sap ewm · s4hana · clean core · ai · supply chain · 2026
Clean Core First: Why the Autonomous Enterprise Hype Actually Matters for EWM Deployments
Agentic AI sounds distant from warehouse ops—but the clean core S/4HANA push behind it directly shapes how EWM gets deployed. Here's why that matters.
Clean Core First: Why the Autonomous Enterprise Hype Actually Matters for EWM Deployments
The autonomous enterprise framing feels abstract until you realise it's quietly reshaping the conversation around every S/4HANA project—including EWM rollouts.
Cognitus recently published a piece on how they're positioning SAP transformation work around the autonomous enterprise shift. Reading it as someone deep in EWM implementations, my first reaction was: this isn't directly about warehouse management. My second reaction was: but it should be on your radar anyway.
Here's why.
The Clean Core Argument Is Not New — But It's Getting Teeth
For years, "clean core" sounded like SAP's way of telling consultants to stop doing things the hard way and customers to stop customising everything in sight. In practice, most projects nodded at the principle and then went and built Z-programs anyway.
What's changed is that agentic AI — the idea of AI systems acting autonomously across business processes — actually requires clean, well-governed data to function. You can't build reliable automation on top of a messy master data landscape or a tangle of custom logic that nobody fully understands anymore.
This is where EWM deployments connect to the bigger picture.
When I'm on a warehouse rollout, the data foundation issues surface fast. Storage type configurations referencing warehouse numbers that don't align with the S/4HANA plant structure. Handling unit data that was migrated half-heartedly. LGTYP, LGPLA, and LETYP values that vary across sites because nobody standardised them before go-live. These aren't abstract data governance problems — they're the reason putaway strategies misfire and yard management processes stall.
If you want automation to work — not AI-autonomous-enterprise-level automation, just basic RF-driven goods receipt without exceptions — you need that foundation clean.
What "Agentic AI" Actually Means in a Warehouse Context
I'll be honest: fully autonomous AI agents managing warehouse operations is not something I'm configuring in EWM today. That's a longer horizon.
But the intermediate steps are already visible. Embedded analytics on warehouse KPIs. Predictive labour demand based on inbound delivery schedules. Exception-based workflows where the system flags the problem and a human resolves it — instead of a human hunting for the problem manually.
None of that works well if your warehouse order creation rules are inconsistent across activity areas, or if your work centre assignments haven't been thought through properly. The AI layer, whenever it fully arrives, inherits whatever mess is underneath.
The Cognitus framing — clean core and data governance as foundational to autonomous capability — is correct. It's just that most warehouse teams encounter this truth through a different path: a misrouted HU, a failed RF transaction, or a stock discrepancy that takes three days to trace.
Where This Connects to Modern Tooling
I'm building markus-mildner.com partly as a practical experiment in this exact space — applying modern tooling, AI assistance, and clean architectural thinking to work I actually do.
The pattern I keep running into: the value of automation compounds fast once the data layer is solid, and almost doesn't compound at all when it isn't. This applies whether you're talking about an EWM warehouse process, an S/4HANA deployment, or a personal platform running on AWS.
In a warehouse rollout, that means the configuration work that feels slow and unglamorous — getting your storage section search strategies right, aligning activity area definitions, building your HU type assignments carefully — is exactly the work that makes everything downstream easier. Including any future automation layer.
What I'd Take From the Cognitus Piece
A few concrete takeaways for anyone working on SAP transformations right now:
1. Treat clean core as an EWM enablement strategy, not just a corporate directive. Every workaround you build today is a liability when you need to extend the system. That Z-enhancement to the putaway strategy feels necessary now. Two years from now it's a migration blocker.
2. Data governance has to happen before go-live, not after. Retroactive master data cleanup is brutal. Warehouse number alignment, plant assignments, HU type hierarchies — these need decisions made early, with the business, not fixed post-cutover.
3. The AI conversation is worth tracking even if it's not your immediate project. The direction is clear. SAP is building toward more autonomous processing. Understanding the architectural requirements now — clean APIs, consistent data, minimal custom core logic — means you're less likely to be building the wrong thing today.
The Cognitus article is worth a read if you're thinking about where SAP's transformation narrative is heading. The warehouse-specific implications aren't spelled out there — but they're real.
If you're working through similar questions on a current EWM or S/4HANA project, I'm curious what you're seeing on the ground. Drop a comment or reach out at markus-mildner.com.
