Three Steps to Implement Artificial Intelligence in Industry
The productivity of Estonian industrial enterprises lags behind the European average. The use of artificial intelligence (AI) helps to increase productivity and competitiveness by making production management fast and data-driven.
The Achilles’ heel of Estonian industry is low efficiency. Companies are struggling – many have automated and robotized much of their production, yet they still lag behind Western European and Asian manufacturers. The productivity of the Estonian industrial sector workforce is only 58% of the EU average*. The competitiveness of companies is increasingly determined by the ability to use artificial intelligence (AI). However, the leap to using AI is much more complex than automating a single specific job. I propose the following initial steps that every industrial enterprise could take.
Start with data collection
Maximizing the potential of artificial intelligence requires that AI solutions can continuously learn from the data generated within the company. Therefore, one should start with data capture and consider how to store data generated at all stages of production in a data warehouse. Data collection must be automated by the production line or through separately installed sensors. Data accuracy is important; for example, it is necessary to differentiate between the time spent on line setup and production. Therefore, manual entry of production data is not suitable, as it can lead to errors and inaccuracies.
The data collected in production is used for regular updates of artificial intelligence, as production changes over time, and thus, data changes as well. Depending on the dynamics of the company, updating the AI solution may occur once a day, week, or month, as well as several times an hour. Buying AI does not mean purchasing a standard software package, but rather setting up self-learning software within the company.
AI brings the most benefit in changing processes, not just individual stages
The second step is to understand that processes need to be changed. The previous major technological leap occurred with the introduction of industrial robots and automation. This was relatively straightforward – companies identified processes that required a lot of manual labor and replaced them with automated solutions.
The use of AI in industry does not just mean changing individual production stages. Efficiency is increased through the management and organization of production and business processes using AI. However, this requires the restructuring of current work processes. Moreover, a large part of efficiency is hidden outside workbenches and production lines. Visiting Estonian manufacturing companies, I have noticed that low technological level work organization, such as order management, production planning, quality control, and workflow forecasting, currently absorbs a significant portion of Estonian industrial resources. Therefore, productivity remains low.
By managing with AI, the above processes can be automated. For example, when making a new price quote or accepting an order, a company using AI has already begun production
planning and knows with high precision the cost, raw material, and time required for that order. Processes become technologically more complex but logically simpler.
Real-time process management is a competitive advantage
The third step is adapting to changes in real-time. For example, AI-driven production management does not require extensive work to reorganize production with each change (such as order cancellation, production line failure, delayed delivery, or employee illness) because the work of multiple planners is done by AI in seconds or faster. Changing the production plan 20 times a day is no longer a problem but a competitive advantage that allows for efficient and highly productive operations.
A good example is how quality management has changed
Quality control on the production line, such as removing defective products, already works in many companies, but few can manage quality throughout the entire production chain.
Quality management involves understanding the causes of defects, learning from them, and preventing similar problems in the future. In other words, monitoring and improving process quality. This could be done before, when the company had the capability to hire data analysts and perform high-level data processing. AI has made this significantly more accessible and allows for finding very complex root causes of quality problems.
For example, in the food industry, a one-degree temperature difference at one stage or a slightly deviant moisture content at another stage may not affect a product individually, but combined, they can cause a quality problem. Proper data collection and quality control with AI help identify the causes of quality problems and change the production process to prevent future errors.
The stage of quality assessment at the end of production has become a continuous goal of the production process.
Based on the experiences of entrepreneurs who have used AI consulting, I can say that it is reasonable to establish a grand plan but start with changing one specific production or business process stage. If the main problem of the company is production planning, consider integrating sales process and production management with AI into one process. If the problem is quality management, consider solving it along with managing a specific production line. Once smaller problems are solved, they can be integrated. For example, data collected during quality control regarding production line speed and product parameters will allow for more accurate prediction of the production time and cost of similar products in the future.
Martin Rebane
Head of AI at Tartu Science Park.
AI Expert at AI & Robotics Estonia
*Based on data from 2022, “Estonian Industrial Policy 2035,” Ministry of Economic Affairs and Communications 2022