AI coding agents are everywhere. The problem is no longer "can an agent write code?" — it's "who's watching all these agents?" JetBrains Central is a direct answer: an open system that sits between your team and every coding agent it uses, providing governance, execution infrastructure, and shared context across the entire software delivery lifecycle.
The framing from JetBrains is blunt. Of the 11,000 developers surveyed in their January 2026 AI Pulse study, 90% already use AI at work. 22% use coding agents specifically, and 66% of companies plan to adopt them within a year. But only 13% use AI beyond the editor — in code review, CI/CD, or release pipelines. The gap between individual productivity and organizational impact is where Central lives.
Three Core Capabilities
- Governance and control — policy enforcement, identity and access management, observability, auditability, and cost attribution for agent-driven work. Some of this is already live via the JetBrains Central Console.
- Agent execution infrastructure — cloud runtimes and computation provisioning so agents run reliably across environments, not just inside your editor session.
- Agent optimization and context — a semantic layer that continuously aggregates information from code, architecture, runtime behavior, and organizational knowledge, giving agents system-level understanding rather than prompt-level guessing.
No Lock-In by Design
Central explicitly supports agents from outside the JetBrains ecosystem. Claude Agent, Codex, Gemini CLI, and custom-built solutions all plug in. Developers can initiate workflows from JetBrains IDEs, third-party editors, CLI tools, or a web interface. The architecture is layered — connect new tools and models without replatforming.
This is a notable posture from a company that has historically built tightly integrated, proprietary toolchains. JetBrains is betting that the control plane is more valuable than the agents themselves — that the company managing the orchestration layer wins, regardless of whose model is doing the code generation.
Code With Me Is Gone
The same announcement carried a quieter, more controversial footnote: JetBrains is retiring Code With Me, its real-time pair programming feature. Version 2026.1 is the last IDE release to include it, and the public relay infrastructure shuts down in Q1 2027.
The backlash was immediate. "We are a two-man company who work from home and we pair-program the entire time. This will be devastating," wrote one developer. Others questioned the logic of removing features from a product that could, in theory, have AI agents maintain them. The message is clear, even if it stings: JetBrains sees human-to-human pairing giving way to human-to-agent collaboration — and it's reshaping the product around that bet.
Air: The Human-Agent Workspace
Alongside Central, JetBrains is building out Air — a workspace where developers organize tasks, run agent-assisted workflows, and review results. A forthcoming Air Team layer adds multi-person coordination on top, letting teams manage work between humans and agents with shared visibility. Air is the front end; Central is the back end.
The Early Access Program launches Q2 2026 with a limited group of design partners. For organizations already juggling multiple coding agents across teams, Central is positioning itself as the missing infrastructure layer — the thing that makes agent sprawl manageable before it becomes a governance problem. Whether JetBrains can execute on "open but centrally managed" will determine if this becomes the Kubernetes of AI coding agents or just another dashboard.