The two most common approaches to incorporate specific data in a LLM-based application are via Retrieval-Augmented Generation (RAG) and LLM fine-tuning.