The overwhelming Python-centric AI ecosystem doesn't require Java developers to switch languages, as powerful Java-native tools exist for AI integration.
LangChain4j, inspired by Python's LangChain, provides a Java-native framework for integrating AI models, with Quarkus offering simplified integration features.
Begin by adding dependencies and using annotations like @AiService to define prompts, parameterize questions, and automatically map model responses to Java objects.
LangChain4j provides out-of-the-box chat memory to maintain conversational context, enabling follow-up questions and parallel conversations using a memory ID.
Use function calling, also known as tools, to allow the AI model to invoke your existing Java methods and services by describing them with the @Tool annotation.
The Multi-tool Calling Protocol (MCP) enables an AI model to autonomously decide which external tools to call in sequence to fulfill a user's request within a Java environment.
Protect against misuse like prompt injection by using input and output guardrails to sanitize requests and responses, ensuring the model behaves as intended.
Use Retrieval-Augmented Generation (RAG) to supplement the model's knowledge with your own documents by loading them into a vector store for relevant context retrieval.
A demonstration of a car rental chatbot showcases how to integrate a database, an external weather service via MCP, and custom documents via RAG to create a comprehensive AI assistant.
32:02Kevin Dubois
31:02Daniel Oh
30:01Mario Fusco
31:53Timo Salm
31:59Juarez Junior
30:38Alex Soto
32:13Cedric Clyburn
32:14Timo Salm & Sandra Ahlgrimm
Jobs that call for the skills explored in this talk.
