ReflectiChain: Epistemic Grounding in LLM-Driven World Models for Supply Chain Resilience

REFLECTICHAIN introduces a Generative Supply Chain World Model (SC-WM) that encodes heterogeneous supply networks into a 6-dimensional graph-latent space, addressing the epistemic gap in AI agents within supply chains. The framework employs Double-Loop Learning to distinguish between epistemic and aleatoric uncertainties, achieving a 33.0% improvement in Rationale Consistency Score on the Semi-Sim benchmark, while maintaining 82.3% operability under adversarial conditions. This model's anti-fragile behavior and operational mechanisms provide practitioners with advanced tools for enhancing resilience in supply chain management using LLMs and RL.