AI agents on the JVM
summary
Java ecosystem runs agents without rewriting in Python, showing embeddings and LLM calls stay on the JVM and integrate via Spring AI, LangChain for J or higher-level frameworks like Kook. Kook and Embabel offer GOAP planning (A*), graph execution with automatic checkpointing, and a rollback/undo tool registry for failed tool calls. Native tool calling lets agents invoke application services directly, reducing orchestration and cost compared to always calling LLMs.
AI-written summary · about
chapters (6)
- 0:00Introduction to AI agents
- 0:53Why JVM for AI agents
- 2:37Frameworks overview
- 4:02Kook framework deep dive
- 8:34Choosing the right framework
- 9:38Attracting AI engineers to JVM
related talks
54mGenAI Grows Up: Building Production-Ready Agents on the JVM
51mReliable AI Agents Using Domain Modeling with Koog in Java
34mAI Agents: Transforming Autonomous Decision-Making
53mUnleash the Power of AI Agents with Java Building Intelligent Application
27mEngineering Scalable AI with Java
28mAI Agents Graph: Your Following Tool in Your Java AI Journey
40mAgents in Action: Advancing Open Source AI Missions
45mWhy Do We Need an Agent Framework?