Embabel 1.0 GA brings Goal-Oriented Action Planning to Java agents, from Spring's creator
InfoQ reported on August 3 that Embabel reached 1.0 general availability, giving Java and Kotlin developers a way to define agents as typed domain objects — goals, actions, and the conditions connecting them — rather than hand-coding prompt-and-tool-call sequences. It borrows Goal-Oriented Action Planning from video game AI: a runtime planner works out how to get from current state to goal using typed actions with preconditions and effects, and replans dynamically when conditions change mid-task. Co-creator Rod Johnson, who founded the Spring Framework in 2003, frames the layering precisely: "Spring AI exists at the level of the Servlet API, while Embabel is more like Spring MVC." For the large population of teams running Spring Boot services, this is the first credible typed agent programming model on the JVM.
Source
↳ Follow the thread