I believe their moat is that they have early access to OpenAIs newer models as they come out so they are always a bit ahead. Combined with some big partnerships

I believe their moat is that they have early access to OpenAIs newer models as they come out so they are always a bit ahead. Combined with some big partnerships they have already landed perhaps, which could yield highly valuable proprietary datasets to train v2 of whatever they are doing.