I'd like to have a better way to analyse how the knowledge base documents are used. Something like langsmith would be great. There's no way to see how the documents are embedded / retrieved, making the knowledge base pretty hit and miss imo. Correct me if I'm wrong pls