Building Agentic AI Systems for Real-Time Inventory Management in Retail

Abstract

This study investigates the development and adoption of agentic AI systems for real-time inventory management in retail. Using a mixed-methods approach, data were collected from 100 retail professionals through surveys and five domain experts via semi-structured interviews. The survey captured inventory practices, operational bottlenecks, manual workflow burdens, data integration levels, and perceptions of AI, while expert insights provided context on design trade-offs, deployment challenges, and organizational readiness. Results reveal that 37% of participants experience stockouts or overstocking weekly, and 72% spend substantial time on manual inventory tasks, highlighting inefficiencies in current systems. Data integration and AI adoption remain limited, with trust, transparency, and workforce concerns influencing willingness to adopt autonomous solutions. Despite these barriers, 71% of respondents expressed optimism or cautious interest in AI adoption, emphasizing the need for structured training, workshops, and vendor support. Expected benefits include improved inventory accuracy, reduced manual workload, faster replenishment, and enhanced forecasting. This study provides actionable insights for building agentic AI systems in retail, bridging operational challenges with technological potential for autonomous inventory management.