Junie: NEW AI IDE Is Quite POWERFUL! AI Agentic IDE - Cursor + Windsurf Alternative!

Key Concepts

• Juny: A smart AI coding agent developed by JetBrains, integrated directly into JetBrains IDEs and Android Studio.
• Code Mode: Juny executes coding tasks autonomously.
• Ask Mode: Juny brainstorms, explores ideas, and provides explanations.
• Metabot: A large language model agent for quantitative finance, built using Juny.
• Project Guidelines: A set of instructions and structure defined for Juny to follow during project development.
• GPT-5: The underlying technology powering Juny, providing advanced understanding and reasoning.
• Human-in-the-Loop: The ability for developers to review and approve changes made by Juny.
• JSON Web Token (JWT): A standard for creating access tokens for authentication.

Juny: A New Smart AI Coding Agent

The video introduces Juny, a new AI coding agent developed by JetBrains. Unlike many AI assistants that simply generate code snippets, Juny is deeply integrated within JetBrains IDEs (like IntelliJ IDEA, PyCharm, etc.) and Android Studio. This integration allows Juny to understand the structure, logic, and relationships within a project, enabling it to propose execution plans, write and refactor code, and run tests directly within the IDE.

Code Mode vs. Ask Mode

Juny operates in two primary modes:

• Code Mode: In this mode, Juny executes tasks autonomously, writing and modifying code based on instructions.
• Ask Mode: This mode allows developers to brainstorm, explore ideas, and get explanations from Juny. It can be used to research best practices or understand existing code.

Key Features and Capabilities

• Seamless Integration: Juny integrates directly into JetBrains IDEs, avoiding workflow disruption.
• Contextual Understanding: Juny understands the project's structure, logic, and relationships.
• Execution Planning: Juny can propose execution plans for complex tasks.
• Code Generation and Refactoring: Juny can write new code and refactor existing code.
• Testing: Juny can run tests directly within the IDE.
• Syntax and Semantics Checks: Juny leverages JetBrains' built-in checks to ensure code quality.
• GPT-5 Powered: Juny is powered by GPT-5, providing advanced understanding and reasoning capabilities.
• Human-in-the-Loop: Developers can review and approve changes made by Juny.

Metabot Case Study: Quantitative Finance Agent

The video highlights a case study where Juny was used to build Metabot, a large language model agent for quantitative finance. Over a week, Juny handled tasks such as refining trading models, risk engines, and backtesting pipelines. The key takeaway is that Juny didn't just generate code snippets; it continuously evolved the project, monitored performance metrics, and maintained a stable workflow within the JetBrains IDE.

Solitary Card Game Example: UI-Driven Projects

Another example showcases Juny's ability to build a full solitary card game in a single shot. This demonstrates Juny's capability to manage complex UI-driven projects.

Getting Started with Juny

1. Download and Install: Download Juny from the link in the description and install it as a plugin for your JetBrains IDE (e.g., IntelliJ IDEA, PyCharm).
2. Account Setup: Open the Juny extension within the IDE and log in with a JetBrains account or register a new one.
3. Project Setup: Open a project and accept Juny's terms to begin using it.

Building a Task Management App with Juny

The video demonstrates building a task management application using Juny within IntelliJ IDEA.

1. Define Project Guidelines: Use the "Create Project Guidelines" feature to describe the desired application structure, features (task management, note-taking, AI features), and testing requirements in natural language.
2. Juny Generates Guidelines: Juny generates a predefined list of project guidelines based on the description.
3. Task Execution: Instruct Juny to follow the project guidelines and build the application. Juny creates a to-do list and executes the tasks, generating code live.
4. Human Review: Review the code changes and approve them.
5. Build and Run: Have Juny build and run the project.

The demonstration results in a functional task management application with features like dark/light mode, a task list, a notes board with draggable sticky notes, and a calendar.

Juny's Ask Agent: Architecture and Implementation

The video showcases the "Ask Mode" by asking Juny about the best architecture for a task management REST API with user authentication and team collaboration. Juny provides a thorough answer, referencing web search capabilities and its own context. The response includes goals, constraints, recommended architecture, data model, navigation, and REST API design.

The video then demonstrates switching back to "Code Mode" and instructing Juny to implement the recommended architecture, specifically creating JWT authentication with refresh tokens. Juny executes the task, building the authentication system.

Juny on GitHub

Juny is also available on GitHub (currently in early access). This allows Juny to work on repositories directly, picking up issues, opening pull requests, and solving tasks without requiring the IDE. This enables version-controlled and collaborative development.

Conclusion

Juny is presented as a powerful and integrated AI coding agent that goes beyond simple code generation. Its deep integration with JetBrains IDEs, contextual understanding, and ability to plan and execute complex tasks make it a valuable tool for professional developers. The video emphasizes Juny's potential to automate various tasks, improve code quality, and accelerate the development process. Juny is not just a coding assistant; it's a coding companion that understands your entire project and workflow.