Google’s I/O developer announcements on May 19, 2026 show a direct push into agentic software development. The company introduced Antigravity 2.0, Managed Agents in the Gemini API, native Android support in Google AI Studio, and new integrations between AI Studio, Android, Firebase, and Antigravity.
The engine is Gemini 3.5 Flash, which Google says outperforms Gemini 3.1 Pro across almost all benchmarks while running four times faster than other frontier models.
What changed
Antigravity 2.0 is a standalone desktop app for orchestrating multiple agents in parallel. Antigravity CLI gives terminal users a lighter surface. The SDK lets developers define custom agent behavior and host it on their own infrastructure.
Managed Agents in the Gemini API are the bigger platform move. Google says developers can create an agent with a single API call, give it tools and code execution in an isolated Linux environment, and resume that environment in later interactions.
Why this matters
The developer-agent race is moving from “a model can code” to “a platform can host, supervise, and resume agent work.” Persistent environments, custom instructions, skills, and tool access are the pieces that make agents useful across real projects.
Google also added native Android app generation and Play Console test-track publishing from AI Studio. That points toward prompt-to-app workflows becoming part of mainstream developer tooling.
What to watch next
Watch whether Managed Agents become reliable enough for CI-like workflows, not just demos. The winning platform will be the one developers trust to keep state, run code safely, and hand back reviewable changes.
