Abstract

This study presents a mathematically grounded framework for integrating Enterprise Resource Planning (ERP) systems with Artificial Intelligence (AI) to enable automated risk assessment and decision support in strategic project planning. The proposed approach models enterprise operations as a high-dimensional state vector derived from ERP data streams, capturing financial, operational, and procurement indicators. A formal Enterprise Risk Function (ERF) is introduced to quantify risk as a nonlinear aggregation of transformed features with stochastic components. Supervised machine learning models, including XGBoost and neural networks, are employed to approximate the risk function and generate real-time predictive risk scores. These predictions are subsequently embedded within a constrained optimization model that determines optimal managerial decisions by minimizing expected cost under operational constraints.
The framework is evaluated using ERP-based datasets, demonstrating significant improvements in prediction accuracy, robustness, and decision efficiency compared to traditional rule-based methods. Sensitivity analysis confirms the stability of the model under parameter variations, while optimization results show measurable reductions in project risk exposure and operational cost. The findings highlight the effectiveness of combining ERP data integration, machine learning, and optimization theory into a unified decision support system. Despite challenges related to data quality and model interpretability, the study establishes a scalable foundation for intelligent, real-time, and risk-aware strategic planning in complex enterprise environments.