This project provides a comparative playground for Java AI libraries, demonstrating Retrieval Augmented Generation (RAG) and tool-use capabilities for customer support applications. It targets Java developers interested in integrating LLMs into their applications, offering a practical comparison of popular frameworks like LangChain4j, Quarkus, Spring AI, and Semantic Kernel.

How It Works

The application simulates a customer support agent that can access terms and conditions via RAG and execute Java methods (tools) using an LLM. It showcases different backend implementations (LangChain4j, Quarkus, Spring AI, Semantic Kernel) with a Vaadin Hilla/Spring Boot frontend.

Quick Start & Requirements

• Run via IDE (Application.java) or mvn command.
• Requires Java 17+ and an OpenAI API key set in the OPENAI_API_KEY environment variable.
• Links: LangChain4j Customer Support Agent example

Highlighted Details

• Comparative implementations of LangChain4j, Quarkus, Spring AI, and Semantic Kernel.
• Demonstrates Retrieval Augmented Generation (RAG) for accessing documents.
• Features tool-use capabilities, allowing LLMs to call Java methods.
• UI built with Vaadin Hilla, backend with Spring Boot.

Maintenance & Community

The project acknowledges contributions from the LangChain4j, Quarkus, Spring AI, and Microsoft teams.

Licensing & Compatibility

The repository does not explicitly state a license.

Limitations & Caveats

The project is presented as a demo/playground, implying it may not be production-ready. It relies on an external OpenAI API key.