JetBrains Fleet 2.0 and AI Assistant integration details

JetBrains Fleet 2.0 provides IntelliJ-grade refactoring through a distributed architecture. This lightweight editor works as a polyglot tool for collaborative coding. The AI Assistant integration reaches more users because the AI Free tier provides 3 AI credits every 30 days. Users with existing JetBrains subscriptions access the AI Assistant without additional costs. These tools use frontier models like Claude 4.0 Sonnet, Claude 4.5 Haiku, Gemini 2.5, Gemini 3.0, Azure GPT-5, and Qwen 2.5. The AI Assistant integrates into all JetBrains IDEs, including IntelliJ IDEA, PyCharm, WebStorm, GoLand, Rider, CLion, and others. It provides code completions, AI chat, code generation, test creation, and documentation. It also supports autonomous workflow agents that handle tasks like Jira implementation and code review. The system provides image to code features by generating code from UI mockup screenshots. It also offers coaching through learning-oriented explanations of what the AI did and why. The AI Assistant analyzes error messages and code context to identify bugs and recommend fixes. This assistance includes refactoring suggestions and code explanation. It also assists with legacy code modernization by analyzing applications and replacing deprecated APIs. JetBrains claims the AI Assistant is the first vendor-neutral offering because it uses multiple large language models. The system automatically handles which model receives which queries to optimize tasks. The 30% average code autocomplete rate reported by JetBrains puts it among the more accurate autocomplete tools available. The interface for Fleet remains clunky compared to VSCode, which many users find frustrating.

JetBrains Air and the Agentic Shift

JetBrains Air repurposes the abandoned Fleet IDE codebase to create an agentic development environment. This tool, currently in public preview for macOS, orchestrates multiple agents like OpenAI Codex, Claude, and the in-house Junie. JetBrains Air utilizes the open Agent Client Protocol to facilitate vendor-neutral communication between different AI agents, which allows developers to orchestrate tools like Claude and Gemini within a single, isolated macOS workspace while maintaining agent isolation. Air bundles terminal access, Git, code navigation, and preview into one workspace. The Junie CLI functions as a standalone, LLM-agnostic coding agent for terminals, IDEs, and CI/CD pipelines. Junie supports models from OpenAI, Anthropic, Google, and Grok via bring-your-key configurations. It emphasizes codebase structural understanding to avoid what JetBrains calls "Shadow Tech Debt" and includes next-task prediction and MCP support. Junie provides one-click migration from Claude Code and Codex. This environment allows for the concurrent management of multiple AI agents in isolated sessions. Because the tool manages multiple agents, it handles complex tasks by assigning them to specific models. Pricing for Air starts at $10 per month. You should understand that the codebase remains the foundation for this new orchestration layer. Will the transition from Fleet to Air signal the end of the lightweight editor market?

AI Credit Limits and License Tiers

The current AI credit distribution determines how much assistance a developer receives. Different license tiers dictate the monthly limits for AI interactions. AI credits are calculated dynamically for each interaction based on the number of tokens used from both the input and the response. Token usage varies depending on the length, complexity, and context of a request. The AI Free tier provides 3 AI credits every 30 days. The AI Pro tier costs $20 per month and provides 20 AI credits. Users who select the AI Ultimate tier must pay $60 per month to receive 70 AI credits. The AI Enterprise tier also costs $60 per month and provides 70 AI credits. The AI Trial provides either 10 or 20 AI credits per trial period. For developers already paying for a JetBrains IDE, the AI Assistant adds no incremental cost. This makes it cheaper than adding GitHub Copilot at $10 per month or Cursor at $20 per month. Users can also select which model they use for different tasks to suit their specific requirements.