What Remains, What Changes
When people talk about artificial intelligence, the conversation quickly turns to models, agents, and new tools. That is understandable. Yet many companies are starting the discussion in the wrong place.
Before exploring potential AI use cases, we should answer a more fundamental question: Are our data and process landscapes actually ready for AI?
After all, AI is only as useful as the information it works with. If data is missing, inconsistent, or lacks the necessary business context, even the best model will fail to deliver reliable results.
ERP Remains at the Core
We often hear the question of whether AI will eventually replace ERP systems. I believe that is the wrong way to look at it.
ERP remains where an organization’s core business processes come together. It is where orders, deliveries, financial postings, planning data, and many other transactions are processed. It contains the structures, rules, and relationships that businesses rely on in their day-to-day operations. Even in the age of AI, ERP therefore remains the company’s central system of record.
AI can build on this foundation. It can analyze information faster, identify patterns, make recommendations, and automate individual process steps. But it cannot replace the reliable foundation on which those decisions depend. Put simply: AI can transform the way we interact with ERP, but ERP itself remains the foundation.
Data Quality Becomes a Decisive Factor
Discussions about AI often focus on the capabilities of the models. In practice, however, the bigger challenge frequently lies elsewhere: in the data.
Incomplete master data, different terms for the same business concept, poorly maintained information, and unclear responsibilities are not new problems. But AI makes them more visible, and potentially more consequential. When automated processes rely on inaccurate or incomplete data, errors are not only carried forward; they can be propagated and amplified much faster.
This is why the challenge is not simply to store data correctly from a technical perspective. Business clarity matters just as much: What does a particular piece of information mean? Who is responsible for it? Which rules apply? And is the information reliable enough to support an automated decision?
Any company seeking to use AI effectively must address these questions, not as a prerequisite to be completed once, but as an ongoing responsibility.
Automation Requires Clear Accountability
AI will change many tasks. Routine activities can be automated, information can be prepared more quickly, and decisions can be better supported. But that does not mean people will disappear from business processes.
Critical processes will continue to require experts who can interpret results, recognize exceptions, and take responsibility. AI can suggest options, highlight risks, and recommend actions. However, responsibility for important decisions must not be delegated entirely to an AI system. This is particularly true when the underlying data is ambiguous, specific customer or business circumstances need to be considered, or a decision could have far-reaching consequences.
The role of people will therefore change. Less time will be spent on simple data entry and standardized approvals. Instead, process expertise, sound judgment, and the ability to critically assess AI-generated results will become increasingly important.
Go-Live Is Not the End
Another point is often underestimated in transformation projects: go-live is not the finish line. Companies investing in S/4HANA, cloud technologies, or AI are not building a final target architecture for the next ten years. They are creating a foundation that must be continuously developed. New requirements emerge, processes change, additional data sources become relevant, and technologies continue to evolve.
Building a modern system landscape is therefore only part of the challenge. Equally important is the ability to keep developing it over time. This applies to technology, data, and processes alike. It also applies to the organization: Who takes responsibility? How are changes decided? How are new applications tested and governed?
Do Not Start with the Tool
When companies ask me how they should begin their AI journey, my answer is rarely to start with a particular tool. A meaningful starting point is a specific problem or process. Where is unnecessary manual effort being spent today? Where is information missing that would enable better decisions? Which recurring tasks could genuinely benefit from AI? At the same time, the data foundation, architecture, integration, and governance must be considered from the outset. Otherwise, an AI pilot will remain an interesting isolated experiment that cannot be scaled across the organization.
The goal should not be to introduce as many AI applications as possible, as quickly as possible. It should be to implement the right use cases on a reliable foundation.
The Future Is Not a Choice Between ERP and AI
ERP remains the reliable foundation for data, transactions, and business processes. AI can make that foundation more accessible and intelligent, simplify processes, and support employees in making better decisions. However, the more companies automate, the more important high-quality data, clear processes, and well-defined accountability become. AI does not make ERP obsolete. On the contrary: the more intelligent and autonomous business processes become, the more important a powerful ERP system and a reliable data foundation will be.
If you would like to dive deeper into the topic, listen to the latest episode of the podcast “The Future of ERP“, titled “Will ERP Die in the Age of AI?”.
