Developing Consumer Trust in Sustainable Agentic Commerce: Conditional Signals, Auditable Governance, and Justified Non-Purchase Advice

Agentic artificial intelligence (AI) changes digital retail from consumer-led search to delegated decision making, increasing both the potential value and the agency risks of recommendation. This conceptual article asks the following: Under what architectural and governance conditions can an autonomous shopping agent earn calibrated consumer trust while supporting economically and environmentally appropriate consumption decisions? Using a purposive, multidisciplinary synthesis of signaling theory, principal–agent theory, the ability–benevolence–integrity model of trust, trustworthy AI, recommender system governance, and sustainable consumption research, we develop a process model linking four architectural features to consumer perceptions, delegation, reliance, and decision outcomes. The features are scoped provenance verification, disclosed and audited commercial incentives, faithful contrastive explanations, and active evaluation of retention, repair, refurbishment, and purchase. We argue that justified non-purchase advice can function as a credible signal of consumer alignment only when the operator is the signal sender, the advice creates observable opportunity or accountability costs, and compensation and audit arrangements make imitation more costly for a misaligned operator. The framework distinguishes perceived trustworthiness from overall trust, willingness to delegate, actual reliance, recommendation appropriateness, and lifecycle sustainability. A five-layer architecture, a worked decision example, and four testable propositions specify how missing data, conflicting preferences, uncertainty, authorization, privacy, and accountability should be handled. This article treats the applications as illustrations rather than empirical cases and concludes with limitations and an empirical research agenda focused on calibrated—not merely increased—trust.