The Importance of Scale in AI

In the early days back 2016 when you started, very few people were building AI models to do something with. But these days almost every company everywhere uses AI. Not everybody wants to run their own in-house operation and that seems where your second push at scale comes in where you provide the full service. I'm really curious when you think back. Originally you were not doing that. You were just focused on data quality, data annotation. What was the moment where you decided actually that's not enough?