Had a question come up recently and not sure how to answer. Has anyone had this issue before?
In using the AP Document Recognition, we have experienced Acumatica bypassing the needed invoice on page 1 and instead grabbing header data from page 2 of the pdf document which is usually a supporting sub-invoice to the actual invoice that should be addressed on page 1. This does not seem to improve with machine learning after several invoices from the same vendor. We are using version Acumatica 23.111.0019. Have your customers reported this and do you know if it has been improved in subsequent Acumatica versions.