"Previously, when using our old OCR solution, every time a new invoice came in, the team had to spend 5-10 minutes mapping it. After mapping, they generally wouldn't need to do it again, unless the vendor changed the invoice layout or design"
Industry
Software Solutions
HQ Location
Stockholm, Sweden
Formpipe simplifies complex invoicing workflow with AI OCR scanning
Previously, whenever they received a new invoice, the Formpipe finance team would spend 10 minutes OCR mapping invoice fields to their AP software. Recently, however, the team realised they were wasting valuable time each day mapping invoice fields manually. On top of that, manual OCR mapping errors led to instances of delayed payments and incorrect archiving – problems that ate into the team’s efficiency. Thankfully, they didn’t have to look too far for the solution as it was right in front of them: Lasernet.
Lasernet, Formpipe’s document management platform, makes managing complex documents simple. The beauty of Lasernet is that it’s designed for the way you work. Instead of having to adapt to the software’s way of doing things, it’s completely flexible to your needs. At Formpipe, we practice what we preach. That's why our finance team uses Lasernet to optimise invoice management: significantly reducing errors and cutting down the time each invoices takes to process.
However, the team discovered there was even more that could be done to simplify the invoice management workflow. By integrating Lasernet with Azure AI’s Document Intelligence, an AI-powered OCR solution, the team found their workflow became much simpler.
Introducing automatic AI OCR scanning
Document Intelligence takes care of mapping documents automatically, significantly smoothing out the team’s invoicing process. The good news is that the AI OCR scanning process is delightfully simple to set up and run. Users can choose from a wide variety of prebuilt models or custom AI OCR scanning modules, allowing for ultimate flexibility.
Custom models give users more control, allowing them to label and build a model that extracts a specific schema from documents such as invoices. Stefan Kristiansen, Business Solutions Consultant at Formpipe, explained why he generally prefers custom extraction models. “For most cases, I prefer custom extraction model because we can actually set up our own specific fields we need it to find every time,” he said. This level of precise control ensures Lasernet can confidently extract the information they need from any invoice format.
Training Document Intelligence’s mapping AI tool
Document Intelligence’s AI OCR mapping tool requires some initial training, but don’t worry, it doesn’t take long.
Here’s how it works: After uploading a new invoice type to Document Intelligence Studio, the user spends a few minutes labelling the data in the document. This trains the software to recognise distinct types of information, such as the company name, customer name and address, invoice number, invoice date and more.
Training the software for a new invoice type is well worth the time investment. “A good idea would be training between 20 and 50 documents that you know you’ll be receiving, and then from there on it should have enough training so the solution is ready to run,” Stefan said. Once you’ve trained Document Intelligence on one type of document, the user won’t have to do it again.
Once Document Intelligence has mapped the data, it’s transferred to Lasernet’s document design engine. From there, users can create a variety of documents using the data from the invoice, such as PDFs, Word documents, Excel spreadsheets, JSON files, XML files and more. There’s also the option to create PDF files with the metadata attached — perfect for uploading to an archiving solution. In the case of the finance team, after automatic scanning, the documents are sent to Rillion Prime, an AP automation tool.
In one simple workflow, the finance team can receive an invoice in any file format, automatically OCR scan the document using AI, process the data, choose a suitable output and then archive the document. A once complex process made simple with real ROI from day one.
A game-changing process improvement
Stefan explained just how transformational Azure’s AI OCR capabilities have been for the invoicing process:
“Previously, when using our old OCR solution, every time a new invoice came in, the team had to spend 5-10 minutes mapping it,” he said. “After mapping, they generally wouldn't need to do it again, unless the vendor changed the invoice layout or design, in which case they would have to remap it.”
Thanks to automatic AI OCR document mapping, the finance team are saving 10 minutes of manual work for every new type of invoice they receive. That doesn’t sound like much on the face of it, but it saves more time than you think. In this use case, it saves 3.7 hours a month, which adds up to just over 44 hours a year — valuable time that can be much better spent elsewhere. Additionally, the time savings extend beyond manually mapping invoices. It also allows for new finance team members to have a more straightforward onboarding experience, as they no longer need extensive OCR mapping training. It’s these kinds of knock-on benefits that can really boost the team’s productivity, making day-to-day operations smoother for everyone involved.
