Google's Developer AI Push: Antigravity, Managed Agents, and WebMCP

Google's I/O 2026 developer keynote on May 19 was dense with infrastructure announcements aimed squarely at teams building AI applications. The headliners: Antigravity 2.0, managed agents via the Gemini API, and a WebMCP origin trial in Chrome 149.

The short version: Google is building the scaffolding for AI agents at the infrastructure level—from a new CLI and SDK for agent orchestration to browser-level protocols for connecting web apps to AI models.

Antigravity 2.0 and the Antigravity SDK

Antigravity 2.0, Google's agent orchestration framework, ships alongside a new Antigravity CLI and Antigravity SDK. These tools are designed to make it easier to deploy, manage, and connect AI agents across services. The CLI in particular lowers the barrier for developers who prefer terminal-first workflows to building and running agents without a UI.

Google AI Studio also received updates, though details on the exact scope were not fully enumerated in the keynote. The broader signal: Google wants AI Studio to become the default workbench for teams prototyping and deploying Gemini-powered applications.

Managed Agents via Gemini API

The Gemini API now supports managed agents—a hosted layer where Google handles the operational overhead of running agents, including state management and task orchestration. This is a direct play for developers who want agentic capabilities without building the infrastructure themselves.

Alongside this, Android-specific AI tooling expanded with Android CLI, Android Skills, and Android Bench with Gemma 4. A Migration Agent preview was also announced, aimed at helping developers port existing applications to newer architectures.

WebMCP: Connecting the Web to AI Models

The most technically interesting announcement for web developers was WebMCP, now in an origin trial in Chrome 149. WebMCP is a browser-level protocol that allows web applications to communicate directly with AI models—local or remote—using a standardized interface. It is an early-stage feature, but it points toward a future where AI model access is a native browser capability, not a third-party API bolt-on.

Chrome DevTools for agents was also announced, giving developers observability into how AI agents operating within the browser behave. The HTML-in-Canvas API origin trial rounds out the browser-side announcements.

FAQ

What is Antigravity 2.0?
Antigravity 2.0 is Google's agent orchestration framework, updated at I/O 2026 with a new CLI and SDK. It is designed to help developers build, deploy, and manage AI agents within the Google ecosystem.

What is WebMCP and why does it matter?
WebMCP is a browser protocol in origin trial in Chrome 149 that lets web apps talk directly to AI models. If it advances past origin trial, it could standardize how web applications integrate AI model access at the browser level.

What are managed agents in the Gemini API?
Managed agents in the Gemini API let developers deploy agentic workflows with Google handling the operational infrastructure—state, orchestration, and execution—rather than building those systems themselves.