Spring Creator Launches AI Agent Framework for Java Virtual Machine

Rod Johnson, the founder of the Spring Framework, has launched Embabel, an open-source framework designed for authoring AI agentic flows on the Java Virtual Machine (JVM). Written in Kotlin, Embabel aims to not only compete with Python-based agent frameworks but to surpass them, offering seamless integration of generative AI into Java applications, particularly those built on Spring.

Key Features of Embabel

• Planning Step: Embabel introduces a unique planning step using a non-LLM AI algorithm for deterministic and explainable planning.
• Rich Domain Model: Encourages the use of Kotlin data classes or Java records to ensure type-safe, tool-able prompts that survive refactoring.
• Spring Integration: Designed for close integration with the Spring Framework, making it a natural choice for Java developers.

Beyond the JVM

While Embabel is currently focused on the JVM, Johnson has plans to expand the framework to TypeScript and Python in the future. This expansion aims to bring Embabel's innovative approach to a broader audience.

Addressing Critical Needs

Johnson highlights the importance of explainability, discoverability, and safety in AI agent frameworks. Embabel addresses these needs by:

• Allowing the mixing of models.
• Injecting guardrails at any point in a flow.
• Managing flow execution and composability at scale.
• Safely integrating with existing systems like databases, where LLM write access could be dangerous.

Ambitious Goals

"It’s early, but we have big plans," said Johnson. "We want not just to build the best agent platform on the JVM, but to build the best agent platform, period."

For more details, check out the Embabel GitHub repository and Johnson's Medium post.