Invoice processing · Production

SaltPay uses Nanonets AI to integrate invoice processing with SAP ERP

The problem

SaltPay was transitioning to SAP ERP and their existing document processing provider did not support SAP, leaving them facing the prospect of manually managing thousands of invoices.

First attempt

SaltPay's existing document processing provider did not support SAP's system, forcing them to find a replacement.

Workflow diagram · grounded in source
1
Invoice sent to Nanonets
trigger
“Once Saltpay sends an invoice to Nanonets”
2
AI extracts invoice fields
ai_action
“Nanonets provided an AI tool to capture information from invoices and upload that information directly into SAP. SaltPay trained the AI to extract specific information from each invoice”
3
Business logic validation
validation
“Nanonets automatically validated certain number of conditions based on the business logic provided by Saltpay”
4
Team member approves extraction
human_review
“a SaltPay team member manually approves the data extraction in Nanonets”
5
Upload approved data to SAP
integration
“which uploads the data directly into SAP”
6
Reviewer approves fields in SAP
human_review
“The reviewer then manually approves the fields in SAP”
7
Ongoing model training
feedback_loop
“the SaltPay team is currently working on increasing accuracy by continuing to train Nanonets platform with new invoices”
Reported outcome

Nanonets AI correctly populates 75% of the invoice fields needed in SAP, with a two-pronged human approval workflow giving SaltPay confidence their system correctly stores information, and ongoing training continuing to improve accuracy.

Reported metrics
invoice fields correctly populated by AI75%
Confidence in correct data storageconfidence that their system correctly stores the information
Reported stack
NanonetsSAP
Source
https://nanonets.com/customer-success-story/saltpay-uses-nanonets-to-integrate-sap-to-manage-vendors
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Nanonets AI correctly populates 75% of the invoice fields needed in SAP, with a two-pronged human approval workflow giving SaltPay confidence their system correctly stores information, and ongoing training continuing…

What tools did this team use?

Nanonets, SAP.

What results were reported?

invoice fields correctly populated by AI: 75%; Confidence in correct data storage: confidence that their system correctly stores the information (source-reported, not independently verified).

What failed first in this deployment?

SaltPay's existing document processing provider did not support SAP's system, forcing them to find a replacement.

How is this invoice processing AI workflow structured?

Invoice sent to Nanonets → AI extracts invoice fields → Business logic validation → Team member approves extraction → Upload approved data to SAP → Reviewer approves fields in SAP → Ongoing model training.

Related invoice processing cases