Efficient use of supply chain analytics for decision making remains a major challenge for planners. Critical tasks such as database querying, KPI analysis, demand forecasting, and performance diagnosis require heterogeneous expertise across data engineering, operations research, and domain knowledge. To bridge this gap, we propose an agentic system that links business decision‑making with technical expertise. A coordinator agent interprets user intent and delegates sub‑tasks to specialist agents.
The architecture supports both exploratory analysis and deterministic workflows, allowing planners to move seamlessly between ad‑hoc questions and structured processes. Domain logic is encapsulated within specialist agents and their prompts, yielding a scalable, modular, and auditable design while reducing functional extension cost through prompt‑centric development.
Evaluation on a testbed that mimics multi‑echelon inventory management shows that the multi‑agent design achieves roughly 90% accuracy, comparable to a single‑agent baseline, yet cuts input token usage by about fourfold, markedly improving scalability and cost‑efficiency. Case studies illustrate interpretable suboptimality detection and automated forecast optimization, demonstrating how agentic architectures can combine open‑ended exploratory analysis with deterministic supply chain analytics workflows, offering a practical path toward more accessible and extensible decision‑support systems.
Review