First-time extraction
First-time extraction, or first-pass extraction, is when a document from a new supplier is extracted for the first time. Because the system has never seen documents from this supplier before, there is no history of user interaction to confirm uncertain field values. This can make initial interpretation less reliable and increase the chance that fields require manual verification.
To address these challenges, the Advanced extraction engine includes an option to use an AI-powered, pre-trained document library to support first-time extraction. This library has been trained on a wide variety of invoice formats and layouts, enabling it to recognize common header and line-item fields even when no prior learning exists for the supplier. By combining this general knowledge with the extraction capabilities of AP Essentials, first-time documents are interpreted more accurately, reducing the need for manual corrections and improving processing efficiency.
You enable the Advanced extraction engine and the Use pre-trained document library for first pass extraction setting in the Extraction settings. The first time you enable this feature, it can help improve the first-time extraction of header fields, with potentially greater improvements for line-item fields.
This feature is currently only available on a private private preview basis.
The pre-trained document library is designed to extract the following AP Essentials fields. If your document type includes fields not listed below, their field values remain empty and their status is set to Complete, unless a field setting or validation rule applies.
|
|
|
The pre-trained document library is available only for accounts payable processes. It does not apply to remittance advices or XML documents.