The AI personalization engine that rebuilds the storefront for every visitor - one swipe, click and scroll at a time.
Most online stores greet every visitor with the same homepage. Product Genius, a small AI company in Cambridge, Massachusetts, was built around the opposite bet: that the storefront should be different for each person who lands on it. Its software installs on a Shopify store and turns the browsing experience into something closer to a social feed - an endless, scrollable stream of products, videos, reviews and questions-and-answers assembled on the fly for the individual doing the scrolling.
The mechanics borrow openly from TikTok and Instagram. As a shopper swipes, clicks, lingers and scrolls, the system reads those signals and reorders what comes next. The company describes the engine as a "scroll-bot," and its stated goal is narrow and measurable: lift the revenue a store earns per session. Product Genius reports a median lift of about 20% per session, and in controlled A/B tests against brands' own websites, an average revenue lift of 36% across 25 retailers and millions of U.S. shoppers.
Behind the shopping feed is a research claim. The company calls its approach Large Interaction Models, or LIMs, and positions them as a counterpart to the large language models that dominate today's AI conversation. Where an LLM is trained on text, a LIM is trained on interactions - the accumulated record of how people move through stores. The pitch is that these models learn continuously, in minutes rather than months, and can work without the enormous proprietary data reserves that big platforms take for granted.
That framing matters because of who the company wants to serve. "Advanced AI shouldn't be limited to big tech," founder and chief executive Ben Vigoda has said. Amazon and the largest marketplaces have long personalized what shoppers see. Product Genius is trying to hand a version of that capability to independent brands running on Shopify - the kind of merchants who could never build a recommendation system of their own.
Figures reported by the company; A/B lift measured against brands' existing sites across 25 retailers.
Instead of a static homepage, each visitor gets an individualized feed of products, AI-generated videos, reviews and Q&A - reordered in real time from behavioral signals.
DTC and retail merchants from independent stores to enterprise. Featured names include Betsey Johnson, Pier 1, Dressbarn and Club Furniture.
Traditional stores show everyone the same layout, leaking revenue on shoppers whose interest shifts within a single visit. Product Genius aims to catch that intent before they scroll away.
Advanced AI shouldn't be limited to big tech. We're proud to be partnering with Google Cloud and the Shopify App Store.- Ben Vigoda, Founder & CEO, Product Genius
Recommendation tools are not new to ecommerce. What Product Genius argues is different is the unit of learning.
Conventional personalization often bolts a recommendation widget onto an otherwise fixed page. Product Genius reorders the page itself, treating the visit as a continuous stream of decisions rather than a series of clicks on a template. Its Large Interaction Models are pitched to update in minutes, learn across a shared multi-brand network, and rely on behavioral signals rather than third-party cookies - a design that fits a web moving away from cross-site tracking.
The company also leans on pedigree. Vigoda helped invent early tensor processors, the deep-learning chips that predate Google's TPU, and sits on a DARPA AI committee. The technology grew out of roughly $35 million in DARPA-funded research before it ever reached a checkout page.
Sold through the Shopify App Store · B2B SaaS subscription
Product Genius is truly revolutionary. We've been using this app for over six months and the results have been amazing.- Jeff King, Founder, Club Furniture
MIT PhD and deep-learning hardware pioneer; helped invent early tensor processors, holds 100+ patents and publications, DARPA AI committee member and Kavli National Academy of Sciences Fellow.
Grew an acquired startup to $5B at Office Depot; former Live Nation and Ogilvy executive; advisor across startups with more than $1.2B in combined exits.
Founded Endeca (acquired by Oracle for $1.3B), co-founded Toast ($20B IPO), and founded Parallel Wireless - a deep ecommerce and infrastructure track record.
Product Genius sits in the ecommerce personalization and merchandising layer - the space occupied by tools like Nosto, Rebuy, Dynamic Yield and Bloomreach, and a generation removed from the site-search engines that chairman Steve Papa built at Endeca. Its wager is that the next move isn't a smarter widget on a fixed page, but a page that reassembles itself for each visitor, learning from interactions the way language models learn from text - and doing it for merchants who don't have Amazon's data or engineers.
Ben Vigoda launches the AI lab, building on years of DARPA-funded research into interaction modeling.
The personalized shopper feed ships to Shopify merchants with tiered SaaS pricing.
The platform trains on 470M+ shopping interactions across a growing multi-brand network.
At NRF 2026 in Google Cloud's booth, the company announces LIMs and a 36% average revenue lift across 25 brands.
It provides a Shopify app that uses AI to give each visitor a personalized, TikTok-style shopping feed - mixing product recommendations, videos, reviews and Q&A - to lift revenue per session.
Ben Vigoda, an MIT PhD and deep-learning hardware pioneer who helped invent early tensor processors and sits on a DARPA AI committee, is founder, CEO and chief scientist.
They are Product Genius's core AI approach, positioned as an advance over LLMs that learns continuously from shopper interactions in minutes rather than months, without needing massive proprietary data reserves.
Publicly listed Shopify plans are Pro at $999/month, Advanced at $1,999/month and Enterprise at $6,999/month, each with a 15-day free trial.
The company reports a median +20% lift in revenue per session and a 36% average revenue lift in controlled A/B tests across 25 brands, trained on 470M+ shopping interactions.
Video links open search results and the official app listing; specific hosted videos may vary.