Practical AI for Analysts.

I use AI for the parts of analysis that used to eat the most time: drafting SQL and Python, trying an unfamiliar method, generating competing explanations and poking holes in my first read. It has made me faster, and it has also made it easier to be confidently wrong, because a polished answer can hide a bad join or the wrong denominator. The posts below split the problem into whether the output is technically right and whether the conclusion makes sense for the business, which is the part I don't hand over.