AI eats software: what this means for the SAP world

The debate about AI eats software is often read as a general attack on business software. However, it is not aimed at the ERP core, but at the layer above it - user interfaces, workflows and add-ons. These are the areas through which software has been used and monetized to date. Agentic AI fundamentally shifts this logic. Processes are no longer triggered via screens and manual click sequences, but are executed directly at data and logic level. For SAP systems, this is not a displacement, but a functional enhancement. With the loss of importance of the interaction layer, the ERP core gains in importance. Systems such as S/4 remain the system of record - as an instance for consistent data, clear authorizations and audit-proof processes. This function is not replaced by AI, but rather strengthened. The more processes AI agents automate, the greater the dependency on the quality of the underlying data and process landscape. Incorrect master data, inconsistent logics or special solutions that have evolved will then no longer have a selective effect, but a systemic one. This does not make ERP obsolete. It shifts its function: away from the operational user interface and towards a stable basis for execution. 

From Seats to Execution

The logic according to which software is valued is changing. Traditionally, the value is based on the number of users. Licenses are awarded per seat, efficiency is measured by usage. This logic reaches its limits with autonomous systems. This is because agents work directly in the processes: they analyze data, identify deviations and initiate decisions. They do not occupy a seat. The economic focus shifts from utilization to execution. The decisive factor is no longer the number of people operating a system, but how stable, efficient and traceable automated processes are. The control variables are changing: The degree of automation, throughput times, error rates and the quality of automated decisions are coming into focus. This places clear demands on architecture and control. AI agents require stable interfaces, consistent data and clearly defined processes. Many ERP landscapes are not prepared for this: grown systems, extensive in-house developments, fragmented data distributed across ERP, CRM and production systems. Clean Core is evolving from a modernization topic to an operational requirement. Open architectures are also crucial. The dynamics in the model market are enormous. Model diagnostics is becoming the central architectural principle. Governance is taking center stage. Which tasks can an autonomous system carry out independently? Where is human control mandatory? Decisions and process steps must remain explainable and auditable. Governance becomes a permanent management task.

From project to sovereign operation

The challenge is no longer the introduction of modern SAP systems, but rather their confident operation. Integration, data quality, governance and model control must be continuously monitored and further developed. As a result, the focus of SAP partners is shifting away from UI-oriented implementation and towards resilient data and process architectures as a control foundation. Operating models such as managed cloud services are changing. They are becoming the operational control level for complex, autonomous systems - with monitoring, intervention logic and clearly defined autonomy limits. AI eats software does not describe the end of ERP, but the transition to a new phase. After migration to the cloud and standardization, the focus shifts to operations: stable data, integrated processes and controlled autonomy. ERP remains the foundation. The question is not whether SAP will be replaced by AI, but how companies can set up their ERP landscape in such a way that autonomous systems can work on it reliably, controllably and economically. (Source:
All for One Group)