LlamaTalks

LlamaTalks is a Spring Boot-based chat application that leverages the power of LangChain4j and Ollama to provide an advanced conversational AI experience. The core of the project is its Retrieval-Augmented Generation (RAG) capability, which allows the AI to answer questions based on ingested documents, making the responses context-aware and significantly reducing "hallucinations".

The application supports real-time streaming of responses, persistent conversation history, and a modular architecture that makes it easy to integrate with various frontends or other services.

Key Features

• Conversational AI: Powered by Ollama and LangChain4j for state-of-the-art interaction.
• Retrieval-Augmented Generation (RAG): Ingest and query your documents to provide context-aware, accurate responses.
• Streaming Responses: Real-time message streaming using Server-Sent Events (SSE).
• Persistent History: Stores all messages and conversations in a PostgreSQL database.
• RESTful API: A clean API for easy integration with any frontend or service.

Architecture

The application is built on a modern Java stack, designed for scalability and maintainability.

+-------------------+     +-------------------+     +-------------------+
|   Frontend/App    | <-> |   LlamaTalks API  | <-> |   Ollama Server   |
+-------------------+     +-------------------+     +-------------------+
                                    |
                                    v
                          +-------------------+
                          |    Vector Store   |
                          +-------------------+