What kind of success rates are you seeing with Oracle Fusion Document IO vs. IDR

Summary:

We're currently evaluating document IO processing solutions to streamline our invoice ingestion and want to get some real-world feedback from the community. Specifically, I am trying to compare the success and extraction accuracy rates of Document IO against Oracle Fusion Intelligent Document Recognition (IDR). I would love to know:

• What kind of straight-through processing (STP) or data extraction accuracy rates are you achieving with each for touchless processing?
• How well does Document AI handle complex, line-item heavy, or poorly scanned supplier invoices compared to Oracle Fusion IDR?
• How do the two compare regarding complexity machine learning training time, and the volume of samples required to get reliable results?