AI-supported variant management activities – insights from an industrial case study

ABSTRACT: Variant management faces increasing complexity that challenges traditional rule-based configuration approaches. This contribution explores how AI can support the generation of configuration rules (1) by comparing two solution concepts – a deterministic Python-based and an LLM-based approach. Following a structured early-stage AI system development methodology, the research investigates (2) how AI can be methodically integrated into variant management and (3) how implementation factors differ between both approaches. The results reveal distinct trade-offs between transparency and efficiency.