How JetBrains AIR Redefines AI‑Powered Development: A Hands‑On Guide

When AI‑assisted programming becomes the default workflow of an IDE, what does the development tool look like? JetBrains answers with AIR, an AI‑first IDE that rebuilds the IDE around AI rather than merely plugging it in.

AIR Overview

AIR is a brand‑new independent IDE built from scratch by JetBrains. Its core principles are:

AI is the default workflow, not an optional add‑on.

Emphasis on determinism and explainability, avoiding “black‑box” magic.

Developers retain full control over the generated code.

Quick Start: Three‑Step AI‑Native Workflow

JetBrains AIR is an Intelligent Agent Development Environment (ADE) that lets you delegate coding tasks to an AI agent while keeping full oversight. The workflow revolves around “Define Task” and “Review Result”.

1. Install and First Run

Currently macOS‑only; Windows/Linux support is planned for 2026. After launching, sign in with a valid subscription to reach the welcome page.

2. Open a Project

You can either open a local folder or clone a Git repository:

Open: click Open and select the folder.

Clone from Git: click Clone from Git , enter the repository URL and destination path.

On first open, AIR asks whether to trust the code (executing untrusted code is risky).

Preview – limited functionality, no code execution.

Trust – allows execution (recommended only for trusted projects).

3. Define a Task

Interact with the AI agent via chat mode, providing a clear description such as “Add pagination to the order module, 10 items per page”, and optionally reference a file, e.g. src/orders/service.py. AIR also supports a “plan mode” where the AI first generates an execution plan.

Task configuration lets you set execution environment, AI model, and permission mode.

Permission modes include:

Ask permission – prompts on first tool use.

Auto edit – automatically accepts file edits.

Planning – only analyses code, no edits.

Full access – skips all permission prompts.

4. Run and Manage Tasks

Task status can be Running, Waiting for user action, Finished, or Canceled/Archived. Multiple tasks run in parallel without interfering with each other.

5. Review and Commit Changes

In the Review tab, use the diff panel to compare new and old code (unified or split view) and add comments next to line numbers, just like a code review.

6. Advanced: MCP Server Integration

Model Context Protocol (MCP) lets AIR connect to external tools such as databases or APIs. Add a JSON configuration under Settings → AI → MCP Servers:

{
  "mcpServers": {
    "alphavantage": {
      "command": "uvx",
      "args": ["av-mcp", "YOUR_API_KEY"]
    }
  }
}

Example: connect to the Alpha Vantage API.

Conclusion

AIR lowers the barrier for human‑AI collaboration by making AI the native workflow of the IDE. Its clear three‑step process, configurable permissions, parallel task execution, and MCP extensibility provide a controlled yet powerful AI‑driven development experience.