Getting started with Antigravity
summary
Demo shows the Antigravity Editor and Agent Manager using Gemini Flash plus a specialized Computer Use model to run browser actions, take screenshots and verify results. The agent builds a medal-table then a heat map, commits files, runs parallel tasks, and provides diffs with one-click revert/versioning. Browser extension, VS Code import, domain allow-lists and sandboxed file access limit agent permissions. The builds succeed and include recorded playback for visual verification.
AI-written summary · about
chapters (17)
- 0:00Countdown
- 1:05Intro
- 3:20What are the friction points of deploying code without an agentic tool?
- 5:44What was it like to build and deploy code before Antigravity?
- 11:14[Demo] Getting started with Antigravity
- 24:58Antigravity browser extension explained
- 30:05[Demo] Creating a heat map with Antigravity
- 32:40What will the evolution of "traditional" AI-assisted editors look like?
- 35:32How do developers build today using Antigravity?
- 40:23When should developers use Antigravity vs. the Gemini app?
- 43:51Are the models on the backend for Google AI Studio and Antigravity different?
- 48:28The future of work for developers with agentic AI
- 52:52Is versioning or rollback supported in Antigravity?
- 54:52Is it possible to import projects from VS Code into Antigravity?
- 55:58How does Antigravity handle permissions, access control, and boundaries for AI agents?
- 57:29How do Andy Zhang and Kevin Hou use Antigravity in their workflows?
- 1:01:05Summary and next steps for developers
related talks
25mDefying Gravity
34mAutomate building Chrome Extensions with Antigravity and DevTools MCP
45mAgentic Dev: Stitch and Jules in the Antigravity of the Gemini Constellation
29mTools and Roadmap for AI-Native Developers
37mBuild agents with Gemini API
01h 03mBuild & deploy AI-powered apps
30mLive Demo – Develop and Deploy a Trustworthy Multi-Agent System
40mDevelop faster with Gemini Agents in Android…