>He loves being able to just dump in a bucket of seemingly unrelated data, and getting actionable BI from it.

Sorry, just don't trust accurate LLM data analysis even from the latest claude model. Have had some embarrassing bloopers over the last year. It is just better to code a deterministic tool to do the data analysis.

Even for the code. It will sometimes shit a very complex code for something with a very easy solution, and if you ask why it is so complex, it will invent some reason why. Always on 3 points, the first is true but have nothing to do with the subject, the second is plainly false, and the third is weird bullshit.

Don't get me wrong, I still use it, but the more you do, the more I see the obvious limitations.