Fresh Servant - machine learning to boost production

Application development facilitates everyday work, and improves the flow of information

Fresh Servant, a provider of ultra-fresh food solutions for store shelves and professional kitchens, is guided by three core principles: the product it offers its customers must be tasty, beneficial to health and of high quality. These principles have helped the business grow consistently, right up to today’s digital management of data and effective knowledge management.

Fresh Servant’s collaboration with Pinja started in early 2020 with the development of a data warehouse and reporting and expanded relatively quickly into application development and machine learning and forecasting. The expansion from one area to the next of efficient data processing has been quite natural and logical, as all developments are ultimately interlinked.

“We’ve been keen to invest in development because automating routine operations improves speed and accuracy, among other things. At the same time, it brings more depth to the work,” says Carita Haapasalmi of Pegasso Oy, who leads Fresh Servant’s data and tool development.

The app development has helped Fresh Servant to move demand and production planning from Excel to mobile applications. The applications communicate with Fresh Servant’s ERP system and the underlying data warehouse, helping reduce manual work and diversify operations. The development work brought relief to previously difficult and cumbersome data management.

“Pinjans have clearly wanted to look at things from our point of view, and produced value-adding solutions. I also appreciate their solution-oriented approach, and the cooperation has broadened our views too,” says Juha Tiitto, Head of Design and Development at Fresh Servant.

One of the key benefits of the development work is that demand planning data can now be shared seamlessly across all internal processes. The applications also make practical work easier by enabling the acknowledgement of faults on the production lines using tablets. This allows us to collect valuable data on production line failures and faster processing of cause-and-effect relationships.

“Pinja’s implementations have been successful in all respects. The successful applications have given us a whole new range of opportunities to develop other activities more broadly,” Titto says.

Machine learning extends the time span of forecasts

The latest area of collaboration between Fresh Servant and Pinja is the use of machine learning to predict sales. Initially, Fresh Servant used people to do it, then mathematical formulae based on previous year’s sales figures. The development project introduced machine learning, and the forecasts introduced in fall 2022 have taken the forecasting work in a very promising direction. Demand forecasts already proved to be more accurate than those produced by hand and mathematical calculations during the testing phase.

“We are in the early stages of development, but we already see machine learning increasing the variety of demand forecasts, extending the time horizon of forecasting and deepening our understanding,” Tiitto says.

In a mutual machine learning development project between Fresh Servant and Pinja, different machine learning methods have been tested and iterated together to see which one works best for which customer group and product. Machine learning-based forecasts can minimize manual work, but also take information entered by people into account, such as campaigns to boost sales, where appropriate. The door is also open to the possibility of applying machine learning to other areas of business in the future.

The cooperation between Fresh Servant and Pinja, which started with data warehouses and reporting development, and gradually expanded to application development, machine learning and forecasting, has shown a natural progression. As a result, Fresh Servant’s key business data are used to comprehensively and in a variety of ways improve production efficiency, supply security and sales development – and ultimately to maintain and improve customer satisfaction.

“We sourced the whole package from Pinja, because they already had a good understanding of the food industry, and they know our data and processes well. It’s easy to trust their expertise. The solutions provided by Pinja also communicate with each other,” reducing our work, says Haapasalmi.