Google Antigravity

The platform introduces a dual-view architecture: an Editor view that provides a familiar coding environment similar to VS Code with an AI agent sidebar for task-based coding, and a Manager view the product's most novel feature that functions as a control center for orchestrating multiple AI agents working in parallel across different workspaces and tasks simultaneously.

The Manager view enables asynchronous, multi-agent software development at a scale not possible in traditional single-thread AI coding assistants.

To address developer trust concerns around AI-generated code, Antigravity introduces an Artifacts system where agents produce verifiable deliverables task lists, implementation plans, test screenshots, browser recordings, and code diffs rather than raw, opaque tool calls.

This transparency layer allows developers to audit agent work at each step and maintain oversight without abandoning agentic autonomy.

The browser integration is a defining capability of Antigravity: built-in Chrome extension functionality allows AI agents to interact with web applications exactly as a human tester would clicking buttons, filling forms, navigating pages, and verifying UI behavior.

This transforms the browser into both a preview surface and a live QA testing environment, dramatically compressing the feedback loop between development and quality assurance.

Google Antigravity launched in a free preview phase with generous Gemini 3 Pro usage included, enabling developers to explore agentic coding workflows without upfront cost. Its combination of agent orchestration, browser-native testing, and a verification-first trust model positions it as Google's answer to the rapidly expanding AI-native development environment market.