Traditional OCR can require many heavy, manual processes. Palsgaard experienced this as well. Before implementing Lasernet Input Management, their accounting team handled supplier invoices manually.

Documents were circulated as copies with a number and matched once the purchase had been confirmed. Payment was then processed manually based on an invoice.

Today, the situation is completely different. Palsgaard uses Lasernet AI Input to process supplier invoices across multiple companies and countries, enabling them to double their revenue with the same accounting team. The solution has been implemented in the Netherlands and across 12 finance companies in Denmark and is a central part of the company’s financial processes. Invoices are automatically read and processed:

“The vast majority of supplier invoices go through without any issues – and users do not have to do anything. The accounting team spends significantly less time on manual mapping and has freed up time for other tasks, as the system automatically recognizes and processes the invoices” says Torben Strand Johansen.

One of the solution’s key strengths is that both the setup and ongoing maintenance can be handled without extensive technical knowledge. The AI training itself requires only limited effort and can be carried out by superusers, Torben explains:

“Tabellae’s consultants handled the initial setup, which took a couple of days. Afterwards, your consultant walked me through what the solution can do and how she had trained the AI model. I then created an AI model for each country and language – that took just one day. This keeps the AI models separate, and those responsible for the companies in each country can train their AI model without affecting anyone else’s.”

The solution can also be continuously fine-tuned without extensive development work:

“I spent about a week configuring the basic elements. Once the solution was live, I received the invoices that failed due to too much or too little data. I have also configured it to identify which bank account it is in our Dynamics 365, or whether it is an EAN or SWIFT number. It checks this every single time an invoice comes in. If the information differs between invoices, it stops the invoice in Lasernet Client and notifies us that something has changed since the previous invoice,” Torben continues.

“If you have no previous experience with Lasernet, there is of course a learning curve. But when it comes to training the AI model, superusers can easily map the required elements,” Torben explains.

Claus Christensen, Global Enterprise Architect, continues: “If you have a finance department without a developer like Torben, I still think it is possible. It is more a question of what resources you have in place to handle invoices that may fail, unless you outsource this directly to Tabellae.”

Fast Go-Live Without Downtime

Once the solution was ready, the transition from the previous input solution to Lasernet AI Input was quick and straightforward – without extended periods of parallel operation:

“We basically just shut down one solution and launched the other. The implementation took only a few days and a couple of meetings, and then we were up and running. Users could quickly continue their work in the new solution without delays. However, you need to make sure you have resources available to configure anything that does not work, or outsource it to Tabellae. Some suppliers may have unusual invoices that will never process automatically and therefore need to be handled manually,” Torben explains.

Claus agrees and adds that the rest of the organization quickly came on board: “Looking at the user interface, it has been easy to get our users to switch to this solution – even though the user interfaces are different.”

After implementation, the benefits quickly became apparent. Automation reduced the manual workload while providing the company with a more consistent and global setup for invoice processing.

“I see a clear business case compared to how we worked before. We have now moved to a solution where the underlying ‘engine’ is developed by Microsoft, and I believe that is the future. In addition, we now have a consistent way of receiving invoices globally, while the workload is also distributed more effectively across the organization.

The accounting team saves time, and the solution also enables us to work much more globally. If an invoice comes in during the evening, an employee in another time zone can, in principle, complete the processing because very little manual work remains. We can distribute tasks more flexibly across the organization and move work to where it makes the most sense – for example, if one part of the organization is under pressure, another can take over, regardless of whether the invoices come from the Netherlands, Denmark, or another finance company,” Claus continues.

• 95.5% recognition on the first pass
• Less manual work for the accounting team
• Easy global rollout
• Fast implementation without operational disruptions
• Future-proof AI-based solution powered by Microsoft technology