Data solution delivered in record time strengthens piecework processes

Two months. That is all it took from the initial meeting between EVU and Mjølner until a fully functional solution was rolled out.

“The short iteration made a huge difference. Mjølner put two creative and highly skilled specialists together with me, while I brought the business challenge. We started out with a completely different solution on the drawing board, but through dialogue we discovered where the real value was. That was pretty cool.”

– Arne Kirkeby-Thomsen, Chief Consultant at EVU

EVU is the Education Secretariat for the Electrical and Plumbing Industries and works, among other things, to ensure quality, education and shared frameworks in the electrical and plumbing sectors. An important part of this work concerns piecework times, which help create an objective and agreed foundation for electricians’ pay.

In Akkord+, electricians and companies can gain an overview of piecework, registrations and time standards. But behind a well-functioning piecework system lies an extensive and complex data task.

The challenge was that product data had previously been handled manually.

Until recently, EVU employees visited construction sites and used stopwatches to measure how long a specific installation task took. To ensure objectivity, each task was measured manually 75 times. These many field studies were extremely time-consuming and limited how many new materials could be time-measured. In short, scaling was impossible.

With a backlog of 130,000-140,000 materials — and only 21,000 materials measured — reaching the finish line was practically impossible.

For EVU, the problem was not only administrative. If new materials do not have a clear and accurate time standard, it can create uncertainty in day-to-day work. Electricians and employers risk lacking a shared reference point, which can ultimately lead to doubt, negotiations and potential disagreements.

“It is critical if we do not have the right time measurements. Valid product data forms the very foundation for agreements, workflows and trust between the industry’s parties. That is why Raadata+, with its high data quality, has made a significant difference,” explains Arne Kirkeby-Thomsen.

Raadata+ makes it possible to retrieve and structure product data directly from wholesalers’ product catalogues. Using an LLM, or Large Language Model, product data is processed, classified and structured. The LLM can recognise patterns and connections in product data and thereby help build a flexible data structure in which different product types can have different attributes. A socket outlet and a nut, for example, do not have the same characteristics, which means the data must be structured dynamically.

The project shows how an AI model such as an LLM can be used to sort, qualify and create structure in complex product data. EVU can quickly access the data needed to assess new materials, creating a stronger foundation for linking materials to correct time standards. This means the stopwatch can be retired, while the task is solved more systematically and scalably than before.

“Raadata+ is very much about data quality. If we are to succeed with AI and the digital solutions of the future, we first need a data foundation we can trust,” says Arne Kirkeby-Thomsen.

From previous projects, EVU has learned that work with AI quickly loses value if the data foundation is not good enough. Preparation is crucial. This learning has been carried into Raadata+, where data governance and data quality have been a focus from the start.

AI has not been a goal in itself, but the next natural step. Only when the data is good enough can AI be used responsibly and value-creatingly. The strong structuring of the project has meant that, after delays with a previous supplier, EVU has now reached the point it wanted to reach:

“The impressive pace of the project has meant that I have been able to move directly, without delay, to the AI-based product comparison that is central to our relevance and credibility in the industry,” says Arne Kirkeby-Thomsen.

Raadata+ is an example of how significant business value does not necessarily require a lengthy development project or a large budget.

The value of Raadata+ lies in the fact that EVU can now work with product data at a scale that was previously unrealistic. New materials can be assessed more quickly and on a more informed basis, and EVU is better equipped to keep pace with ongoing developments in the market.

For electricians, this means less uncertainty about new materials. For employers, it means a more consistent and reliable foundation. And for EVU, it means the organisation can maintain its neutral role and continue to deliver data that both sides of the industry can trust.

“This enables us to work with data at a completely different scale than before. That is crucial if we are to keep up with the development of materials while maintaining the credibility of our time standards,” says Arne Kirkeby-Thomsen.

Through a targeted digital effort, EVU has succeeded in lifting a heavy maintenance task while also paving the way for more advanced use of data and AI.

The collaboration between EVU and Mjølner built on an already strong partnership established through the development of Akkord+. With Raadata+, there was once again a need for technical understanding, business insight and the ability to quickly test ideas.

“I have valued Mjølner’s open-minded approach to finding new solutions. It turned out to be an offshoot of the project that truly created value. The rapid execution from prototype to production-ready product, which became fast and stable through just a few iterations, has resulted in a solution I am very satisfied with,” emphasises Arne Kirkeby-Thomsen.

For EVU, it was important that the solution was not made larger or more complex than necessary. The goal was to create concrete value quickly while also building a foundation that could be developed further.

This approach placed demands on the way the collaboration was structured. The project required an open and exploratory process in which ideas could be tested, the direction adjusted and the solution shaped according to the value it was intended to create.

Mjølner contributed technological guidance, development capacity and a pragmatic approach to turning a complex data challenge into a usable solution.

“I am very pleased with the collaboration with Mjølner. We have developed solutions together for several years, and Mjølner is a valuable technology partner that we greatly appreciate,” concludes Arne Kirkeby-Thomsen.