Can traditional RAG solve modern Knowledge problems?