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

arXiv:2606. 10359v1 Announce Type: new Abstract: AI agents in supply chains face a fundamental epistemic gap: large language models (LLMs) interpret policies but lack physical grounding, while reinforcement learning (RL) optimizes flows but is semantically blind to unstructured constraints.