Google I/O Unveils Gemini 3.5 and Agent-First Development Tools
Google is embedding AI agents directly into Search, Android, and developer workflows used by billions of people daily.
Gemini moves from chat to action
At Google I/O on May 19, the company unveiled Gemini 3.5 Flash, the first in a new series of models combining frontier intelligence with the ability to act. The model is generally available through the Gemini API, Google AI Studio, and Android Studio, and became the default model in the Gemini app and AI Mode in Search on launch day.
Google also announced a redesigned Search box that accepts text, images, files, videos, and Chrome tabs as input, and can reason across all of them. The company is merging AI Overviews and AI Mode into a single, seamless AI Search experience that flows from a query to a search results page with an AI Overview, to follow-up questions in AI Mode.
Antigravity 2.0 for agent orchestration
Google shipped Antigravity 2.0, an agent-first development platform that includes a standalone desktop app, CLI, SDK for Python, TypeScript, and Go, plus a Managed Agents API and Enterprise Agent Platform. The platform allows developers to spin up specialized sub-agents to tackle complex workflows, with built-in sandboxing, credential masking, and hardened Git policies.
The company also introduced new developer tools including the Android CLI, which enables AI agents to tap into Android Studio capabilities. Google released Android skills to help language models execute best practices for complex workflows, and published Android Bench, a leaderboard for LLM performance on Android development tasks.
What this means for developers
The developer keynote emphasized tools designed for the era of AI agents that can independently navigate complex tasks across entire workflows. Google introduced WebMCP, a proposed open web standard that allows developers to expose structured tools so browser-based AI agents can execute tasks with greater speed and reliability. An experimental origin trial starts in Chrome 149.
The company also released Modern Web Guidance, a set of expert-vetted skills for coding agents that supports over 100 use cases and integrates with Baseline standards. Google stated that 75 percent of all new code at the company is now AI-generated and approved by engineers, up from 50 percent in late 2025.