Follow us
Breaking
Product Reviews

JetBrains retires Fleet to build agentic developer environments

JetBrains will discontinue Fleet on December 22, 2025, to focus resources on Air, an agentic development platform. This strategic pivot shifts the developer role toward supervising AI agents that perform testing and refactoring tasks.

Share

The end of Fleet and the pivot to Air

JetBrains will stop Fleet downloads through the Toolbox app and other channels on December 22, 2025. The product spent three years in public preview but failed to find a clear market position between the established IntelliJ platform and the popularity of Visual Studio Code. I view the decision to discontinue Fleet as a necessary move because maintaining two separate IDE product families caused user confusion and diluted company focus. JetBrains now directs its resources toward Air, a platform built for agentic development where developers act as guides for AI agents that perform testing and refactoring. This shift follows a trend where 68% of CIOs plan to transform their data architecture. Fleet was designed as a lightweight, multi-language editor that could function as a full IDE with a single click, which was its main competitive advantage against tools like VS Code. The technical components of Fleet, including the frontend, the backend, the workspace, and the Fleet System Daemon, will now integrate into other JetBrains products like IntelliJ IDEA, PyCharm, WebStorm, and Rider. The UI uses modular "islands" that separate the editor from tool panels to reduce distractions.

AI accuracy and the data readiness problem

Air currently functions only on macOS and requires a subscription to Anthropic to operate. I would skip this preview if your workflow relies on Windows or Linux. The transition from Fleet to Air is a fundamental shift in software development because the human role changes from a primary coder to a supervisor of autonomous AI agents that execute complex tasks like refactoring. This transition happens while data readiness remains a massive hurdle for the industry. While 88% of data leaders say their data is ready for AI, 43% of those same leaders name data readiness as their biggest barrier to business success. I find it difficult to trust an agent when 92% of SVPs and VPs have acted on AI recommendations they later suspected were wrong because of bad data. The accuracy of these agents depends entirely on the underlying information, yet 21% of marketers say their CRM data is well prepared for AI. Even with advanced models, performance varies significantly based on how the data is structured. For example, Claude Sonnet 4.6 and GPT-5.3 Codex both answered 64.5% of questions correctly from raw tables, but Claude Sonnet 4.6 rose to 90.0% after three data models were added. JetBrains utilizes models from OpenAI, Google, AWS, and SpaceXAI, as well as its own Mellum and Mellum2 models. To support these services, JetBrains collects pseudonymized operational metadata and detailed code-related data, though it does not use customer content to train its models unless the user agrees.

Feature Fleet (Legacy) Air (Current Preview)
Platform Support Multi-platform macOS only
Primary Workflow Manual coding Agentic guidance
AI Model Dependency JetBrains AI Service Anthropic subscription

KMP abandonment and community reaction

JetBrains will also end Kotlin Multiplatform support in Fleet within three months. This decision follows feedback that users preferred KMP support in mature IDEs like IntelliJ IDEA or Android Studio. I see this as another instance where JetBrains chooses to strengthen existing tools rather than forcing users into a new, unproven environment. The community reaction to the Fleet discontinuation remains skeptical, especially after the project failed to replace the heavy IDEs it intended to challenge. Some developers noted that Fleet’s Vim plugin was too basic to be useful, even failing to allow file saving in certain versions. You know that JetBrains is known for supporting almost every programming language, but this pivot toward AI agents requires a complete change in how developers work. This move ignores the reality that 51% of CIOs name a missing governed data foundation as a barrier to scaling AI.

The company’s decision to stop the Fleet project also follows a period of price increases for all IDEs that took effect on October 1, 2025. These increases ranged from 11% to 30% depending on the product and subscription type. Customers can lock in current prices by renewing early for multiple years to avoid the higher rates. I find it hard to believe that a developer would abandon their stable IntelliJ setup just to manage an AI agent that might fail due to poor CRM or campaign data. Will developers actually accept a role as AI managers instead of programmers?

Share

Technewsdaily

Senior tech writer covering AI, gadgets and cybersecurity. Breaking down the news that matters, every day.