Once again, we sat down for a Q&A with Raffaello Lepratti, Global VP of Industrial Manufacturing at Mendix, to cut through the hype and talk about what’s actually happening on the factory floor today.

Raffaello knows a thing or two about evolving manufacturing technology. He has worked in manufacturing for 25 years, serving in various roles in automotive manufacturing, manufacturing operations management, and business development. He holds a degree in electrical engineering as well as a PhD in Advanced Human-Machine Collaboration.

Using GenAI across the industrial value chain

Across the industrial value chain, where do you see generative AI being the most prevalent or the most utilized?

Raffaello Lepratti (RL): That’s the critical, $1 million question.

Right now, I’m seeing GenAI being used a lot to create or distill documentation. Take instructions that a worker or operator has to work through and accomplish. These are, typically, fixed and are created prior to any operations. It’s static.

But conditions in the factory or a production line can change rapidly. As an example, when a new tool is introduced, that requires updated safety considerations for the operator.

GenAI makes documentation more dynamic. You can make changes that consider the new tool, for instance, or adapt to the new situation. GenAI lets you more quickly embed a new update to the documentation or a generate a new document. It shifts your manufacturing applications from static and pre-planned to dynamic.

Another place I’m seeing GenAI being put to use is in B2B2C and customer service. Businesses usually have a KPI around time-to-resolve for customers’ product issues. If a customer can provide a description of the product’s behavior, that goes through the company to the manufacturer, and it creates a record. It can be a slow process to get to the resolution.

GenAI can help by taking that record and immediately producing a set of instructions an organization needs to go through to accelerate their time-to-resolve. It’s analyzing claims and then producing repair protocols.

We’re also seeing early adoption where GenAI helps interpret operational data – summarizing trends, anomalies, or incidents in plain language for faster decision-making.

GenAI provides immediate guidance while the human service representative remains in the loop to make the final decision.

Real-world use cases with Mendix, Siemens, and AWS

What are some use cases you’re seeing where organizations are using Mendix, Siemens, and GenAI?

Raffaello Lepratti (RL): Well, the first example that I gave you was easy to come up with because it’s that’s actually happening with Mendix customers using Opcenter.

These customers use Opcenter, Siemens’ Manufacturing Execution System (MES) to execute production by guiding operators through the instructions of their asks. This is where they’ve implemented the ability to generate instructions dynamically out of any documentation that’s required for a job or task.

Mendix works with a lot of different infrastructures. One customer of ours uses AWS services because AWS is their cloud provider. We came up with how we can link AWS to Opcenter, and that’s where Mendix came into play. It allows the customer to easily integrate AWS Bedrock service into the application that creates those instructions.

This could apply to other infrastructures too, like Azure. The beauty of Mendix is the flexibility to connect to services from different hyperscalers.

Another way to look at it involves machine learning. If you’re looking at outcomes during the Statistical Process Control (SPC), you’re looking at the requirements, the results, the trends—anything that has to do with predicting a deviation in quality. Machine learning can learn from this. It can inform the manufacturer what the right path is or create alarms.

That’s a different level of GenAI, but it’s generated from machine learning. This is an important capability of Mendix, our Machine Learning kit. You don’t have to leverage an external service to use it in an application.

And Opcenter is the foundational part of everything here, because it gives the context. The MES gives all the information someone would need about a product and it tells you what the next production step in the sequence is. It gives you the conditions about the tools required to accomplish the task, the regulations around that task, and other specifics.

It’s all the important data and know-how and if we’re connecting any service and training the GenAI service or use our own ML kit to make use of all that data in the MES, then the customer can create an immediate impact.

What’s critical here is that Mendix isn’t just calling an AI service – it’s embedding intelligence directly into the operational workflow where decisions are made.

The combination of Mendix and Opcenter, or any Siemens Xcelerator product, is not just about the ability to connect multiple systems all the time. It’s about making sure you can easily consume the data by embedding any service that the customer wants or needs to embed with an infrastructure they already have.