This process might be automated by ingesting sample metadata into an LLM and getting it extract enriched metadata. We expect this features to