JetBrains Introduces Air, an Open System for Agentic Development

JetBrains on September 22, 2026 introduced JetBrains Air, an open system of products for agentic software development that spans its IDEs, team delivery workflows, and organizational governance, with the company’s Junie coding agent supported across all Air surfaces.

Three Products Under One System

The announcement, authored by Kirill Skrygan on the company blog, describes JetBrains Air as an open, coherent system of products for developers, teams, and organizations, inside and beyond JetBrains IDEs. The system is multi-surface and multi-service; each product solves a distinct problem, but the products work better together.

The announcement names three products. Air in JetBrains IDEs is an agentic development experience for directing and orchestrating agents and verifying their work inside JetBrains IDEs. Air Teams coordinates and automates software-delivery workflows involving developers and autonomous agents. Air Governance, formerly JetBrains Central, handles organizational policy, visibility, auditability, cost management, and accountability for AI-assisted and agent-driven development.

The launch brings together six months of public experimentation. JetBrains introduced JetBrains Central in March 2026 as an open control and execution system for agent-driven development, then rolled out Central CLI, shared context, cloud agents, automations, governance, and AI cost controls for teams and organizations.

JetBrains said Air expands the company beyond the individual developer workbench it has focused on for 26 years toward the wider system through which agentic work is initiated, executed, coordinated, reviewed, and governed, while adding that the IDE remains important to its future.

Availability and Access

The JetBrains Air product page presents the system as working with Claude Agent, Codex, Junie, Copilot, OpenCode, and any agent connected via the Agent Client Protocol (ACP), across JetBrains IDEs, web, CLI, and mobile surfaces, with mobile marked coming soon. Cloud runs are already available for some customers in JetBrains IDEs and the browser, with rollout to remaining customers planned gradually over the coming months, according to the page.

Air in IDEs, distributed as a plugin, lets developers keep several projects open, each with its own sessions, and track unread updates, changed files, and outgoing commits across all of them. The tool finds compatible agents on the machine for one-click connection, and additional agents can be added from an agent registry. Diffs open directly from an agent session, and developers can leave comments on specific lines for the agent to act on. For model access, users can bring an existing provider subscription or API keys, or use JetBrains AI with credits billed at public API rates, switching between subscriptions even mid-session.

Air Teams is offered through early access. It provides shared projects in dedicated cloud environments, automations that delegate repeatable tasks such as code review, release notes, and fixes to coding agents, per-project controls for VM size, internet access, and secrets, and MCP server connections made once and shared by the team and its agents.

Air Governance supplies rules and permissions for models and agents set for a whole organization or tailored to specific teams or individuals, flexible AI spend limits, support for bringing keys from Amazon Bedrock, OpenAI, Anthropic, or other providers, and usage analytics. JetBrains also lists Air Context, whose semantic indexing the company says speeds execution and lowers the cost of working in production-scale repositories, and Air Gateway, which brings terminal agents such as Claude Code and Codex into Air.

Multi-Vendor Design and the Agent Client Protocol

JetBrains said multi-vendor support is a foundational design principle of Air. The company argued that models vary in what they are good at and that rankings change every few months, that teams inside the same company already make different choices, and that organizations should not have to choose between using the best available tools and understanding what is happening inside their own engineering. Air does not require customers to use JetBrains’ own agent, the company said, and Air Governance is designed to extend visibility and cost governance across providers and tools.

Much of that openness runs through the Agent Client Protocol, an open protocol that defines how IDEs and AI coding agents communicate, developed openly by JetBrains and Zed with a stated goal of no vendor lock-in. JetBrains’ ACP page lists agents working over the protocol including Junie, Gemini CLI, GitHub Copilot, Codex, Cursor, Mistral Vibe, OpenCode, Kimi CLI, Qwen Code, Factory Droid, Cline, and Kiro CLI, and clients including JetBrains IDEs, Zed, and Neovim via plugin. According to the announcement, ACP standardizes the connection between an IDE and an agent’s full harness, including its planning, logic, tools, model routing, and observability, and an ACP Registry lets developers discover and run a growing range of compatible agents while working inside JetBrains IDEs.

Verification Argument and Release Plans

JetBrains argued that as agents take on more of the execution, the bottleneck shifts from producing change to understanding, verifying, and owning it: code becomes cheaper to generate but more expensive to verify, while responsibility for what ships remains with the people and organizations that ship it. The company said its 26 years of deterministic code intelligence provide a foundation for making agentic work more reliable and that it is seeing promising results from giving AI agents access to that intelligence, describing the benefit as economic as well as technical because agents otherwise spend time and money rediscovering information a codebase already contains.

JetBrains said Air will develop through a rolling series of releases, each spelling out what customers can use now, what is entering preview, and what remains longer-term direction. Planned extensions include mobile and remote experiences for initiating, monitoring, reviewing, and continuing agentic work; richer context drawn from code, architecture, repositories, runtime behavior, and organizational knowledge; and more work triggered by repository events, schedules, and delivery processes rather than by a developer opening an editor and issuing a prompt. The company said it will not name future products before their scope and availability are confirmed.