Google Antigravity 2.0 desktop app
- Google introduced Antigravity 2.0 at I/O on May 19, 2026, expanding its agent-development toolkit with a standalone desktop app, CLI, SDK and Managed Agents API. - Google’s product page says Antigravity 2.0 is a “command center” to manage multiple local agents in parallel, with background workflows and scheduled tasks. (antigravity.google) - Developers can download Antigravity from Google’s product site and documentation, where Google lists CLI migration guidance and SDK setup details. (antigravity.google)
Google did not unveil Antigravity 2.0 first in an isolated X post on May 20. The company announced the product at Google I/O on May 19, 2026, as part of a broader agent-development push that also included a new Managed Agents API, an updated CLI and an SDK. Google described Antigravity 2.0 as a standalone desktop application built for an “agent-optimized experience,” while its product page says the app is designed to orchestrate multiple local agents in parallel. (antigravity.google) The May 20 X post appears to be a social-media amplification of that launch rather than the original source of the release. Google’s own pages and documentation already describe the same core functions shown in social clips: parallel agent work, background workflows, subagents and integrations across Google’s developer stack. ### If this was already announced by Google, what is actually new here? May 19 is the key date. In Google’s I/O developer materials, the company said it was extending Antigravity into a larger ecosystem that now includes Antigravity 2.0, Antigravity CLI, Antigravity SDK and the Managed Agents API. (cloud.google.com) Google Cloud executives Addy Osmani and Alan Blount wrote that the toolkit is meant to let developers “build locally and deploy securely to the cloud on a shared protocol layer.” (antigravity.google) Google’s developer highlights page said Antigravity 2.0 is a new desktop platform, while the CLI and SDK broaden how developers can work with the same agent harness. That places the desktop app inside a larger product family rather than as a one-off demo. ### What does the desktop app itself do? Google’s Antigravity product page says Antigravity 2.0 is “your command center to manage multiple local agents in parallel.” The same page says users can group conversations into projects, operate across multiple workspaces and automate routine tasks with scheduled messages. (cloud.google.com) Google’s documentation and codelabs describe Antigravity more broadly as an “agent-first” development platform in which autonomous agents can plan, code, browse the web, validate work and iterate on engineering tasks. A getting-started codelab published April 30 said the product was available as a preview for personal Gmail accounts and required local installation on Mac, Windows and certain Linux distributions. (blog.google) ### Where do the multi-agent and background-task claims come from? Google’s own product copy says the agent-first surface is built to “orchestrate multiple subagents across all your environments” and to run “complex, long-running workflows in the background.” The page also says other agents can simultaneously handle research, implementation and testing. (antigravity.google) Google’s Japanese I/O developer highlights page adds more detail, saying Antigravity 2.0 supports orchestration of multiple agents, dynamic subagents for parallel workflows and scheduled tasks for background automation. (codelabs.developers.google.com) That description closely matches the capabilities circulating in social posts. ### Did Google say anything about Claude Code, GitHub workflows or local orchestration APIs? Google’s official materials confirm local orchestration, subagents and an SDK, but they do not, in the sources reviewed, explicitly present Claude Code or GitHub workflows as launch-day headline integrations. (antigravity.google) The product and documentation pages instead emphasize Google AI Studio, Android, Firebase, Google Cloud projects, MCP servers, plugins, skills, hooks and sidecars. Google’s CLI documentation says the terminal product brings the same core capabilities as Antigravity 2.0, including multi-step reasoning, multi-file editing, tool calling and conversation history. (blog.google) Google also says the SDK provides programmatic access to the same agent harness used in its products. ### How does this fit into Google’s broader AI rollout? Google tied Antigravity 2.0 to Gemini 3.5 Flash and its enterprise agent stack at I/O. The company said the Managed Agents API lets developers launch agents in isolated Linux environments with a single API call, and said those agents are powered by the same Antigravity harness. (antigravity.google) Google also introduced a new AI Ultra subscription priced at $100 a month, with Google saying the plan includes five times the Antigravity usage limit of the AI Pro plan. (antigravity.google) TechCrunch and Google’s own I/O materials both linked that pricing to the broader Antigravity launch window. ### Where can developers see what ships next? Google’s Antigravity product site now lists downloads for Antigravity 2.0, the CLI, the SDK and the IDE. The documentation hub also includes migration pages, CLI setup and SDK quick-start materials, which are the clearest public markers of what Google is shipping beyond the I/O stage demos. (blog.google) (antigravity.google) (techcrunch.com)