JUG Oberpfalz

Welcome to the JUG Oberpfalz Online Live Coding Sessions!
Our online sessions are held in English and streamed live on YouTube. The recording will be available afterwards. Join the live chat, ask questions, and follow along as we explore what the next generation of software development might look like.

This session is hosted by Johannes Rabauer with special guest Bartolomeo Sorrentino from Softphone srl, creator of LangGraph4j.

Session Topic
In this session, we build a real agent live: a train-delay watcher that pulls genuine departure data from the Deutsche Bahn public API, decides whether a delay needs a human's sign-off before rebooking, and has to cope with the API itself failing partway through — no mocked data, nothing scripted in advance.

Before the build, Bartolomeo Sorrentino, the creator of LangGraph4j, joins for an hour of conversation: what a tool-calling loop in LangChain4j already gives you for free, what it doesn't, and where LangGraph4j's graph layer — built on top of LangChain4j and Spring AI, not against them — starts pulling its weight.

We'll also touch on how it compares to Embabel's goal-planning model and Temporal's crash-recovery model.
Then, screens open: we take a bare tool-calling loop and push it until it needs a real decision point — a branch on an actual delay, a retry on a failed call, a human approving a rebooking — and rebuild it as a graph with a real checkpoint backend.

What we'll do

• Talk with Bartolomeo Sorrentino about why he built LangGraph4j and how it fits next to LangChain4j and Spring AI
• Draw the line between a bare tool-calling loop and a graph with branching, cycles, and human-in-the-loop approval
• Compare LangGraph4j's explicit-graph approach to Embabel's goal-planning model and Temporal's crash-recovery model
• Build a train-delay watching agent live against the real Deutsche Bahn public API — no mocked data
• Branch the agent between an automatic minor-delay notice and a human-approved rebooking
• Handle a real failed API call as part of the agent's own logic, not hidden behind a fallback
• Answer questions from the community

Topics

• LangGraph4j
• LangChain4j
• Spring AI
• Java
• AI agents
• Human-in-the-loop
• Embabel
• Temporal

Whether you're a Java developer wiring up your first LLM tool call, a software architect weighing LangGraph4j against Spring AI or Embabel, or simply curious what a "controllable" AI agent actually requires, you're welcome to join.

See you in the live chat!

Embrace Your Java Journey

Dive into the pulsating heart of the Java community at JCON EUROPE 2027, the premier in-person Java community conference in Germany, lovingly organized by the Java User Group Oberpfalz in collaboration with JAVAPRO Magazine.

https://2027.europe.jcon.one/