JetBrains Launches Air to Manage Many AI Agents

JetBrains has spent 26 years building tools for the individual programmer sitting at a keyboard. On Monday the company signaled a different future. It launched JetBrains Air, an open system of products built for software work done by people and AI agents together, the company announced.

One place to run many agents

Air's core idea is that a single developer may soon be supervising several AI agents at once, and existing editors were not built for that. The environment lets a programmer start, watch and steer multiple agents in parallel, and it works with tools from across the industry rather than only JetBrains' own. At launch it supports OpenAI's Codex, Anthropic's Claude Agent, Google's Gemini CLI and JetBrains' in-house agent, Junie.

Underneath sits JetBrains Central, an open control and execution layer the company introduced earlier this year, Techzine reported. Central handles shared context, cloud-run agents, automations, and the governance and cost controls that larger teams ask for once agents start doing real work.

Built on a false start

Air did not appear from nowhere. It grew out of Fleet, an earlier JetBrains editor the company set aside, and it reached public preview in the spring before this week's wider launch. The pitch has shifted from a better code editor to something closer to a control room for a team of software agents.

The move puts JetBrains on the same ground as a growing field trying to make autonomous coding practical rather than a demo. We wrote recently about an open-source coding agent closing the gap with paid rivals, and about a platform that lets AI agents run a business. The common thread is orchestration. The hard part is no longer getting one agent to write code. It is managing many of them without losing control of the result.