Companies must learn to deal with artificial intelligence (AI) or risk going under. For most business areas, AI is an important part of the solution for creating more efficient, precise, and user-friendly systems. And yet, many organizations find it difficult to implement. The following four principles show how companies can master the path to becoming data-driven AI champions.1
The AI market is booming. Analysts expect global sales of this technology to more than double to about 1.2 trillion dollars by 2028. Many industries from the automotive industry to the energy sector and logistics companies are investing in AI solutions. Nevertheless, many companies still limit their use of AI to individual applications. However, highly specific stand-alone implementations are particularly difficult to transfer to other areas. Therefore, the first principle in dealing with AI is: focus on merging all operational systems. For every new system that companies introduce, they should make sure that it can communicate with the other existing systems through interfaces. Incidentally, this applies regardless of whether AI is used or not. Those who take this principle to heart ensure that all data flows together across the company and can be used to train AI applications.
Digitalization is transforming function-based systems into data-based systems.To prevent isolated applications, inaccessible data silos, and unknown data formats, the AI era requires information to be provided in an easily consumable format. How can the information from a new system help other applications? How must data be prepared in order to enable its incorporation in the company’s own AI model in the future? Being service-oriented makes all data sources, including legacy data from ERP or CRM systems, usable in the company and facilitates the scaling of AI solutions. One example is Dematic, a supplier of automated conveying and sorting systems. In cooperation with T-Systems, Dematic provides its customers with fully automated high-bay warehouse systems. The AI in the system uses all available customer data to recognize patterns and improve processes.
AI applications and platforms deliver their full benefits when as many departments and users as possible can access them. To ensure a successful AI strategy, companies should therefore promote the usability of AI applications. For this step, it is necessary to design end-user-friendly interfaces that make the AI platform easily accessible for users in the company. An intuitive chatbot enables employees to access AI-driven information for their tasks. For example, sales staff might use the AI platform to ask the chatbot for relevant customer data for their next call. Increased usability has a beneficial side effect: the better the workforce gets at using AI, the faster the AI investment generates profitable added value.
Anyone working with AI models must be aware of the importance of data governance and risk management. For example, the recently adopted EU AI Act formulates clear administrative requirements that companies must fulfill when using AI. Still, regulation offers companies enough leeway to experiment. Considering this, company managements should define what they want to achieve with AI and then give their specialist departments a free hand in the actual development. Employees can set up their own AI applications with the help of low-code tools. Outstanding examples of this can be seen with many of our customers: for example, consider a team in a manufacturing unit that uses AI to improve assembly line processes or the HR department that uses its own AI application to analyze job applications. The proactive and creative use of AI creates numerous small innovations that would previously have required intervention from entire development departments.
Based on my experience, I can say that data-driven transformation will only succeed if companies define a clear strategy. The use of AI does not necessarily have to be the immediate goal. However, the sooner companies deal with the availability and harmonization of data sources, the easier it will be to use them later. A strong data basis and an open, courageous AI culture are the best ingredients for successful digitalization. If you are interested in AI, want to find out more, or are already in the course of implementing your own AI road map, please feel free to write to me. I look forward to our dialog and your ideas.
1 How to move beyond a monolithic data lake to a distributed data mesh, Zhamak Dehghani, 2019, martinfowler.com
