Watch. Comment. Buy. AI Powers Live Shopping’s Encore
Live shopping is the television home shopping experience rebuilt for the internet.
A host goes live on a platform, demonstrates a product, takes questions from viewers in the comments and offers a limited-time deal. Viewers buy with a single tap, without leaving the stream. The format collapses the distance between discovery and transaction to mere seconds.
In China, live shopping generated roughly $900 billion in sales in 2025, nearly the size of the entire eCommerce market in the United States, consumer intelligence firm NIQ estimated in a Friday (July 17) press release. In North America, by contrast, 68% of consumers said they’ve never bought anything through social media.
A format that drives hundreds of billions of dollars in Chinese eCommerce has yet to gain any real footing in the U.S.
The gap is not about the content. It is about discovery. Thousands of live streams run simultaneously. Inventory changes by the minute. Auctions end in seconds. A viewer who joins late misses the product that would have converted. Matching the right buyer to the right stream at the right moment requires personalization at a speed that static recommendation systems cannot deliver. That is the problem Whatnot, a live-shopping marketplace valued at more than $11 billion, is trying to solve.
Whatnot processes more than 500,000 hours of live video and millions of user interactions each week. On Wednesday (July 15), the company acquired Shaped, a machine learning startup that builds real-time recommendation and search infrastructure, TechCrunch reported Wednesday. Financial terms were not disclosed.
Shaped Founder and CEO Tullie Murrell, who worked on recommendation systems at Meta before founding Shaped in 2021, will join Whatnot alongside nearly a dozen engineers and researchers to lead a new applied AI research group, the report said.
Live Shopping Requires AI to Match Buyers to Streams in Seconds
The challenge Shaped must meet is specific to live commerce. Unlike traditional eCommerce platforms where product catalogs are relatively stable, Whatnot’s inventory is in constant flux, according to the report. A buyer’s interests can shift during a broadcast.
“By combining Shaped’s technology with Whatnot’s existing systems, we can make recommendations faster, more responsive and more personalized,” said Emmanuel Fuentes, vice president of data and AI at Whatnot, per the report.
Shaped’s systems bring together customer data with AI models and machine learning to offer personalized search and discovery, the report said.
The combination could help Whatnot suggest the most relevant streams to users based on previous purchases and real-time behavior signals.
Whatnot launched more than 35 new product categories in 2025 and more than 45 in the first half of 2026, with sellers surpassing 1 billion orders, according to the report.
Each new category adds complexity to the discovery problem. A shopper who lands on a live feed at the wrong moment—after an item is sold or before the item they want appears—gets little value and might not return.
AI Personalization Could Close the Gap Between Content and Purchase
The PYMNTS Intelligence report “The Agentic Commerce Deep Dive: Payment Infrastructure and the Path to Trust for Merchants,” a collaboration with Visa Acceptance Solutions, found that 48% of online shoppers used AI to research their most recent purchase.
Meanwhile, AI-referred traffic to U.S. retail sites grew 393% year over year in the first quarter of 2026.
The consumer appetite for AI-mediated shopping is building. Real-time AI can surface the right stream for the right buyer at the right moment and update as behavior shifts within the session. By acquiring Shaped, Whatnot is betting that solving the discovery problem is what moves live commerce from a niche format into a mainstream channel.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
