AI models

This section outlines how AI models are used to identify key invoice attributes such as company, vendor, and add-on charge codes during the DII process. These models analyze extracted data from invoice documents and match them against configurations to automate and improve accuracy in invoice processing.

Model	Input to AI	AI model process	Output from AI	Notes
Company identification	Extracted Bill-To Address, Company Name, Finance Enterprise Group	Compares Bill-To Address and Company Name against process-level address and Payables company name. If multiple process levels share the same address, the process level name is included to refine similarity.	Company ID, Process Level, Vendor Group	Returns the company with the highest confidence score. If multiple results have the same score, the first one is selected. Minimum score: 70%; otherwise, returns No Similarity.
Vendor identification	Extracted Remit-To Address (or Vendor Address), Vendor Name, Vendor Group	If Remit-To Address is available, compares it with Vendor Location and Name. If not, compares Vendor Address and Name with current Vendor data in FSM. If no address is available, compares based on Vendor Name similarity.	Vendor ID, Remit-To Code (if available), Vendor Group	Returns the vendor with the highest confidence score. If multiple results have the same score, the first one is selected. Minimum score: 70%; otherwise, returns No Similarity.
Add-on charge code identification	Extracted Add-On Charge Description, Identified Company	Compares the extracted description with the Search Description field in the Add-On Charge business class in FSM, filtered by company.	Company ID, Add-On Charge Code, Add-On Charge Search Description	Returns the code based on similarity. No confidence score is applied. If multiple results match, the first one is selected. If no match is found, returns No Similarity.