Why Google Antigravity 2.0 Split Its Most Popular AI Tools
TL;DR
Antigravity 2.0 splits the product into four parts: a desktop app for agent orchestration, an IDE for coding, a CLI for terminal workflows and an SDK for custom integrations. The new desktop app becomes the hub for scheduling, parallel sub-agents and complex agent runs. The IDE still exists, but now as a separate download. The rollout created confusion because some auto-updates replaced the IDE with the desktop app. Existing projects and settings are described as preserved in separate directories.
Nauti's Take
This is not just a product update, it is Google admitting agents do not fit neatly inside an IDE drawer. Serious AI builders need orchestration, terminal workflows and integrations as separately controllable surfaces.
The rollout was messy, but the architecture call is right.
Briefingshow
Google is moving Antigravity away from a single AI coding editor toward multi-entry agent infrastructure. That helps teams that want to operate agents across workflows, but it makes onboarding less obvious. Users who simply wanted to keep coding suddenly had to understand product architecture first.
Video
Sources
