When to Productionize a Model Too Late?

The vendor lock-in concerns is often a huge one we see as well with teams. I don't think it's an difficult concern necessarily at the start up end, but definitely at the print enterprise end. We tend to have three kind of three to four distinct stages when we start to productionize a model. The first thing is to put value next thing is to like to deliver the first iteration to make sure it's good. And then it's probably building out some of those more rigorous controls around it and going that.