How industrial AI is changing the shop floor

AWS advertises with the slogan “Built for Industrial AI.” What do you understand by that?

Jan Metzner: We are currently seeing two central areas of development: Physical AI and agent-based maintenance systems. Physical AI is about how artificial intelligence can support manufacturers - both machine builders and producing companies. What is interesting here is that simulation is brought together with what is happening on the shop floor. The intelligence can run in the cloud and optimize production in advance or act directly on the shop floor.

What concrete use cases do you currently see?

Metzner: One particularly exciting field is maintenance agents. Here, added value can be achieved very quickly by bringing together maintenance data with manuals and documentation. This helps employees find information much faster and solve problems more efficiently.

Development is now going even further. Production lines are logically divided into individual areas, each of which has its own agents. These agents communicate with one another. This is slowly replacing traditional SCADA systems. For the actual control, a PLC is still required, but sequence control no longer has to be rigidly programmed. This makes production lines significantly more flexible.

The simplification and acceleration of decision-making processes is currently considered one of the most important drivers for the use of AI.

Metzner: Absolutely. At the Hannover Messe, we had a small production line on display that we only physically assembled on Sunday afternoon. Since each element has its own agent, we were able to implement that very quickly.

It is always about the concrete added value. As soon as the added value is clear, that drives innovation.

Another major topic is data sovereignty. What are you working on here?

Metzner: European manufacturers have been able to host their data in our AWS region in Frankfurt since 2014. In the meantime, we have the European Sovereign Cloud in Brandenburg.

It is technically completely independent of the commercial infrastructure and can survive autonomously in an emergency.

This is particularly relevant for highly regulated industries and the public sector. Many mechanical engineering companies, for example, now have customers or projects in the defense environment in which exclusively European solutions are required.

Industrial AI forms the basis for a resilient industry

Artificial intelligence alone does not make industry fit for the future. What is decisive is whether companies connect AI and digital twins into a new industrial architecture. Sabine Scheunert of Dassault Systèmes explains how this can succeed.

Anja Ringel

Is the topic of data sovereignty in high demand among companies?

Metzner: That depends heavily on the respective company. A globally positioned automobile manufacturer naturally views the topic differently than a specialized mid-sized company.

Especially among smaller manufacturers or in the defense sector, this is clearly an issue. That is why we provide a framework with which companies can assess their actual need for sovereignty. This also includes external audits. In this way, every customer can make a well-founded decision about which requirements are really necessary.

What role does the use of generative AI play in this?

Metzner: Data sovereignty also plays an important role in the field of AI and machine learning.

The following applies here: It does not always have to be the most powerful AI model. Often, a more cost-effective model is sufficient. What matters is whether it reliably fulfills the specific use case.

At your trade fair booth, you demonstrated agent-based production control. How does it work?

Metzner: There are several agents, each responsible for parts of a production facility. If, for example, a robot fails, an optimization agent analyzes the situation and asks: "Where can I build the products instead? What must be completed first?"

Users can communicate directly with the agents and, for example, request information on capacity utilization or raw materials. This is the vision of Industry 4.0, which is now becoming feasible because we no longer have to program everything rigidly.

Do these systems also work in multiple languages?

Metzner: Yes, that is an essential requirement. In modern plants, people of many nationalities work. Therefore, user interfaces and assistants must be available in different languages.

We have also gathered this experience in our Amazon Fulfillment Centers. There, employees can now switch the user interfaces to their respective language.