Embabel Agent

Embabel is an agentic AI framework for the Java Virtual Machine. It is written in Kotlin with a usage model intended to feel natural from Java, and its argument is that the qualities which matter once AI becomes mission critical, namely context handling, reliability, type safety, performance, observability and integration with existing enterprise systems, are strengths of the JVM rather than of an experiment focused ecosystem. The framework models work as Actions, Goals, Conditions and a domain model. Instead of a state machine or a fixed sequence, a planner formulates the sequence of actions dynamically using an algorithm that is not an LLM, then replans after each action completes. That lets a system combine known steps in an order nobody programmed, adapt to what the previous step revealed, and make its own decisions about parallel execution. Because conditions are usually inferred from data flow in the code, developers rarely handle them directly. Prompts are type safe: a call through the operation context creates a typed domain object from an instruction. Spring Boot integration, MCP tool support and several model providers, including Bedrock, Minimax and Z.ai, are covered in the reference documentation, and both Java and Kotlin starter projects are provided. The framework is free and open source under the Apache 2.0 license and is published to Maven Central. No commercial tier is advertised. Embabel comes from a team that includes Rod Johnson, the creator of the Spring Framework. The documentation hub is labeled beta and is itself built on Embabel, answering questions about the framework in natural language.