JetBrains, the company behind IntelliJ, PyCharm, and a developer tool ecosystem used by millions, has announced Central, a governance and execution platform for AI coding agents in enterprise environments. The Early Access Program launches in Q2 2026 with a limited group of design partners.
Central addresses a specific operational problem: enterprises are adopting AI coding agents fast, but without infrastructure to track costs, enforce policies, or coordinate multiple agents across teams. The New Stack framed the product as JetBrains’ answer to “the cloud ROI crisis repeating itself” in agentic AI, where unchecked adoption leads to spending blowouts and painful rationalization.
What Central Does
The platform has three layers, according to JetBrains’ announcement.
Governance and control covers policy enforcement, identity and access management, observability, audit logging, and per-agent cost attribution. Some governance features are already live via the JetBrains Central Console.
Agent execution infrastructure provides cloud runtimes for running agents reliably across environments. Instead of each developer managing their own agent setup, Central handles sandboxing, resource allocation, and failure recovery.
Agent optimization and context builds a shared semantic layer across repositories and projects. When multiple agents work on related codebases, they access accumulated knowledge rather than starting cold each session. JetBrains says this enables “intelligent routing and task optimization, selecting the most appropriate models, tools, and execution paths for different tasks.”
Central supports JetBrains’ own Junie agent alongside Claude, Codex, Gemini CLI, and custom-built agents. It integrates with Git, CI/CD pipelines, Slack, Atlassian products, and Linear, according to Zen Van Riel’s technical overview.
The Data Behind the Launch
JetBrains cites its own AI Pulse survey from January 2026: among 11,000 developers worldwide, 90% already use AI at work, but only 22% use AI coding agents. Meanwhile, 66% of companies plan to adopt coding agents within the next 12 months. Only 13% of developers report using AI across the full software development lifecycle, such as code review and release pipelines, per the JetBrains blog post.
The gap between current adoption (22%) and planned adoption (66%) suggests the tooling and operational overhead is the bottleneck, not the models themselves.
Air: The Agent-Native IDE
Alongside Central, JetBrains launched Air (available at air.dev), an agentic development environment built for delegating coding tasks to multiple AI agents concurrently. Unlike traditional IDEs that bolt AI features onto a code editor, Air is designed around agent workflows. It supports running agents locally, in Docker containers, in Git worktrees, or in cloud sandboxes, according to Van Riel. Air is built on Fleet, JetBrains’ earlier IDE experiment.
Why It Matters
JetBrains is the first major developer tools incumbent to ship a dedicated governance layer for AI coding agents. The timing is notable. Enterprises are grappling with agent governance from the risk side: VentureBeat reported this week that OpenClaw has 500,000 instances running with no centralized kill switch. Central approaches the same governance gap from the solutions side. For teams already running multiple coding agents in production, the question is no longer whether they need governance infrastructure, but who will provide it.