Digitalization rarely fails because of ideas, but mostly because of reality: SAP Core, non-SAP systems, production, data sources everywhere, backlogs full, resources scarce. Then AI comes along and everyone hopes for a big leap forward. But without clean data and processes, it often remains just a good demo.
In the session at ITOK26 of IT Onlinemagazin, we talked to Adrian Dexheimer from Mann & Schröder Cosmetics Group about exactly that. And what sticks is not so much a tool as an approach: think side by side, prioritize use cases, deliver quickly, but with clear guidelines for governance, security and operation.
The crucial point: AI needs a foundation
A practical learning from the conversation: In AI projects, a lot of time is not spent on the model itself, but on making it suitable for the enterprise. Data connection, roles, authorizations, cost control, deployment, operation. It is precisely these topics that ultimately decide whether AI works in everyday life or only in the proof of concept.
That’s why it’s worth looking at the foundation underneath: Integration, data, workflows and applications that neatly connect these building blocks.
From theory to practice: use cases that really work
Mann & Schröder did not stop at one pilot. In the talk, several implemented applications were mentioned that are left lying around for a long time in many organizations because they are too small for a major project, but too important for “we’ll do it later”.
Examples from the session
-
Print data process with documents and PDF handling directly in the browser
-
Connection of a laboratory database with feedback of measured values to SAP
-
Master data pool for product data including synchronization
-
Field service solution for orders and status information with SAP access
Why workflows are more than just a technical detail
Workflows create clarity.
As soon as a process is clearly described, roles, statuses, handovers and responsibilities become visible. This not only helps with implementation, but also with acceptance and maintainability. The application is then not “an app”, but a clean process with comprehensible logic.
Cooperation that works
Instead of handovers between the business department and IT, a tandem was described: Process Owner in the business department and Product Owner in IT. This sounds simple, but in practice it makes all the difference because requirements are not “thrown over the fence”, but are created together.
Conclusion
If you wait for the one big transformation project, you lose time. If you start side by side, consistently deliver use cases and take governance seriously, you can achieve results faster without touching the SAP core and without inheriting chaos later on.
