Google no longer expects developers to leave their preferred editor to use Antigravity, its autonomous coding agent platform. The company said this week that Antigravity now comes bundled into eligible Gemini Enterprise subscriptions, alongside new extensions that bring the same agent workspace into VS Code, Visual Studio 2026, JetBrains and Zed.
That is a different move than the one AI Insiders covered previously, when Google added file-based custom agents to the standalone Antigravity IDE. This announcement is not a capability added to Google’s own editor. It is Google conceding that developers were not going to switch editors just to get agentic coding, so the agent now travels to wherever they already write code.
Eligibility is narrower than the headline suggests. Google says the bundling applies specifically to Gemini Enterprise Standard and Plus license holders, not to every enterprise tier, and both the subscription bundling and the extensions are live today for customers on those two plans. Teams outside that band still need a separate add-on to get Antigravity.
For the audience that actually decides whether agent tools enter a regulated engineering org, the administrator controls matter more than the editor list. Google names five levers: sandbox and MCP (Model Context Protocol) permission limits that confine what an agent can touch, monthly budget thresholds enforced through Google Cloud Billing with pooled quotas across teams, opt-in overage caps, Workforce Identity Federation and Application Default Credentials for login without manual API keys, and a single toggle for full audit logging of prompts and agent actions.
Sandboxing, tool permissions, spend, identity and an audit trail are the five items a security team checks before letting an autonomous agent touch production code. Building all five into the subscription itself, rather than leaving them to third-party wrappers or ad hoc scripts, is what makes this an enterprise-readiness release rather than a feature release.
Google’s evidence for impact so far comes entirely from itself. The company cites AirAsia generating more than half of its production QA code with Antigravity, a figure that appears only in Google’s own post and in a quote from an AirAsia executive, not in any independently verified benchmark. Accenture, Cognizant and other named partners are quoted in similarly promotional terms, with no usage numbers attached to any of them.
Shipping extensions into rival editors instead of forcing developers into a dedicated Antigravity app concedes that the interface was never going to be the differentiator. Distribution is now the strategy: meet developers inside the tools they already use, and compete at the agent layer beneath the editor rather than the editor itself.
Engineering leaders evaluating agentic coding platforms should treat the administrator control set, not the list of supported editors, as the actual diligence checklist. Confirm sandbox scope, per-team budget enforcement and audit log retention before letting a pilot expand past a handful of developers.
Google announced the update on its Antigravity blog in a post titled “Bringing Antigravity to Gemini Enterprise,” which carried no dated byline at the time it was scraped.
