Today, large language models are becoming an integral part of almost every IT solution. However, their use is often accompanied by certain limitations, such as the relevance of information or its depth and specificity. One of the ways to overcome these limitations is the method of working with LLMs - RAG (Retrieval Augmented Generation).