Accounts payable · Production

Financial services provider in Paris automates invoice processing with Nanonets AI

The problem

A Paris-based financial services provider processing over 20,000 invoices per month across hundreds of different formats could no longer scale manual processing. Invoice images contained junk data and multiple invoices per page, and attempts to standardize invoice formats caused friction in vendor onboarding.

First attempt

Prior OCR solutions including Amazon Textract and Docparser were tried but encountered either accuracy issues or a lack of flexibility in handling new document formats.

Workflow diagram · grounded in source
1
Object Detection
ai_action
“Object Detection will detect the presence of 3 invoices in the image below. The object detection model has been re-trained with the clients images and has an accuracy score of ~96%.”
2
Image Classification
ai_action
“Image Classification will identify the different types of documents as present in the image below. The image classification model has been re-trained on the clients documents and has an accuracy score of ~98%.”
3
OCR Data Extraction
ai_action
“Nanonets OCR AI will capture the required data fields from each document, based on the type of document as detected in the classification model above.”
4
Fail-safe Validation
validation
“Additional checks are set up at each step of the workflow, as a fail-safe mechanism to ensure highest veracity before data is exported to their internal systems.”
5
Export to Internal Systems
integration
“before data is exported to their internal systems”
Reported outcome

Nanonets' three-step AI workflow achieved an accuracy score of ~96% for object detection and ~98% for image classification, with fail-safe validation checks at each step before exporting data to internal systems, enabling customers to reduce the cost and turnaround time of accounts payable processing.

Reported metrics
Object detection accuracy~96%
Image classification accuracy~98%
Monthly invoice volumemore than 20,000
cost of AP processingreduce the cost of Accounts Payable processing
Show all 5 reported metrics
object detection accuracy~96%
image classification accuracy~98%
monthly invoice volumemore than 20,000
cost of AP processingreduce the cost of Accounts Payable processing
payment turnaround timeturnaround time of payments
Reported stack
NanonetsNanonets OCR AIAmazon TextractDocparser
Source
https://nanonets.com/customer-success-story/financial-services-provider-france
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Nanonets' three-step AI workflow achieved an accuracy score of ~96% for object detection and ~98% for image classification, with fail-safe validation checks at each step before exporting data to internal systems, enab…

What tools did this team use?

Nanonets, Nanonets OCR AI, Amazon Textract, Docparser.

What results were reported?

Object detection accuracy: ~96%; Image classification accuracy: ~98%; Monthly invoice volume: more than 20,000; cost of AP processing: reduce the cost of Accounts Payable processing (source-reported, not independently verified).

What failed first in this deployment?

Prior OCR solutions including Amazon Textract and Docparser were tried but encountered either accuracy issues or a lack of flexibility in handling new document formats.

How is this accounts payable AI workflow structured?

Object Detection → Image Classification → OCR Data Extraction → Fail-safe Validation → Export to Internal Systems.

Related accounts payable cases