Agentic AI-Driven Enterprise Decision Support for Cross-Domain Business Intelligence

Abstract: Existing enterprise decision-support systems often suffer from fragmented data management, limited cross- domain knowledge integration, and insufficient autonomous analytical capabilities, resulting in reduced decision accuracy and inefficient intelligence generation. To address these challenges, this paper proposes an Agentic Intelligence Framework (AIF) for cross-domain business intelligence that employs hierarchical multi-agent collaboration to improve autonomous reasoning, knowledge synthesis, and decision support across enterprise environments. First, heterogeneous enterprise data originating from retail, finance, and networking domains are integrated through Retrieval-Augmented Generation (RAG) and domain-specific knowledge repositories, enabling contextual knowledge retrieval and semantic enrichment. Second, a hierarchical multi-agent architecture comprising Research Agents, Analytics Agents, Validation Agents, and Insight Generation Agents is designed to collaboratively investigate complex business scenarios, validate analytical outcomes, and maximize the quality and reliability of generated intelligence. Third, an adaptive reasoning and workflow orchestration mechanism is introduced to dynamically coordinate agent interactions, continuously refine analytical hypotheses, and uncover hidden relationships among distributed enterprise data sources. Finally, explainable AI techniques are incorporated to provide transparent reasoning paths and interpretable recommendations, thereby improving user trust and supporting informed decision-making. Experimental results demonstrate that, compared with conventional AI-based business intelligence frameworks, the proposed AIF achieves higher analytical accuracy, improved knowledge discovery capability, reduced decision latency, and enhanced explainability across retail, financial, and networking applications.

Keywords: Agentic Artificial Intelligence; Multi-Agent Systems; Business Intelligence; Retrieval-Augmented Generation; Large Language Models; Explainable Artificial Intelligence; Cross-Domain Analytics; Enterprise Decision Support.