I

think this is one of the good things about some of the data work that you do.

So, for example, let's say you are a data warehouse developer and you primarily work in SQL.

The surface area of what your AI has to do, I would argue, is pretty small compared to one-shotting GitHub, right? Correct.

And that's good, right? It's not hard to be like, oh, make sure you write some tests or make sure you do the lint command before you submit.

But equally, I think there are some areas which are a little bit harder to maybe put in.

So, for example... let's say you're a scale up that's grown very quickly and you never decided how you're actually going to model your data if you're doing like facts and dims or if you're doing like data fault or whatever like i feel like if you don't tell the ai exactly how you want that then like there are little things that you know might creep in like maybe you have like a naming convention where instead of one underscore you put in two underscores and suddenly you start writing all this code and everything works everything's perfect but then You know, for sanity's sake, half your columns now have one underscore instead of two underscores.

I don't know what your thoughts are.