Whatnot Accelerates Live Shopping Personalization with Strategic Acquisition of Real-Time AI Startup Shaped.

The dynamic landscape of live shopping demands a new paradigm of technological agility, a reality underscored by Whatnot, the burgeoning live shopping marketplace, in its recent strategic move to acquire machine learning startup Shaped. This acquisition, announced last week, is a clear indicator of Whatnot’s intensified focus on enhancing its product recommendation systems, particularly by addressing the critical factor of speed in delivering personalized suggestions to its rapidly engaging shopper base.

For Chief Product Officer Tom Verrilli, the pursuit of superior discovery and recommendation capabilities has been a consistent priority. Under his stewardship, Whatnot has already made significant strides, dramatically reducing the latency of its recommendations to mere minutes. This operational efficiency means the platform’s technology can now rapidly ingest and process real-time data – such as a product selling out on a live stream – and instantly recalibrate its recommendations to buyers, ensuring unparalleled relevance in a fast-paced environment. This constant drive for improvement in its recommendation architecture ultimately led Whatnot to Shaped, a company founded in 2021 that positions itself as a "real-time retrieval engine for search, feeds and agents," having previously lent its expertise to notable entities like Vox Media and QVC, including the development of a personalized news feed for New York Magazine.

The Unique Imperatives of Live Commerce Discovery

Discovery, according to Verrilli, has historically been a distinguishing feature for Whatnot. The inherent nature of live commerce presents a unique set of challenges for recommendation systems that far exceed those of traditional e-commerce or streaming platforms. Unlike static online catalogs where inventory remains constant for extended periods, live shopping environments are characterized by fluid, rapidly changing inventory. Sellers operate without foreknowledge of specific questions a shopper might pose, and buyers are often in a competitive rush, eager to secure coveted collectibles and rare items before others.

"Genuinely, the thing that sets live commerce apart and makes it a uniquely hard recommendations problem is the speed," Verrilli articulated. "Inventory changes second by second. We actually don’t know when a show is going to end, right? So you can’t necessarily turn through [things] like if you’re recommending movies." He elaborated on the stark contrast with a movie recommendation scenario, where a system might infer a user has an hour free on a Friday night and suggest long-form content. In live shopping, the unpredictable flow of a show means recommendations must be adaptable and immediate. Whatnot’s systems have therefore always been engineered for maximum speed, now boasting latency for recommendations down to "literally minutes," with the capacity to update model features at a minute-by-minute interval.

This emphasis on speed, or minimal latency, is paramount. Verrilli stressed its importance in maintaining Whatnot’s core value proposition to sellers: the ability to generate higher revenue and move inventory swiftly. "One of the things that makes sellers choose Whatnot more than anything else is that you can make more money on Whatnot than anywhere else, and that’s because you can be running auctions quickly," he explained. This contrasts sharply with traditional retail models where sellers purchase bulk inventory, upload it, and passively await sales. On Whatnot, items are expected to sell due to the presence of highly-intent customers and rapid inventory turnover. In such a dynamic ecosystem, recommending products based on yesterday’s sales becomes obsolete. Recommendations must instead reflect the current show’s inventory and the seller’s immediate offerings, effectively matching buyers and sellers in real-time.

Whatnot’s Pre-Acquisition AI Investments and Organizational Structure

Prior to the acquisition of Shaped, Whatnot had already established a robust internal framework for AI integration. Emmanuel Fuentes serves as the company’s VP of Data and AI, a role he has held since the early days of the company. However, AI’s influence at Whatnot extends beyond a dedicated department; it is woven into the fabric of nearly all development teams. Internally, AI tools are prevalent within the engineering organization, as is common in advanced software companies. Crucially, AI is also live and operational in a wide array of customer-facing tools.

Whatnot’s philosophy regarding AI is distinct: it aims to augment, rather than replace, the human element. "We don’t believe that the best use of AI is trying to replace the seller who is actually merchandising the goods," Verrilli stated. The company firmly believes that the human connection forged between a live seller and a buyer is the cornerstone of trust and expertise, a key factor in why buyers spend an impressive average of 95 minutes per day on the platform.

Instead, AI is deployed to alleviate the operational burdens faced by sellers. The selling process, with its demands for inventory uploading, management, shipping logistics, and timely packing, can be arduous. Whatnot has embedded AI into various aspects to simplify these tasks:

• AI assists in transforming an image directly into a product listing.
• It is integrated into recommendation products to ensure correct shipping labels and providers are used.
• AI is leveraged in scanners to help sellers efficiently mark inventory as checked off.
• And, as the central focus of the Shaped acquisition, AI is already integral to its existing recommendation system.

Shaped’s Expertise and the Strategic Synergy

Shaped’s specialization in "real-time retrieval engines" presents a powerful complement to Whatnot’s existing infrastructure and strategic objectives. Founded in 2021, Shaped has quickly carved a niche in developing technologies that allow for instantaneous data processing and highly personalized content delivery across various platforms. Its work with media giants like Vox Media, where it helped create a more personalized news feed for New York Magazine, and its collaboration with retail behemoth QVC, underscore its capabilities in handling complex, real-time data streams and user behavior patterns to optimize discovery.

The acquisition of Shaped is expected to deepen Whatnot’s ability to extract nuanced insights from live interactions and transactional data, translating them into even more precise and timely recommendations. This synergy could lead to several significant advancements:

• Enhanced Feature Engineering: Shaped’s technology can enable Whatnot to incorporate an even broader array of real-time features into its recommendation models, such as instantaneous changes in audience sentiment, specific questions asked during a live stream, or micro-trends emerging within seconds.
• Sophisticated Model Updates: The ability to update recommendation models at an even finer granularity than the current minute-level, potentially approaching near-instantaneous adjustments, will further refine the relevance of suggested products as events unfold in a live show.
• Scalability of Personalization: As Whatnot continues its rapid growth, Shaped’s expertise will be crucial in scaling these advanced personalization capabilities across millions of users and thousands of concurrent live streams without compromising performance or latency.
• Proactive Discovery: Beyond reactive recommendations, the combined technology could foster more proactive discovery mechanisms, anticipating buyer interests based on subtle cues and guiding them to relevant live shows or products before they even express a clear intent.

Broader Industry Context: The Live Commerce Boom and AI’s Role

Whatnot’s aggressive investment in AI-driven recommendations is set against the backdrop of a global live commerce boom. Initially popularized and scaled in China, where live shopping accounts for a substantial portion of e-commerce sales (estimated to be over $400 billion in 2023, according to Coresight Research), the model is gaining significant traction in Western markets. Platforms like TikTok Shop, Amazon Live, and Instagram Shopping are all vying for market share, making differentiated user experience, particularly through effective discovery, a critical competitive advantage.

Analysts predict the U.S. live commerce market to reach over $60 billion by 2026. In this burgeoning, competitive environment, companies that can most effectively connect buyers with products they desire, in real-time, will emerge as leaders. Whatnot, with its strong community and focus on niche collectibles, has carved out a unique position. By doubling down on real-time AI, it aims to solidify this position and attract an even wider audience and seller base. The general trend across e-commerce giants, from Amazon to Netflix, indicates a massive investment in AI and machine learning for personalization, recognizing its direct impact on customer engagement and revenue. Whatnot’s move aligns perfectly with this industry-wide strategic imperative.

Ethical Considerations and Seller Autonomy

The growing pervasiveness of AI inevitably raises questions, particularly concerning creative autonomy and potential misuse. Whatnot, acknowledging these concerns, maintains a balanced approach. Regarding AI-generated creative, such as show thumbnails, Verrilli noted that sellers currently have the freedom to create their own visuals to merchandise their shows. "We think it’s really important that sellers have the capacity to merchandise themselves and describe their show and their brand themselves. If they want to use AI tools to do that, I’m OK with that. We have no intent to try and manage that out of the system." While expressing a personal preference as a customer for "a real image of a seller and some of their goods," Verrilli emphasized that customer preferences vary, and the platform supports seller choice. This stance reflects a commitment to seller empowerment while navigating the evolving landscape of AI-assisted content creation.

Strategic Outlook: Prioritizing People, Empowering Experience

Looking ahead, Verrilli’s priorities as Chief Product Officer remain firmly anchored in the core user experience, rather than abstract technological goals. "I actually don’t have AI priorities – I have seller and buyer priorities," he asserted. This perspective frames AI as a powerful enabler for achieving business objectives focused on user satisfaction and efficiency.

For sellers, the ongoing objective is to "continue to take the work out of being a seller." This translates into continued deployment of AI across various operational touchpoints:

• Listing Flows: Streamlining the process of getting products online.
• Show Management: Simplifying the complexities of running a live show.
• Data Analysis: Empowering sellers with AI-driven insights to understand their performance better, identify trends, and improve their business strategies. This could include personalized recommendations for optimal show timings, pricing strategies, or inventory management based on their unique sales history and audience demographics.

For buyers, the focus remains on "discovery so that people can find great sellers." This includes:

• Enhanced Routing: Leveraging AI to more effectively guide buyers to the most relevant live shows and products.
• Post-Show Support: Utilizing AI to assist with any queries or support needs that may arise after a show, improving the overall customer journey. This could involve AI-powered chatbots for common questions, or intelligent routing of complex issues to human support, ensuring a seamless experience from discovery to post-purchase.

In essence, Whatnot’s acquisition of Shaped is not merely a technological upgrade but a strategic reinforcement of its commitment to delivering an unparalleled live shopping experience. By mastering the intricate dance of real-time data and AI-driven personalization, Whatnot aims to solidify its position at the forefront of the live commerce industry, ensuring both sellers and buyers thrive in its dynamic marketplace. The integration of Shaped’s "need for speed" with Whatnot’s existing infrastructure promises a future where discovery is not just efficient, but truly instantaneous and profoundly personalized.