Agentic AI with Quarkus, LangChain4j and vLLM

Explore the emerging patterns of Agentic AI systems in this 48-minute Devoxx talk by Mario Fusco and Daniele Zonca. Gain a theoretical overview of AI agents and learn about the two main architectural categories: workflows with predefined code paths orchestrating LLMs and tools, and agents where LLMs dynamically control their own processes and tool usage. Understand the security and safety challenges of implementing Agentic Systems in mission-critical scenarios, including the importance of on-premise model serving and guardrailing techniques. Through practical examples, discover how to implement these patterns using Quarkus with its LangChain4j extension on premise, leveraging vLLM and Kubernetes for deployment.