TLDRocket
Sign in

Google’s AI coding agent just escaped its own IDE

The New Stack Amanda Caswell

Google put Antigravity into VS Code, JetBrains, Visual Studio, and Zed. That makes its coding agent easier to use — and easier to bill and control.

Based on reporting by The New Stack, Amanda Caswell — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Google is pushing Antigravity out of its own desktop app and into the editors developers already live in. The company said Thursday it now has extensions for Visual Studio Code, Visual Studio, JetBrains IDEs, and Zed, and it is also opening access through eligible Gemini Enterprise subscriptions.

That changes the feel of the product. Instead of moving a project into Antigravity 2.0, developers can open an agent chat in a side panel, review inline diffs, inspect plans, and hand off multi-step work without leaving their usual workflow. One Antigravity account follows them across those environments, so there’s no separate sign-in mess or license juggling.

Google is clearly aiming at enterprise teams that don’t standardize on one editor. The VS Code extension is available now in Microsoft’s marketplace on macOS, Linux, and Windows. Support for Visual Studio 2026 and .NET solutions is in preview, and the JetBrains lineup starts with version 2026.2.1, alongside Zed.

But the bigger story may be the guardrails. Admins can enable Google’s developer tools for Gemini Enterprise Standard, Plus, or Standard Emerging Market plans, and they can cap monthly spend per Google Cloud project. When quota runs out, Antigravity either stops or falls back to pay-as-you-go pricing. Google says user- and team-level controls are coming later this year; for now, everyone in the same edition, project, and region draws from a single pool of credits that resets on a rolling seven-day basis.

That matters because agentic coding burns tokens fast. Google says a nontrivial engineering task can use 150,000 to 200,000 tokens, and multi-agent handoffs add more each time work moves between agents. The company is not alone in worrying about that bill: Microsoft has already introduced AI token budgets for its engineering groups, and other internal budgets at companies like Uber and Amazon have made for ugly reading.

My take — AI-written commentary, not fact-checked reporting

This is the real enterprise AI play: not a smarter agent, a tighter leash. Everyone loves the demo where the bot ships code; fewer people want to talk about the billing meter sprinting while the bot does it. Google knows that control planes sell better than wizardry when the work has to survive contact with a company budget.

Read more about this at: The New Stack

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.