Classifying patent applications, recommending ways to operate plants, and predicting the purchase prices of raw materials—the use of artificial intelligence is already part of daily work at Evonik today.


This kind of interaction between human beings and machines is already working in many areas. Production is a case in point. “At the moment, artificial intelligence cannot operate a production plant entirely on its own,” says Hendrik Ewe, Group Leader Data Engineering at Technology & Infrastructure. “Instead, it operates like an experienced colleague who gives me assistance and advice.” This “colleague” analyzes the relationships between dozens or even hundreds of factors, such as the quality of the raw materials, their reaction behavior, and the humidity of the air in the production plant. The algorithms identify relationships between these factors that are difficult for human beings to detect. On the basis of these relationships, they recommend that employees carry out specific actions.

The aim is to ensure that the plant functions as efficiently as possible and delivers an optimal product yield. Today the AI in production plants even uses data that extend beyond the production operations themselves. “Because of the way we’ve trained the program, it can include market prices in its recommendations,” says Ewe. Has the price for a relevant precursor increased? In that case, AI might recommend a production process that is more energy-intensive but requires fewer raw materials.


Thomas Paul’s work also focuses on prices. He’s a data scientist at Marketing & Sales Excellence. As a member of a team that also includes people from Procurement, he developed an AI-based app whose algorithms predict the future purchase prices of raw materials. “The program autonomously analyzes past prices as well as external influencing factors that will play a role in the future,” he explains. If the price of a certain raw material appears likely to increase in the near future, the purchasers at Evonik buy rather sooner than later, thanks to the AI app. That saves money and reinforces the Group’s competitiveness.


Virtual Formulation Assistant COATINO® enables paint and coatings formulators to obtain AI-based additive recommendations for countless applications and individual guiding formulations for pigment concentrates.

AI could also play an important role in the area of research and development. For example, the Business Line Coatings & Additives now offers COATINO, the first-ever language assistant for the paint industry. And SciTAI (Scientific-Technical Support by Artificial Intelligence) gives the plastics experts at the Business Line High Performance Polymers centralized access to the know-how accumulated through in-house research in the area of compound formulations.


Not only product innovations but also AI innovations must be conceived, developed, tested, and often adapted or even rejected. Daniel Wais works in the Digital Labs at Global IT. Together with his colleagues, he develops AI solutions that help to optimize the work done by other Evonik departments.

“One example is a program called Patent Classifier, which - amongst other approaches - uses decision tree learning. It was trained with patents that have already been assessed, and on the basis of these data it creates a ranking,” Wais explains. “With the help of the ranking, the researchers at the business units know at once which patents are definitely relevant to them.”


In order to support the use of AI close to the specific application, the employees at Global IT, Marketing & Sales Excellence, Technology & Infrastructure, and Evonik Digital are part of a global network that includes offices in Germany, the USA, China, and Singapore.

At an industrial group as big as Evonik, in the past it was entirely possible for different groups to be duplicating their efforts, but all of that changed last year.

Since the summer of 2019, the AI experts at Evonik have been working together to pool the Group’s knowledge and resources so that they can expand the utilization of AI. The result of their cross-unit approach is the Cognitive Solutions Agenda for evaluating, prioritizing, implementing, and scaling AI projects. The main focus is on the employees. The experts want to sensitize them to the utility and opportunities offered by AI and help them develop a basic understanding of how AI works.