From XML Brokered Markets to Autonomous Agentic Commerce: A Design Science Framework for AI-Driven B2B Marketplaces

Abstract

More than two decades ago, electronic-commerce researchers proposed intelligent agent-based business-to-business marketplaces in which software agents represented buyers and suppliers within brokered exchanges. This paper revisits that brokered-marketplace concept through a design science lens and proposes an Agentic Commerce Framework in which autonomous AI agents operate within governed marketplace ecosystems. The framework integrates economic foundations, governance, trust and explainability, marketplace orchestration, specialized agents, transaction processes, controlled interoperability, and risk-adjusted human oversight. Six design principles and fifteen research propositions guide future implementation and empirical study. The paper contributes by linking electronic-marketplace theory to contemporary agentic AI, recasting the intermediary as a governance-enabled orchestrator, and establishing a research agenda for autonomous B2B commerce.