Abstract

E-commerce is moving from search & scroll to predict & approve. As AI assistants increasingly act as the first buyer, websites lose relevance and decision systems take center stage. This session explores how agent-led shopping works in practice: from intent detection and shopping graphs to prediction, optimization, and human approval. It shows why retailers must shift their focus from storefront optimization to decision quality, providing high-quality data, clear constraints, and economic targets so AI systems can select them profitably. Drawing on real-world experience from large-scale pricing systems at Zalando and current work at 7Learnings, the presentationoffers a concrete playbook for pricing, marketing, and inventory in a post-website commerce world.