What you're describing sounds a lot like traditional training of an ML model combined with descriptive+prescriptive analytics. What value do LLMs bring to this use case?
Ability for normal people to set up reasoning chains.
Ability for normal people to set up reasoning chains.