Our clients were not aware they were using TensorFlow, so that aspect was transparent. I think we personally chose TensorFlow because it provided us with more of the end-to-end package that you can use for all the steps regarding billing and our models. So basically data processing, training the model, evaluating the model, updating the model, deploying the model and all of these steps without having to change to a new environment.
