Is there away to query your custom rug for context and data so that you can use LLM to answer based on retrieved data?