Can you talk more about the architecture, which model are you using, if it just worked or not, etc.
The OCR model part does not support everything, just some apps but looking to improve it.
