Enabling LLMs to Navigate ERP Systems

Enterprise Resource Planning systems like SAP S/4HANA contain decades of business logic, custom configurations, and complex data relationships. The question is no longer whether AI will transform ERP — but how.

The Challenge

Traditional ERP automation relies on rigid, rule-based logic. Business rules are hardcoded in ABAP, workflows follow predefined paths, and any deviation requires developer intervention. This creates a gap between the flexibility businesses need and what their systems can deliver.

Where LLMs Fit In

Large Language Models bring a fundamentally different approach to ERP interaction:

Natural Language Processing for Documents

Intelligent Decision Support

Agentic Workflows

The Research Perspective

My doctoral research at the University of Mannheim focuses on creating frameworks that allow LLMs to understand ERP data structures, navigate complex module interactions, and generate actionable insights — all while maintaining the compliance and auditability that enterprise systems require.

Conclusion

The future of ERP lies in hybrid systems that combine the structured reliability of traditional business logic with the adaptive intelligence of modern AI. The goal is not to replace SAP — but to make it smarter.