JetBrains has described how it began centralizing AI usage after development-related spending increased roughly tenfold in six months. Rather than restricting engineers to a small set of approved tools, the company built a shared access and accounting layer intended to preserve tool choice while giving teams greater visibility and control over consumption.
The change came as AI usage accelerated across the company. JetBrains said most developers were using between three and five AI tools each month, while token consumption began rising sharply from January 2026 alongside the arrival of increasingly capable frontier models including Claude Opus 4.5 and 4.6.
That diversity made understanding the bill increasingly difficult. JetBrains initially spent four days manually collecting usage and expense data into spreadsheets, producing a snapshot of where money was being spent across providers and tools. The company subsequently automated the process through provider APIs and internal dashboards, giving teams a continuously updated view of current and forecasted expenditure.
Visibility, however, did not provide a convenient way to intervene. The dashboards could show where spending was occurring, but requests still went directly from individual tools to their respective providers.
The next step emerged from an internal tool originally built by one developer. JetBrains expanded the wrapper into Central CLI, providing a common way to invoke both its own and third-party AI tools. Requests made through the CLI are routed through the company’s existing AI platform, creating a shared control point between developers and the underlying model providers.
That changed the role of the platform from observing AI spend after the fact to participating in the traffic generating it. JetBrains can apply its existing AI-credit system to third-party tools, while managers can view spending and configure limits for individual developers, teams, or larger organizational groups.
The company deliberately avoided solving the problem by sharply reducing the number of tools available. JetBrains said it had seen other organizations standardize on one or two AI products, but argued that the pace of change in the market makes it difficult to know which tool will be best for a given task even a few months later.
Centralizing the access layer allows those two concerns to be separated. Developers can continue choosing among supported tools, while accounting, access management and spending controls sit underneath them.
More than 1,000 developers adopted Central CLI within weeks, according to JetBrains, although the system does not yet cover every source of AI expenditure. Some terminal-based agents and personal subscriptions remain outside the managed path, while the company is still working on policies for distributing AI budgets fairly between different users and teams.
The wider cost-management problem is becoming increasingly familiar. The FinOps Foundation has described generative AI as a growing part of the FinOps remit and recommends centralized approaches to tracking AI usage and costs. Other organizations have responded more directly to rising consumption: ITPro reported that Accenture asked employees to curb unnecessary AI use, while Uber introduced monthly limits after exhausting an annual AI budget within four months.
JetBrains is taking a different route. Rather than beginning with restrictions on which tools developers can use, it is attempting to centralize the infrastructure through which those tools are measured and governed.