A Simulation-Based AI-Enabled Supply Chain Decision Framework for Cross-Border Business Expansion

This study develops an AI-enabled decision framework for cross-border supply chain management of small and medium-sized enterprises (SMEs). A SaaS-based platform integrates heterogeneous enterprise data into a unified supply chain state. We employ three task-oriented agents for demand forecasting, route planning, and data-value evaluation, and use an LLM-RAG component to provide semantic decision support for interpreting quantitative outputs. We conduct a 12-period multi-market simulation with four supply nodes and six demand markets, and compare the proposed framework with a traditional experience-driven baseline under identical conditions. The results indicate lower average MAPE and reduced inventory fluctuation under the proposed framework compared with the traditional baseline. The results provide preliminary evidence that the framework may improve demand forecasting accuracy and inventory stability under the simulated conditions.