Agentic Artificial Intelligence for Intelligent Business Decision-Making: A Multi-Agent Framework Integrating Large Language Models, Explainable AI, and Business Intelligence

Abstract

The increasing complexity of modern business environments requires intelligent decision-support systems capable of combining predictive analytics, contextual reasoning, and transparent decision-making. This study proposes an Agentic Artificial Intelligence framework for intelligent business decision-making by integrating Large Language Models (LLMs), Explainable Artificial Intelligence (XAI), machine learning, and Business Intelligence through a collaborative multi-agent architecture. The proposed framework utilizes specialized AI agents for data processing, feature engineering, prediction, explanation generation, and business reasoning, coordinated through an orchestration agent to produce actionable insights.The framework is evaluated using the Online Shoppers Purchasing Intention Dataset from the UCI Machine Learning Repository. Multiple machine learning models, including Random Forest, XGBoost, CatBoost, Artificial Neural Network, and LightGBM, are compared, where LightGBM achieves the best standalone performance with 94.82% accuracy and 0.99 AUC. After integrating multi-agent collaboration, SHAP-based explainability, and LLM-driven reasoning, the proposed framework improves performance to 95.63% accuracy, 0.95 F1-score, and 97.1% decision consistency while providing interpretable business recommendations. The results demonstrate that the proposed Agentic AI framework effectively bridges predictive analytics and intelligent business decision-making by delivering accurate, explainable, and actionable insights. The framework provides a scalable approach for AI-driven decision support across industries such as retail, finance, healthcare, manufacturing, and supply chain management.