Join Huyen Tue Dao as she sits down with Vladislav Tankov at KotlinConf'24 to uncover the latest in JetBrains AI. Gain insights into how JetBrains AI addresses concerns with clarity and reassurance, backed by collaborative efforts with industry leaders like OpenAI and Google.

JetBrains AI Overview

• JetBrains AI is developed in response to the AI revolution, integrating AI capabilities like code generation and contextual assistance into JetBrains IDEs.
• The key announcement is a new Kotlin-specific code completion model, boasting 3.7 billion parameters, optimized for better performance compared to open-source models.

Comparison with GitHub Copilot

• Unlike GitHub Copilot, JetBrains AI offers deeper integration with the IDE, utilizing advanced static analysis and context-aware features.
• It leverages JetBrains' long history of code analysis technologies to provide more integrated and context-sensitive code suggestions and refactoring options.

User Concerns and Reliability

• Addressing concerns about AI reliability and accuracy, Vladislav emphasizes using grounding techniques to ensure current and relevant information is retrieved and verified.
• The integration includes validation through IntelliJ's code analytics engines, enhancing the integrity of AI-generated code completions.

Multiple LLM Integration

• JetBrains AI integrates various large language models (LLMs) from providers like OpenAI, Google, and Meta, selecting the best model for different tasks based on performance and requirements.
• For enterprises, the system accommodates specific needs, allowing customization based on existing contracts and infrastructure.

Exciting Features and Future Work

• Vladislav highlights the powerful new multiline code completion and enhanced documentation systems, which now support Kotlin Multiplatform and Fleet.
• Future developments will focus on deeper integration, providing seamless functionalities such as natural language commands for code actions within the editor.