It takes effort to integrate large language models with RAG or vector databases in an application rather than just slapping a chatbot on