Jakarta EE 11 meets AI: Building intelligent microservices with virtual threads and Jakarta Data

This talk, titled "Jakarta EE 11 meets AI: Building intelligent microservices with virtual threads and Jakarta Data," presented by Luqman Saeed from Payara, focuses on the development of AI-powered microservices using standard Jakarta EE components without adding complex external dependencies. The session highlights how to seamlessly integrate large language models into existing enterprise Java architectures, utilizing Jakarta Data repositories for persistence and storing embeddings in standard JPA entities.

The architecture separates request handling, model interaction, and data access, employing retrieval-augmented generation through vector search results. Concurrency is managed with virtual threads via Jakarta Concurrency 3.1, allowing parallelism for I/O-bound tasks like embedding generation. The discussion concludes with practical considerations for scaling AI microservices, addressing when to incorporate vector databases, GPU-based inference, or distributed architectures, while also comparing declarative and programmatic approaches to AI integration.