JetBrains is making Air Governance generally available. This new management platform lets organizations manage AI agents, models, and providers from a single console. Administrators can enforce budgets, track costs per agent, and measure how much code AI writes. This also works with third-party agents, such as Claude Agent, Codex, and Gemini CLI.
Air Governance is not an entirely new product. JetBrains previously called the platform JetBrains Central, which it introduced earlier this year as an open platform for agentic software teams. Under its new name, it falls under JetBrains Air, the product family for agentic development. Air Governance serves as the organizational layer within that family.
The platform works with Junie, Claude Agent, Codex, Gemini CLI, and other agents that support the Agent Client Protocol (ACP). This applies to IDEs, the CLI, the browser, and mobile devices. Developers can already control these same agents in Air, JetBrains’ agentic development environment. Within Air, multiple agents can work in parallel on tasks within an existing codebase, either locally or in Docker containers.
Why is such a management layer necessary? According to JetBrains, most engineering teams use AI tools from different vendors simultaneously, each with its own contract, dashboard, and management environment. This makes it difficult to see who is using which models and what the associated costs are.
A firsthand problem as a starting point
JetBrains ran into that problem itself. The company’s AI spending increased roughly tenfold in the first half of 2026. The first usage report covering all vendors took four days to compile manually. JetBrains’ internal tools for this purpose eventually became Air Governance. The company now uses it to manage agent usage for about 2,300 employees each month.
In the console, administrators set credit limits for users, teams, and service accounts. These budgets are enforced, not just reported. They also approve agents and configure MCP servers at the enterprise level. With centralized BYOK, organizations link their own accounts with OpenAI, Anthropic, Amazon Bedrock, or Google Vertex AI. They can also simply use JetBrains AI. These settings can be bundled into AI policies by user group.
On the analytics side, the platform displays costs per developer, per agent, and per model or provider. Air Governance also measures how much code humans wrote and how much AI generated or modified.
Pricing and roadmap
There is a free version. According to the product page, the Basic plan includes centralized billing, spending limits, SSO, and AI analytics. The Business plan costs $20 per month per developer with AI access. Individual AI licenses are not required. Developers use a shared pool of AI credits, which organizations purchase separately. Organizations with 100 or more developers can choose the Enterprise plan.
JetBrains is also working on several enhancements. Approved MCP servers will also become available in third-party CLIs. Support for GitHub Copilot and cloud agents is coming via Air Gateway. More comprehensive analytics, credit limits for BYOK models, and cross-team skill management are also in the works.