But of course, this is a prototype that runs on my beer money budget. Deploying this at scale, and fine-tuning will require an investment, as well as an input from a real trained classicists, and ideally a behavior psychologist, because I want this to be a helpful application.