LangChain4j for Beginners

5:00 du. - 6:00 du. (UTC)

Move from "text generator" to "action taker." You'll expose Java methods as tools using @Tool annotations and watch the AI chain them automatically with the ReAct pattern. Then you'll explore Model Context Protocol (MCP)—an open standard for AI-to-tool communication—and build a Supervisor Agent that dynamically orchestrates sub-agents to read files, analyze content, and summarize results

4:00 du. - 5:00 du. (UTC)

Learn how to build safe, reliable, and enterprise-ready AI applications in Java, including how to protect API keys and model endpoints, validate tool output, enforce content filters, and keep LLMs from stepping outside their intended boundaries. See how to design prompts defensively, restrict system capabilities, and use structured interfaces to avoid injection attacks. We'll also explore patterns for safe RAG, secure memory, and audit-ready logging that make your Langchain4j applications trustworthy, enterprise-grade, and ready for real users.Learn how to build safe, reliable, and enterprise-ready AI applications in Java, including how to protect API keys and model endpoints, validate tool output, enforce content filters, and keep LLMs from stepping outside their intended boundaries. See how to design prompts defensively, restrict system capabilities, and use structured interfaces to avoid injection attacks. We'll also explore patterns for safe RAG, secure memory, and audit-ready logging that make your Langchain4j applications trustworthy, enterprise-grade, and ready for real users.

4:00 du. - 5:00 du. (UTC)

In this BONUS finale, we graduate from single agents to full multi-agent systems. We'll explore 8 patterns that power production AI: chain agents like an assembly line, fan-out for parallel expert opinions, loop until a critic approves, and route requests to the right specialist. Then unlock Supervisor agents that delegate like project managers and Human-in-the-Loop gates when humans need the final say. Finally, explore goal-oriented planners that find optimal paths and peer-to-peer meshes where agents collaborate as equals without a boss.