Large Language Models are no longer just answering questions. They are becoming agents that plan, call tools, loop until quality is high, and even ask humans for help. This new agentic approach unlocks complex business workflows, but only if developers can use it in a controlled, enterprise-ready way. In this session, we’ll take a hands-on journey from a simple REST endpoint to a fully functioning agentic workflow powered by the new LangChain4j-agentic module and a local model running with Ollama. You’ll see how to: • Define typed Java agents with @Agent • Wire tools into your agents for real business actions • Orchestrate multi-step workflows with sequence, loop, and parallel patterns • Hand off execution to a Supervisor for true autonomous behavior • Keep everything observable, testable, and ready for production If you’re a Java developer or architect curious about how to move beyond single-prompt demos and bring agentic AI into enterprise systems, this talk will show you exactly how, including runnable Java code, practical design patterns, and the pitfalls to avoid.

Session Speakers

Alex Soto

Developer Advocate at IBM