This blog looks at the create model and train model tasks in SAC Predictive and compares them to a typical Python workflow. It breaks the tasks out into some of the key components and describes how each approach goes about it.
The obvious difference between the two approaches is the trade-off between speed and flexibility; SAC Predictive requires less knowledge and less time to create so is faster, Python is more flexible because you control much more of the generation. Hence particularly the Python approach is just one way of many possibilities.
You can download the PDF here.