Lately I've been experimenting the usage of LLMs for the explainability of SQL stored procedures with exceptionally good results - think about identifying important fields, linking them to existing glossaries, expanding said glossaries, etc. I wonder how much SQL looking like natural language helped for my use case.