@shanesemler most of the scrutiny came from the training data and power consumption, not how proprietary it was or not being able to run it locally.