Zalando's Head of Product Ania Szostek outlines how the company is moving from keyword-based search to conversational style agents powered by LLMs. The approach translates natural language fashion intent (e.g., 'outfit for a trip to Peru') into catalog-executable product constraints like material, color, and silhouette. Key challenges include measuring subjective 'taste,' connecting LLM reasoning to live inventory via RAG, managing latency, and avoiding over-surfacing AI features. Szostek also argues that Product Managers must become T-shaped builders capable of prototyping and owning evaluations alongside engineering and UX teams.
Table of contents
Key insights for digital commerce leadersFrom the front lines: Szostek’s perspectiveFrom keywords to style agentsThe new product manager: T‑shaped and outcome‑drivenMaking LLMs shop the catalogThe contrarian view: Challenging the status quoYour strategic roadmap: What to do nextExpert Q&AConclusionSort: