Manufacturers warn AI readiness depends on verified product data logic

The problem stems from legacy systems built over decades, where original rule creators have long since left. Modernization projects like SAP S/4HANA migrations often focus on applications and infrastructure, treating configuration logic as secondary. This oversight introduces hidden risks: new configuration engines may interpret old rules differently, altering valid product combinations without explicit changes. With industrial models containing tens of variables—each with multiple choices—manual testing is impractical. Formal methods, like compiling rules into mathematical representations, offer a scalable way to verify correctness before AI deployment. The author argues AI should augment deterministic engines, not replace them, ensuring reliability in recommendations while keeping critical logic provably accurate.