Agentic AI is increasingly framed as enabling consumers to delegate commerce decisions and actions to digital assistants, yet consumer-facing evidence still centers on assistive chatbots and recommender-like systems, with scarce evaluation of execution-level delegation. This study provides an evidence-mapping review of empirical work on agentic commerce and synthesizes determinants and outcomes of delegation across three questions: (RQ1) how systems are operationalized (autonomy, task scope, interaction mode, and transaction capability/evidence realism), (RQ2) what facilitates or inhibits delegation, and (RQ3) what downstream outcomes follow for marketing performance and consumer experience. We searched Scopus and Web of Science for English-language, peer-reviewed primary studies (2015–2026) and applied conservative coding rules that distinguish claimed capability from simulated or demonstrated execution. The mapped literature is concentrated in text-based, low-autonomy assistants focused on recommendation and post-purchase support; coverage drops sharply for workflow-level autonomy, cart building, checkout/payment execution, and negotiation. Across studies, findings cluster into two motifs: a utility/assurance pathway in which performance cues and interaction quality increase perceived usefulness, satisfaction, and trust, and a governance pathway in which autonomy cues and system-initiated control trigger reactance/powerlessness and reduce acceptance unless mitigated by safeguards; urgency can attenuate governance resistance. Because most outcomes are intention- or vignette-based, calibration, verification, and error-recovery behaviors remain under-measured. Overall, delegation appears to depend less on maximizing autonomy than on coupling capability with user governance (consent, oversight, recourse, accountability), and we outline measurement priorities for evaluating execution-capable agents.
Stefanos Balaskas (Wed,) studied this question.