Whatnot, a live shopping marketplace, announced the acquisition of Shaped, a machine learning startup founded in 2021 that bills itself as a "real-time retrieval engine for search, feeds and agents". The deal strengthens Whatnot's recommendation system, which Chief Product Officer Tom Verrilli has identified as a core competitive advantage. Whatnot's recommendation latency is now down to just minutes, allowing the platform to quickly ingest real-time information—such as when products sell out during a livestream—and use that data to inform what products it recommends to buyers.
Recommendations in live commerce are uniquely challenging because inventory changes second by second, and sellers must respond to customer demand in real time rather than relying on pre-purchased inventory. For commerce practitioners, this acquisition signals the importance of sub-minute latency in high-velocity selling environments. Whatnot emphasizes that buyers and sellers choose the platform because it can best match inventory to high-intent customers in real time, and faster recommendations directly support this value proposition by ensuring shoppers see what is actually available today rather than yesterday's sold inventory.
Shaped has previously worked with companies like Vox Media and QVC, developing personalized experiences such as a customized news feed for New York Magazine. The acquisition reflects a broader trend among marketplaces to invest in AI-driven discovery as a competitive moat, particularly in categories where inventory turnover and real-time responsiveness are critical to seller earnings and buyer satisfaction.