Whatnot buys Shaped as live shopping pushes harder into AI

Whatnot is buying Shaped, an AI startup built around recommendation and search technology, as the livestream shopping company tries to make its marketplace easier to browse in real time. For investors watching private consumer marketplaces, the deal shows how much of Whatnot’s growth case now depends on matching buyers with the right live auctions before they disappear.

Whatnot announced the acquisition Wednesday. The company said Shaped’s technology will support its personalization and discovery tools as it adds more product categories and serves millions of buyers.

Live shopping creates a harder matching problem than a regular online store. On a traditional e-commerce site, listings often sit in place long enough for search and recommendation systems to sort them. On Whatnot, a show can begin or end throughout the day, inventory can change second by second, and auctions may last only minutes.

A recommendation engine is software that decides what products, sellers or shows to put in front of a user. Whatnot wants that system to react faster as bids, supply and shopper behavior shift during live video sessions.

Emmanuel Fuentes, Whatnot’s vice president of data and AI, told TechCrunch that adding Shaped’s technology to Whatnot’s current systems should make recommendations quicker and more tailored to each shopper. He said live commerce is difficult because available items, shows and buyer intent keep changing during a session.

Fuentes also told TechCrunch that Whatnot has spent six years cutting the delay in its recommendation system from about a day to minutes. The company expects Shaped to help move those suggestions closer to real time. Whatnot said its systems handle more than 500,000 hours of live video and millions of real-time interactions each week, data it uses to improve recommendations.

What Shaped brings to Whatnot

Shaped developed machine learning tools for businesses that want personalized search and discovery. Machine learning refers to software that improves by analyzing data patterns. The company also used large language models, a type of AI system trained to process and generate language, alongside customer data to refine recommendations.

Shaped’s customers included Outdoorsy and QVC, according to TechCrunch. As part of the deal, founder and CEO Tullie Murrell will join Whatnot and run a new Applied AI Research group. Nearly a dozen engineers and AI researchers from Shaped are also joining Whatnot. Murrell previously worked at Meta before starting Shaped.

The acquisition lands during a fast expansion phase for Whatnot. Founded in 2019, the company recently said sellers on its platform have passed one billion orders. Earlier this year, Whatnot raised $225 million in Series F funding, valuing the company at more than $11 billion, after adding 20 million buyers over the prior year.

Whatnot has also been moving beyond its earlier strengths in collectible-style commerce. TechCrunch reported that the company launched more than 35 categories last year, including art, golf and vinyl. It added more than 45 categories in the first half of 2025, with additional subcategories still rolling out monthly.

The AI push is not happening in isolation. TechCrunch noted that resale and marketplace companies such as eBay and Poshmark have also been adding AI features across their platforms. For Whatnot, the bet is more specific: in live commerce, better recommendations could determine whether shoppers find relevant auctions while they are still happening.

This story draws on original reporting from TechCrunch.