Published by Emerging Technologies Laboratory · via ETL Newswire
Technology· 

Google Pushes Agent Tooling Into the Browser With WebMCP Standard at I/O 2026

Alongside Gemini 3.5 Flash and a revamped Antigravity desktop platform, Google proposed WebMCP - a browser-native standard that lets AI agents call structured web tools instead of scraping UIs.

By Theo Okafor, Staff Reporter · Technology Desk

Google used its I/O 2026 developer keynote on May 19 to ship what it framed as an infrastructure upgrade, not a model release. The headline items, according to the Google Developers Blog published the same day, were Gemini 3.5 Flash, a second major version of its Antigravity development platform, and a proposed open web standard called WebMCP.

The architecture decision worth naming is WebMCP. Today, browser-based AI agents mostly navigate websites by simulating clicks and scraping the DOM - the same brittle path that broke every screen-scraping tool before it. WebMCP proposes a different contract: a site exposes its tools - JavaScript functions, HTML forms, registered API endpoints - so an agent can call them directly rather than guess at the interface. According to the Chrome for Developers blog covering the I/O announcements, the experimental origin trial opens in Chrome 149, with Gemini in Chrome support to follow. What most mainstream coverage skipped, according to a developer priority analysis published by DigitalApplied, is that WebMCP is being co-developed by Google and Microsoft inside the W3C WML Community Group - meaning this is not a Google-proprietary play, at least by intent.

The missing piece to name honestly: WebMCP is still a proposed standard in origin trial. It is not a shipping primitive. Whether it clears the W3C process and lands in non-Chrome browsers is an open question, and Google's history of proposing web standards that stall is relevant context here.

Antigravity 2.0 is the other structural move. The original Antigravity shipped alongside Gemini 3 in November 2025. Version 2.0 is a standalone desktop application - separate from the IDE plugin - built around agent orchestration. As described in the Google Developers Blog keynote recap, developers can now spin up specialized subagents to tackle parallel workflows, with built-in terminal sandboxing, credential masking, and hardened Git policies. An Antigravity CLI gives the same agent harness to terminal-first developers. An SDK lets teams deploy the harness on their own infrastructure. And a new Managed Agents endpoint in the Gemini API means a single API call provisions an isolated Linux sandbox where an agent can reason, use tools, and execute code - no local setup required.

Gemini 3.5 Flash is the model underneath all of it. Per the Google developer highlights post, it outperforms Gemini 3.1 Pro across most benchmarks while running four times faster than comparable frontier models. That speed profile matters specifically for agent loops: every turn in an agentic workflow pays a planning cost, a tool-use cost, and a synthesis cost. A cheaper, faster model makes those loops commercially viable at volume in a way that slower flagship models do not.

The competitive read here is straightforward. Anthropic and OpenAI have stayed largely in the model-plus-harness lane. Google owns Chrome, Android, and a seat in the W3C standards process, and it used all three at I/O to push the agent surface below the application layer and into the platform itself. The Antigravity SDK letting teams self-host the same harness Google runs internally is the clearest sign that this is a developer-acquisition play as much as a product launch.

Two caveats for teams evaluating the stack. First, according to the DigitalApplied developer analysis, neither the Antigravity CLI nor the 2.0 desktop app currently meets the bar for organizations that require auditable, open-source tooling in their pipelines. Second, Gemini CLI - the open-source tool that had accumulated over 100,000 GitHub stars - is being retired on June 18, 2026, per the same analysis. Teams depending on it have less than three weeks to migrate.

Sources cited:
- Google Developers Blog - Google I/O 2026 Developer Keynote (https://developers.googleblog.com/all-the-news-from-the-google-io-2026-developer-keynote/)
- Chrome for Developers Blog - Chrome at I/O 2026 (https://developer.chrome.com/blog/chrome-at-io26)
- Google Blog - I/O 2026 Developer Highlights: Antigravity, Gemini API, AI Studio (https://blog.google/innovation-and-ai/technology/developers-tools/google-io-2026-developer-highlights/)
- DigitalApplied - Google I/O 2026 Developer Priority Matrix (https://www.digitalapplied.com/blog/google-io-2026-developer-priority-matrix)
- MarkTechPost - Google Launches Antigravity 2.0 at I/O 2026 (https://www.marktechpost.com/2026/05/19/google-launches-antigravity-2-0-at-i-o-2026-a-standalone-agent-first-platform-with-cli-sdk-managed-execution-and-enterprise-support/)

Reporting by Theo Okafor, Staff Reporter, for the Technology desk · ETL Newswire staff
Read more at the source

This release was originally distributed via ETL Newswire. Visit Google Developers Blog - Google I/O 2026 Developer Keynote for the full story, related releases, and contact information.

Visit Google Developers Blog - Google I/O 2026 Developer Keynote →