Artificial Intelligence–Driven Viable First-Mile Food Supply Chains: A Unified Robust Stochastic Programming Approach with Immune-Inspired Adaptation

This study addresses the problem of designing viable first-mile agricultural supply chains under uncertainty, with a focus on small-scale production systems characterized by territorial heterogeneity and logistical constraints. A unified modeling framework is proposed, integrating a data-driven Rural Viability Index (RVI), a Unified Robust Stochastic Programming (URSP) formulation, and adaptive mechanisms inspired by the Human Immune System (HIS). The approach combines territorial assessment, uncertainty management, and endogenous adaptation within a single decision-making structure. The framework is applied to a fresh pea supply chain in Cundinamarca, Colombia, considering multiple uncertainty regimes and adaptive configurations. The results show that system performance is strongly influenced by the interaction between URSP weighting and the activation of adaptive mechanisms. Configurations with HIS activation consistently achieve higher net utility, improved viability indicators, and greater supply capacity, especially under scenarios with a stronger emphasis on disruptive conditions. At the same time, the analysis reveals trade-offs in environmental impact and labor demand, highlighting the multi-dimensional nature of supply chain viability. The findings demonstrate that viability is not achieved through isolated optimization components, but through the coordinated integration of territorial intelligence, uncertainty modeling, and adaptive response capabilities. From a methodological perspective, the study contributes by operationalizing viability within a unified optimization framework, bridging the gap between predictive analytics and prescriptive modeling. From a practical standpoint, the results suggest that first-mile supply chain design should incorporate adaptive mechanisms and spatial information as core elements to ensure sustained performance under uncertainty.