This article discusses setting up a self-hosted AI chatbot using Docker Model Runner and Couchbase Capella. The chatbot leverages the Llama 3.2 model for inference, storing conversation history in Couchbase Capella, and retrieving previous chats. By running the chatbot locally on your machine, you retain control over prompt customization, metadata storage, and fine-tuning behavior, all while ensuring faster inference and scalability. This setup prioritizes user privacy, allowing businesses to build dynamic AI applications without relying on external APIs. The chatbot's primary interface is a command-line interface, where users can send messages, view past chats, or exit the application. With this combination of Docker Model Runner and Couchbase Capella, developers can create efficient, scalable, and fully controlled AI applications that prioritize user privacy.