Shaping Sustainable Consumer Choices Through Agentic Commerce in the Grocery Retail Sector

Abstract

As grocery retail enters agentic commerce, consumer decisions may increasingly be delegated to AI agents that search, compare, recommend, and assemble shopping carts on behalf of users. Yet, little is known about how sustainable choices are shaped when grocery shopping becomes a human-AI interaction rather than an individual consumer decision. Building on practice theory, customer journey theory, and choice architecture, this study analyses sustainable choice-making through a qualitative case study of the ChatGPT-Instacart integration in the US grocery retail sector. Drawing on observations, interviews, netnographic material, and an expert interview, we examine how participants interacted with an AI agent across pre-purchase, purchase, and post-purchase stages. We find that agentic grocery shopping is shaped through the co-creation of meanings, competences, and materialities: participants bring existing shopping priorities into the interaction, the agent redistributes competences by taking over planning, search, comparison, and cart-building, and the interface structures what can be noticed, selected, corrected, or overlooked. We identify sustainability-related frictions across the journey, including trade-off, information, availability, effort, and outcome frictions. By showing how sustainability can be recognised in the conversational layer but weakened in the transition to product selection and checkout, the study contributes to research on sustainable consumption, AI-mediated customer journeys, and agentic commerce, while offering implications for retailers, platform developers, and policymakers seeking to make sustainable choices visible, actionable, and durable in agentic commerce.