutes, highlighting a streamlined machine learning workflow.
Seamless Cloud-to-Local Deployment
Antigravity enables hybrid workflows where training can be validated in the cloud and then deployed locally. Developers can reuse checkpoints from prior runs and push optimized models to personal devices, including laptops, without manual configuration.
CLI Integration and Automation
The Antigravity CLI extends capabilities from the Gemini CLI, allowing developers to manage agents, trigger training jobs, and automate pipelines entirely from the command line. This supports headless environments and remote GPU workflows, including SSH-based operations.
Expanded Android Development Support
Official Android support has been added, including a built-in Android CLI that simplifies environment setup without requiring Android Studio. Developers can still transition to Android Studio for production refinement, combining automation with traditional tooling.
Agent-Orchestrated UI Testing and Previews
Agents can interact with development tools to open UI previews, test layouts across device configurations, and validate features such as offline functionality. This introduces automated, agent-driven testing workflows for mobile applications.
Modern Web Guidance and WebMCP
Google introduced Modern Web Guidance, a curated set of best practices that helps agents implement contemporary web features. Alongside it, WebMCP enables websites to expose capabilities directly to browser-based agents, effectively making web applications “agent-ready.”
Chrome DevTools for Agents
New Chrome DevTools integrations allow agents to evaluate how generated code performs at runtime. This creates a closed feedback loop where agents can build, test, debug, and refine applications autonomously using MCP servers and specialized tooling.
CONCLUSION
Antigravity 2.0 signals a shift toward fully agent-driven software development, combining rapid prototyping, automated training, and cross-platform deployment into a single workflow powered by Gemini.
Full transcript
Today we're making that same Antigravity harness available to you with Managed Agents in the Gemini API. You can easily start creating your agents by adding custom instructions, tools, and data through a single call to the Gemini API. To help you get started, we've added a set of new custom agents that you can run instantly. Today we're launching the Antigravity SDK. Yeah, It gives developers and researchers the exact same agent harness optimized for Gemini, but with the ultimate flexibility to run it wherever and however you want. You can go from an idea to an Android app directly in AI Studio. No software to install, no SDK to manage, and no local environment needed. To start. Yes. Hallelujah. Okay. Awesome. Yeah. To start, all you have to do is select build an Android app and start prompting. We're introducing Google Antigravity 2.0, a new desktop application that is unabashedly agent-first. So I want to fine tune Gemma 4 to directly give me a bash command response with no additional fluff, so that I can use the response directly in my CI pipeline. I have a data set of prompt to bash commands. Write the training and eval code for doing a LoRA fine tuning over this data set, as well as the code to deploy my custom Gemma 4 model on my laptop. I've SSH'edinto my GPU-enabled VM, and I'm going to pull down the code from main that we just wrote in Antigravity 2.0. I've already had the Antigravity CLI installed on this machine and it's running in a separate tab. No GUI necessary. I just asked the agent to start the training job that we just wrote. I'm going to use the cloud to do a quick sanity check on the first few steps to make sure the loss is trending in the right direction, and everything is healthy. I actually kicked off the same exact run earlier. I let it run for a few hours, so I'm just going to use that checkpoint and ask my agent to deploy that resulting model to my laptop, so that we can all see that this does indeed work. In minutes. we used Antigravity to go from no code to fine-tuning Gemma 4 for on our own custom use case. So we took everything we learned from how you use Gemini CLI and rolled those insights into the Antigravity CLI as well. So starting today, the Antigravity CLI is available to all Gemini CLI users. Today we're also bringing official Android support to Antigravity so you can more easily deliver the most performant experiences for your users, no matter where they are or what kind of devices they have. Setting up the environment without Android Studio is kind of a pain. but now Antigravity has our new Android CLI built in. These screenshots are nice and all, but we need to see how this looks across multiple configurations and whether that AI summary really works offline. Antigravity enables you to orchestrate your agents, but the ideal setup is to use Antigravity, alongside Android Studio. You can easily transition to Android Studio at any time and get that production grade polish. For example, with the agent, I can ask “Open the Compose previews for the home screen in Android Studio.” Then pixels percolating. There it is. There really hasn't been a "solution" to keeping up with new features until now. Today we're launching a new tool to supercharge your agents Modern Web Guidance. A comprehensive and expert-vetted set of skills providing AI agents with a blueprint for modern web features. This looks amazing. But you know what? All of these menus, options, and sliders. So you know what I'm going to ask? Can we make this work seamlessly with agents? With WebMCP, we can now make this web page agent-ready in minutes. This proposed browser standard lets you expose web capabilities to browser-based agents. Okay, so we've added some new features to our website, but could we have the coding agent now test this like a real user would? That all changes with the new Chrome DevTools for agents. Your agent can finally see how the code it has written performs at runtime. With the MCP server, the CLI, and a set of tailored skills, it finally creates a closed feedback loop for agents. That's phenomenal for building, validating, and debugging. We hope you'll take what you've seen today and push it to the limits.