SoK: The Economics of Trustworthy Agentic Commerce — Information, Incentives, and Interoperability across Web2 and Web3

Autonomous AI agents are becoming economic actors: they discover counterparties, delegate tasks, authorize transactions, and exchange value across an emerging stack of Web2 and Web3 protocols. Yet existing maps largely describe what protocols exist rather than explain why their functions are economically necessary, whether their guarantees compose, or whether they support a complete and trustworthy exchange. We address this gap through information economics. We construct a reproducible corpus of 304 academic, protocol, industry, and institutional records and derive an eight-layer taxonomy from the informational and contractual problems arising before, during, and after exchange, including adverse selection, information design, moral hazard, incomplete contracts, and enforcement. We then evaluate representative protocols using an assurance framework covering coherence, interoperability, completeness, and institutional fitness. The analysis reveals a fundamental distinction between connectivity and trustworthy interoperability: identity, delegation, authorization, payment, and interaction mechanisms provide useful local guarantees, but these guarantees do not consistently propagate across layers or the Web2-Web3 boundary. Most importantly, while the first seven layers have active protocols, the stack lacks a native, widely adopted mechanism for agent-speed dispute resolution, leaving execution increasingly machine-speed while adjudication remains platform-specific, human-speed, or jurisdiction-dependent. We contribute a theory-grounded systematization of agentic commerce, an assurance-based framework for evaluating protocol composition, and a research agenda for designing economic institutions capable of supporting trustworthy autonomous exchange.