The ability of Hazy to generate hundreds of thousands, if not millions, of "random" users/accounts with unique attributes is really impressive! I'm able to quickly create models / datasets without having to worry about collecting or cleaning any external data. It helps me save time by creating my own dataset. There are some limitations when it comes to how you can set up your environment. There doesn't seem like there's an easy way to do this unless you're willing to spend hours figuring out what goes where. If you want to get into building predictive insights from large amounts of unlabeled (or semi-labeled) text then using Hazy will definitely be worth trying. Being able to build custom datasets and train models quickly and easily.