Launched in April 2025, Etsy's "algotorial curation" system combines human trend experts with machine learning to build browseable themed collections — like "Modern Farmhouse" or "Literary Girl." Human editors hand-pick roughly 50 seed items per collection, then machine learning expands each collection to around 1,000 items tailored to each shopper.

Human trendspotters at Etsy identify and curate initial seed listings around editorial themes. Machine learning then expands each collection from roughly 50 to approximately 1,000 listings. Google's Gemini multimodal model is used to verify that expanded listings maintain visual and aesthetic consistency with the curator's original intent. Collections are then personalized per shopper based on recent browsing, purchases, and viewed items. In Q1 2025, Etsy launched more than 200 trend collections and 150+ hobby-based collections using this approach. LLM-generated image alt text, a byproduct of this system, reportedly yielded approximately 5% more visits from search engines and 3% higher sales conversions.