Case Hedengren

In the solution proposed by Digia, order data is extracted from PDF documents or email messages using robotic automation. The robot identifies the ordering company and selects the appropriate AI prompt. Order details are then processed using Azure OpenAI capabilities. The automation is configured to recognize specific information from each order, interpreting and extracting the required data through a large language model. It then transfers the data into the Digia Envision ERP system, creates the order there, and reports the details in Excel.

This solution was designed specifically for Hedengren and is unique, yet it is built on a scalable and customizable implementation widely used at Digia. It is a standard solution that can be adapted to a wide range of customer processes.

Nieminen estimates that the solution has already saved approximately 1.5 full-time equivalents in its early stages. Currently, 40% of orders are processed through automation, with a target of reaching 80%.

“It’s already a significant saving, even though we haven’t fully completed the solution yet. We calculated the savings, and they were substantial. It was definitely worth starting."

At present, around 200 email orders are received weekly. With 40% handled by the automation, this means over 3,000 orders per year no longer require manual entry. If the 80% target is achieved, the annual number of automated orders will exceed 6,000, doubling the savings in work hours. The reason the target is still ahead lies in minor challenges related to Hedengren’s data.

“Identifying the customer has been surprisingly easy, even though we have the same customers across different locations. The biggest challenge is recognizing item codes. An order may include the product code in several places, which we use for identification. However, customers might order using a slightly different code or omit a character, like a dot, making it harder for the system to recognize,” Nieminen explains.

Hedengren plans to provide clearer instructions to customers on how to place orders to ensure the automation works more efficiently. Currently, orders arrive in the email inbox of the order processor, who forwards them to the automation if they are in a format the automation can read. The goal is for customer orders to go directly to the automation without intermediate steps.

“First, we want to resolve these challenges, and once it works flawlessly, we can start considering further development ideas,” Nieminen adds.

At Hedengren, automation could also be applied to the quotation process, enabling it to recognize when a quote includes a discounted price instead of the standard price. Eventually, it might even be possible for the automation to generate the quote itself.

The implementation of the order processing automation progressed efficiently: from the kickoff meeting in February 2024 and the signing of the agreement in May, things moved smoothly toward readiness before the summer holidays, when testing began. Marit Nieminen explains that the solution was tested very quickly while it was already in use. By September, the solution had successfully gone into production.

“The collaboration went very well. The process was fast, and communication was excellent. We had regular weekly meetings where we could follow the progress. We got exactly what we ordered, so I can recommend Digia for this kind of partnership,” says Marit Nieminen.

For Digia, leveraging AI to streamline business processes is seen as a critical driver of competitive advantage and a key differentiator for future success.

“With the Digia Dolphin service concept, organizations can identify strategic use cases for AI, implement them efficiently, and achieve significant business benefits. AI is already part of everyday operations in automation, delivering major advantages in routine tasks—whether it’s interpreting orders, advising customers, or preparing quotes,” summarizes Olli Ainasvuori, Vice President of Business Automation Services at Digia.