How VWC Group reduces document processing times for its clients by 87% in FM Module

The Challenge

VWC (Van Wendel Computers) Group is a Dutch-based company founded in 1992. Its portfolio spans a vast selection of products and services, from hardware such as computers, printing & scanning to office supplies to software solutions for various teams.

One such solution, FM Module, serves the needs of accounts payable (AP) teams. FM Module covers the processing of invoices and purchase orders, determining what is on the invoice, which is then processed in customer accounting software and gets deployed in cooperation with the customer’s finance teams.

Previously, the FM Module’s team maintained a set of templates for invoices with customized triggers or formulas (regular expression or regex) assigned to every field. While this system worked initially, it was not very scalable as the VWC’s customer base grew, as each trigger on separate fields needed to be updated and maintained manually. With the customer base doubling in size, so would double the need for work to maintain it.

The need became apparent for an invoice recognition OCR that would be scalable and could extract data from the new, previously unseen documents. 

The Solution

Rossum AI was integrated into the Microsoft Azure cloud that runs the entire process on the backend, with the FM Module team handling the validation and review without having its clients involved. Rossum deployed its AI engine to plug into an existing FM Module’s UI and overall document processing workflow. The document is received and stored in the Microsoft Azure cloud, after which the Rossum API is called to perform data matching and business checks. 

The team was happy with the quality of data recognition right out of the box. Still, to continuously improve the quality of AI output, the FM Module team performs validation in the Rossum UI in case human interference is needed at any point in the process, smoothly running the entire process on behalf of the customer in the backend. 

Customers also have an option to flag a specific document for human validation, which tends to mainly be performed for new document layouts and formats to ensure that the AI learns quickly and gets things right from the start with high confidence. At the end of the process, customers store the final processed invoices in their downstream bookkeeping solution of choice.

The Impact

VWC’s FM module is growing by 30-35% in revenue in 2022, with more and more customers switching to its Rossum-supported software for a scalable, flexible setup and smooth UX. What used to be a half-time job of maintaining rule-based formulas and regular expressions is now reduced to less than 1 hour per day spent by FM Module’s support engineering team on document processing. At the same time, the number of customers has since more than doubled. The initial goal was to reduce each invoice processing time to under one minute.  Now, the team is only spending 9 seconds on average per document due to the combined powers of AI, automation, and Rossum’s agile response times.