Abstract
The digitalization of agri-food markets has outpaced the tools needed to coordinate heterogeneous, selfinterested participants under uncertainty and conflicting goals.Grain markets exemplify this: trade matching, transport scheduling, and storage allocation are optimized in isolation, leaving welfare unrealized.This paper proposes an integrated methodology and decision-support toolset for a grain marketplace platform built on a multi-agent system.Agent decisions become a fuzzy multi-criteria process generating ask and bid signals; a double-auction clearing mechanism couples with a transport-and-warehouse optimizer through an extended Contract Net Protocol, so prices internalize logistics costs; and a bilevel loop ensures convergence.Evaluated on a benchmark calibrated to public datasets and Kazakhstan grain geography, the toolset cuts the logistics cost index by 14.3 percent, raises order fulfilment from 78.4 to 95.2 percent, lifts fleet utilization by 20.9 points, and attains an allocative efficiency of 96.3 percent, equal to 97.4 percent of the centralized optimum, at far lower runtime.