LLM agent driven online auction mechanism for agricultural products

Purpose The study aims to address persistent inefficiencies in agricultural product trading—namely, information asymmetry, market opacity, and trading inefficiency—that hinder supply chain profitability and disrupt market stability. It explores how intelligent systems, empowered by large language models (LLMs), can enhance transaction transparency, optimize decision-making, and improve market performance. Design/methodology/approach This research integrates large-scale language models (LLMs) into an online auction framework through dynamically adaptive intelligent agent systems. These agents simulate real-world market behaviors and leverage machine learning algorithms to enhance auction decision-making. An empirical evaluation is conducted using three experimental groups, with performance assessed across multiple quantitative metrics. Statistical significance is tested using Tukey’s HSD post-hoc analysis. Findings The results reveal that the LLM-based agent system significantly outperforms the other two benchmark approaches on all performance indicators. The system demonstrates superior capability in mitigating the effects of information asymmetry, improving pricing accuracy, and facilitating more efficient and transparent transactions. Research limitations/implications Geographic and temporal data limitations may impact generalizability; further research is needed. Practical implications The proposed model offers actionable insights for designing next-generation online trading platforms in the agricultural sector. By embedding LLMs into intelligent agent systems, market participants—including farmers and buyers—can benefit from enhanced information flow, reduced transaction costs, and more balanced supply-demand interactions. Social implications Promotes fair transactions, reducing market information asymmetry for sustainable agricultural development. Originality/value This study is among the first to incorporate LLMs into auction-based agricultural trading systems, demonstrating the transformative potential of AI in reshaping market mechanisms. The research bridges artificial intelligence with smart agriculture, contributing a novel framework that advances both theoretical understanding and practical innovation in digital agriculture and e-marketplace design.