case

eVerbinding offers e-invoicing platform with AI-driven document conversion solution

Case

Share on linkedin
Share on facebook
Share on twitter
Share on whatsapp
Share on email

About eVerbinding

Connect (eVerbinding) is a Dutch Billing Service Provider, market leader for the Dutch market, with an international ambition. The company’s mission is to provide a fully automated administration service for everyone, making the administration process entirely  machine-to-machine. eVerbinding is on a trajectory to build a network of over 300 financial software providers all connected to the eVerbinding hub. The Peppol Access Point of this hub is connected to the international Peppol network and all other billing service providers connected to this network, allowing a true paperless (machine-2-machine) procurement process. Major Dutch financial software vendors all have a direct integration to eVerbinding.

Challenge

Around 2012, eVerbinding had the vision to build a network of over 300 financial software providers all connected to the eVerbinding hub. Its mission was to solve the following challenges for the Dutch small- & medium sized businesses:
  • Invoices are received through multiple channels (email, paper, fax)
  • Invoices need manual action to get it into the accounting software for processing
  • Lack of invoice governance
  • Compliance requirements from government to eliminate malpractice
  • Increasing pressure from CSR (Corporate Social Responsibility) and commercial necessity
The platform should fully automatically take care of the processing of all invoices and other financial documents.

Solution

eVerbinding has chosen Collabrr as an e-invoicing platform for enterprises to send and receive invoices to their business partners. Major financial software vendors like AFAS, Exact, BaseNet, Visma, Muis and Snelstart have a direct integration to eVerbinding. eVerbinding, part of Lansigt Accountants, has chosen Vanenburg to develop and manage the eVerbinding software and services, based on its B2B collaboration platform Collabrr. To this end, almost 2 million invoices are received from hundreds of buyers and sellers as pdfs or scanned copies on a yearly basis. In addition, Vanenburg built a solution allowing for automated processing of the invoices. Incoming invoices (pdfs and scanned images) are received via API. Leveraging Google Cloud Vision API for text extraction from pdf and images, these documents are subsequently classified based on their type (e.g. invoice, receipt) through Machine Learning based classification algorithms. The invoice entities from the invoice documents are recognized and extracted into XML. A human feedback loop is incorporated into the automatic conversion workflow to verify and provide feedback to the ML service, on the extracted invoice data. In this project we benefit from the strong capabilities of Tensorflow machine learning model development, and Google Kubernetes Engine and Kubeflow for the training and deployment of the Machine Learning models. Finally, the validated invoice entities are updated into the backend accounting systems and ERP systems.

Interested?

Are you wondering how we can modernize your enterprise IT and build your intelligent enterprise application? Contact us today!

Do you have any questions or do you need professional advice? We’re happy to help you!

Contact form

Subscribe to our newsletter to stay up-to-date on what’s new at Vanenburg