GitHub announced on 3 November 2026 that GitHub Copilot is now available in JetBrains Air, JetBrains’ environment for handing off coding tasks, steering them while they run, and reviewing the results.
What the changelog says
- Copilot connects to Air through the Agent Client Protocol (ACP), and you pick it from Air’s agent picker using your existing Copilot subscription.
- You can track several agent sessions at once, with project context and IDE tools, and review changes next to the code.
- Cloud tasks run in isolated environments through Air Teams, started from the IDE or the browser. Each runs in a Docker container on a fresh virtual machine, with traffic going through a proxy and no state carried between tasks.
- Cloud tasks are billed to your Copilot plan, not JetBrains AI credits. If a subscription or credits are unavailable when the environment starts, the task fails.
- Air’s governance and analytics do not include Copilot usage, and Air administrators can disable Copilot but cannot set Copilot spending limits there.
The interpretation
The interesting part is not another place to type a prompt. It is that the workspace and the AI agent are now separate choices. One workspace can host agents from different vendors, and a vendor’s agent can show up in several workspaces. For a founder or product lead, “which AI coding tool do you use?” is turning into “which agent, in which environment, under whose billing and controls?”
The governance gap matters most. Usage in Air is not visible in Air’s own analytics, and spending caps sit elsewhere. That is manageable, but only if someone owns it.
So what
If you commission software, ask your development partner which agents and workspaces touch your code, where usage and cost are tracked, and who can switch them off. AI speed is real; unmanaged AI spend and unclear access are the risks. BuildApps builds with AI-assisted workflows under clear review and ownership rules. See AI-assisted development or start a project.