In early September 2026, JetBrains announced JetBrains Air, an open and coherent product system for developers, teams, and organizations, within its integrated development environments (IDEs) and beyond. The system spans multiple interfaces and services. Each product addresses a distinct problem, but the publisher states they work best together.
JetBrains Air incorporates Air Teams, a collaboration space dedicated to agentic development, meaning work entrusted to agents, artificial intelligence programs capable of executing tasks autonomously. Air Teams provides teams and agents with the shared context, environments, tools, and instructions they need throughout the development cycle.
The company explains it has observed two trends since the launch of Air, presented “several months ago” as a first step. Developers do not need another standalone application, as tools and agents must be integrated into the environment where they already work. Furthermore, as individual developers became more productive thanks to agents, “the real bottleneck shifted to the team level.” JetBrains adds: “We therefore decided to evolve Air beyond a simple standalone desktop application and offer a broader, integrated system.”
Automations Shared by the Whole Team
Air Teams is built on four elements: automations, shared cloud environments, cloud tasks, and projects. Automations are workflows, meaning sequences of steps, that run autonomously in the cloud based on an event or a schedule. Air Teams provides 10 templates, including for code review, bug fixing, dependency updates (the external libraries used by a project), and documentation maintenance.
Each automation relies on four parameters: instructions, an environment, tools, and a trigger. Tools can include Jira, Figma, and Linear via connectors. The trigger can be a GitHub or Jira event, a webhook, or a schedule. In the example given by JetBrains, adding the Bug label to a ticket launches an agent, which reads the ticket, identifies the cause in the code, and opens a pull request containing a proposed fix that “an engineer can review.”
JetBrains presents three examples. For code review, the agent posts comments and a summary, then can approve the pull request or request changes. For ticket fixes in YouTrack, a team member can assign a simple, well-defined task to an agent. Finally, twice a week, an agent upgrades the project dependencies, compiles the code, runs the tests, and opens a summary pull request.
Environments and Projects That Don’t Depend on a Single Person
Each automation and each cloud task runs in an environment. In Air Teams, this configuration is stored in the repository, at the path .air/cloud/startup.sh, so it can be versioned and reviewed with the rest of the code. Once ready, the environment can be saved as a snapshot and shared. JetBrains clarifies that sharing the configuration does not mean sharing credentials: shared secrets can be used without being visible, while personal secrets remain specific to each individual.
Cloud tasks, on the other hand, are not tied to the device that launched them. A user can start a task from their IDE and then resume it on the web, and the agent continues working after the laptop is closed. A team project brings together members, shared environments, MCP connectors, and automations. Each project has its own service account and its own AI credits, and an administrator decides whether automations consume these credits or those of their creators. In the first case, they continue to function after their creator leaves.
