Whatnot Buys Shaped to Power Live Shopping Recommendations
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Whatnot acquired Shaped, a machine learning company specializing in real-time recommendation and search systems, to strengthen discovery and personalization across its live commerce marketplace.
- Shaped founder and CEO Tullie Murrell, previously a Meta employee, will join Whatnot to lead a newly formed Applied AI Research group along with nearly a dozen engineers and AI researchers.
- Shaped combines existing customer data with large language models and machine learning for personalized search and discovery; its client roster included Outdoorsy and QVC.
- Whatnot has spent six years improving its recommendation engine, reducing latency from roughly a day to minutes, and its systems process more than 500,000 hours of live video and millions of real-time interactions weekly.
- Whatnot sellers have surpassed one billion orders since the platform launched in 2019, and the company raised $225 million in Series F funding earlier this year at a valuation exceeding $11 billion.
- Whatnot launched more than 35 new product categories last year and 45 additional categories in the first half of 2025, including art, golf, and vinyl.
- Emmanuel Fuentes, Whatnot's VP of Data and AI, called live commerce "a uniquely hard recommendation problem" because inventory, auctions, and buyer intent shift continuously.
Why it matters: Live commerce is uniquely hard for AI because inventory and buyer intent shift by the second. Whatnot is buying both the tech and a former Meta engineer to lead its new Applied AI Research group — consolidating its recommendation stack as eBay and Poshmark race to integrate their own AI.
