LangChain4j simplifies consuming AI model APIs for Java developers, avoiding the need for deep data science expertise.
Use the Quarkus CLI to bootstrap a new Java application and add the necessary LangChain4j dependency for OpenAI integration.
Define AI interactions using the @RegisterAIService annotation, system messages for context, and user messages with dynamic placeholders.
LangChain4j uses memory to retain context across multiple calls, with the @MemoryId annotation enabling parallel conversations.
Use the @Tool annotation to expose Java methods to the AI model, allowing it to execute business logic like sending an email.
A practical demonstration shows how to generate a haiku using a prompt and then use a custom tool to send it via email, verified with Mailpit.
Enhance LLM responses with your own business data by using an embedding store or Quarkus's simplified 'Easy RAG' feature.
A chatbot demo uses a terms of service document via RAG to correctly enforce a business rule for booking cancellations.
Run LLMs on your local machine with Podman AI Lab and make your application resilient to failures using SmallRye Fault Tolerance annotations.
A final demo shows an application calling a locally-run model and triggering a fallback mechanism when the model service is unavailable.
29:18Daniel Oh & Kevin Dubois
31:59Juarez Junior
32:13Cedric Clyburn
31:53Timo Salm
30:01Mario Fusco
31:02Daniel Oh
41:17Syed M Shaaf
32:14Timo Salm & Sandra Ahlgrimm
Jobs that call for the skills explored in this talk.
