From Data to Decisions: An AI/IT Intelligence Architecture for Asset-Light International Trading SMEs

Asset-light international trading intermediaries face a distinctive information-management challenge:they must coordinate heterogeneous data - commodity price feeds, supplier and shipment records,customs and phytosanitary documentation, multi-regime regulatory compliance and B2B customer creditexposure - across multiple product lines, without owning the physical infrastructure that traditionallyanchors data collection in trading and logistics firms. This paper develops a conceptual informationtechnology and artificial-intelligence (AI) architecture for such firms, using LAEXIM - a proposed asset-lightimport - export and supply-chain-services start-up serving Vietnamese SME producers and B2B buyers -as an illustrative applied case. Adapting a hospital-CIO-style requirements framework (scalability,interoperability, accuracy, regulatory compliance, real-time analytics and ethical governance) to thetrading context, the paper specifies a data architecture in which supplier, logistics, compliance and marketdata are consolidated into a shared analytics layer supporting descriptive, diagnostic, predictive andprescriptive decision-making. Drawing on peer-reviewed evidence that machine-learning models such asLSTM and gradient-boosted trees outperform classical statistical methods in agricultural commodity priceforecasting, and that machine-learning credit-scoring models improve default prediction for small andmedium-sized enterprises (SMEs) in supply chain finance, the paper argues that a modest, phasedanalytics investment can materially strengthen an asset-light trading firm's core value proposition:procurement timing, credit-risk management, compliance efficiency and customer retention. The papermaps this architecture onto LAEXIM's five revenue pillars and three-phase growth roadmap, showing howthe firm's proposed Pillar 5 (Supply Chain Intelligence and Trade Consulting) can be understood as thenatural productization of an internal AI/IT capability built to support Pillars 1 - 4. Implications forpractitioners designing similar SME trading ventures in emerging agricultural economies and for futureempirical validation, are discussed.