First Lieutenant Milo Minderbinder buys eggs from Malta for 7 cents apiece, sells them for 5 cents apiece in Pianosa, in Italy – and contrary to clear mathematics, makes a significant profit from the sales. As everyone who has read Joseph Heller knows, the plot of the case is not to stare at a single trade, but to understand the whole, and not to try to make too much profit: even the loss of individual transactions is acceptable if they enable the overall business to make a profit.
The same understanding of the overall business picture is required in the digitalization of processes and in the development of AI solutions that support business. Partial optimization, such as the pursuit of maximum margin at all stages of operation, can be very deceptive and detrimental to the whole.
In terms of understanding the whole picture, it is important to understand what the top-level objectives of digitalization are. Often when talking about the benefits of digitalization and artificial intelligence, the replacement of manual human work with automated solutions is highlighted: too often the sought-after benefits seem to be crystallized in the reduction of work costs (i.e. labor) (see also Lagstedt, Kauppinen & Lindstedt 2021).
If digitalization and AI solutions are primarily used to reduce costs, there is a risk that companies will fall into a vicious circle of shrinkage. If cost reduction is done by haggling over investments and reducing staff, the company’s mental capacity decreases. As a result, business agility also often decreases. At worst, taking the “slack off” from one process is harmful to the operating conditions of another process.
Shrinkage easily becomes a spiral. For a while after the efficiency measures, the company’s competitiveness improves, but when competitors find more agile and innovative ways to implement the same product or service, it is necessary to take new efficiency measures in order to stay on the market. As investments and innovations fade (and key players move to more innovative companies), the only way to do well in competition is to lower the price. This, in turn, forces to cut back even more and more. In this case, the latter has been chosen from the ‘grow or die’ phrase, and the digitalization of individual processes will only, at worst, exacerbate the direction of development.
If you think about digitization in big picture, the primary purpose should not be to replace people with apps, but precisely to streamline operations. Successful digitalization can even lead to an increase in the number of employees. Of course, then, to be a successful digitalization, labor productivity has increased even more significantly.
A healthier approach to reducing costs is therefore an action aimed for growth and renewal. In practice, better utilization of existing resources and improvement and development of services and products based on accumulated data (product, production or customer data).
Digitization of operations and artificial intelligence solutions enable progress along either of the aforementioned routes. Digitalization, and especially artificial intelligence, are powerful tools, but it is important to think how to use the tool as usefully as possible.
In the AI-TIE project’s artificial intelligence accelerators, which were held in autumn 2021, the companies Teca Oy and Picosun Oy end up with very different solutions, both of which clearly support the companies’ higher-level goals and enable new kind of growth for companies. Teca Oy is one of Finland’s leading suppliers of industrial products, solutions, and services, one of its fields of activity is dust removal solutions. Picosun Oy is an operator specializing in atomic layer growing, which delivers solutions for the needs of the pharmaceutical industry, among other things.
In order for Teca to provide companies with reliable dust removal solutions, the maintenance of the systems supplied, and the replacement of the filters should always take place shortly before the power of dust extraction begins to decrease significantly. The timing of maintenance is always challenging because different customers have different types of dust and dust volumes vary dramatically. In addition, the days or production periods of one individual customer may be different. If the maintenance is arranged only based on the calendar, there is a risk of arriving too late or visiting the place for nothing. Neither option is good because they lead to situations where the service is either poor or too expensive. In addition, the urgent maintenance requests received from the customer suddenly mess up the maintenance schedules and cause significant extra driving in different parts of Finland.
The situation would improve considerably if companies invested in smart filters that can measure the amount of filtered dust themselves and whose condition can also be monitored remotely. Smart filters allow maintenance to be carried out in a more timely manner, but the problem is the significantly higher price of smart filters compared to traditional “stupid” filters.
Picosun, on the other hand, considered in the accelerator increasing the capacity of the semiconductor production clusters they manufactured with the help of artificial intelligence. The objective was to optimize the operation of the robot moving wafers in production. The challenge of the cluster product was that their customers have very different production methods and applications. The change in the product manufactured by Picosun’s customer and the need for prioritization in terms of deadlines for finished products must be taken into account in the optimization. The primary objective was to reduce cluster waiting times and idleness to enhance the capacity of the customer’s production line in an ever-changing production situation.
When Teca considered the entirety of dust removal solution services, the issue was approached from a multifaceted perspective from the point of view of selling solutions, selling spare parts, and organizing service and maintenance services. Based on the customer discussions already held, it was known that, in principle, customers appreciate smarter filter solutions due to better scheduled maintenance and more consistent quality. Although the purchase of smart solutions would cost more than poor solutions, the development of maintenance arrangements with the help of intelligent solutions for artificial intelligence and filter equipment will make it possible to reduce the life cycle costs of a filter.
Smart filters would be able to both improve maintenance service and optimize service operations by utilizing artificial intelligence in selecting optimal driving routes and allocating resources according to appropriate maintenance schedules. In practice, this means a significant reduction in mileage, as maintenance routes can be planned according to foreseeable maintenance needs and the geographical location of maintenance sites and service technicians. Customers get even better service at a lower price than before, while travel expenses are significantly reduced.
Picosun, on the other hand, ended up with a solution in which, based on the known times for the work steps, continuous optimization of the sorting robot’s work order is carried out. The solution does not directly affect the work or processes of the people in production but enables a significant increase in production efficiency. At the same time, the customer’s capacity needs can also be monitored, and bottlenecks identified.
It can be said that in a certain sense the cases presented here are extreme: Teca’s solution affects many operators and processes within the company, while Picosun’s solution focuses on one key production phase in the semiconductor industry. However, it was essential in both use cases that they were chosen with the whole in mind: neither of the cases focused on minimizing the costs caused by individual tasks, but rather on finding new innovative solutions to improve the product and services, which enable the company to grow and develop in the future.
From now on
The operating method chosen for the accelerator proved to be a suitable way forward for both Teca Oy and Picosun Oy. From this and other business cases of the AI-TIE project, more experiences and lessons learned will be gathered in the future to share. They will be summarized as part of an online course to be published in autumn 2022.
With the experience and lessons learned, even companies that did not manage to get involved in the AI-TIE project accelerators will be able to develop their own artificial intelligence use cases. The free, self-paced online course provides learning and analysis tools for understanding the meaning of data and a company’s data resources, as well as developing and selecting a meaningful development target.
The aim of the course is to provide enthusiasm, learning and application opportunities. At its best, the course guides the company to select the most promising objects and use cases to be developed with the help of artificial intelligence. This makes it easy for companies to proceed to, for example, the Proof-of-Concept (PoC) phase, the more detailed planning of an AI project and the selection of a suitable development partner. Let’s bring artificial intelligence to work together!
Translated from the Finnish by Sari Benford and Petteri Saloranta.
Lagstedt A., Kauppinen R. & Lindstedt J. 2021. Vieraskynä: Ovatko säästöt tärkeämpiä kuin toimivat prosessit? Accessed: 24 February 2022.