JetBrains has unveiled Koog, an open-source framework crafted to simplify the development of AI-driven agents using Kotlin within the Java Virtual Machine (JVM) ecosystem.

Released on May 22 and now available on GitHub, Koog introduces a Kotlin-specific domain-specific language (DSL) aimed at streamlining the creation of intelligent agents. This modern toolkit allows developers to harness Kotlin’s expressiveness and conciseness to design autonomous systems more efficiently.

According to JetBrains, Koog was developed in response to the lack of native Kotlin libraries specifically tailored for building agentic AI solutions. The company emphasizes that Kotlin developers should not have to compromise when creating advanced AI systems — they deserve a framework that’s as adaptable and robust as the language itself.

Koog focuses on removing barriers to entry with an intuitive setup process and ready-to-use tools. Key features include:

• Easy onboarding for new users;
• Simplified agent definition and management;
• Built-in strategies for common AI use cases;
• Integration with Model Context Protocol (MCP) for smooth context transitions.

In addition to its core functionality, Koog supports streamed AI responses, efficient handling of extended conversation histories, and other features needed for building production-grade AI solutions.

By offering Koog, JetBrains aims to strengthen Kotlin’s position as a top-tier choice for AI development within the JVM environment.