I wonder the same thing. Code is pretty open ended though so I wonder if it’s not the best example. On the one hand it would definitely be useful for something

I wonder the same thing. Code is pretty open ended though so I wonder if it’s not the best example. On the one hand it would definitely be useful for something like classifying support requests into priority. But would it be worth it to just train your own model? I guess one advantage is you could give it some well informed guidelines without training something on lots of data. There is probably a better example between discrete labeling and code though.