Alex Ruber practically grew up thrifting. His mother, an immigrant who escaped communist Romania and moved to Italy, then Canada, often brought him along to secondhand stores and Sunday flea markets when he was a child. Together, mother and son would hunt for unique items. “I remember getting my first piano literally from a flea market,” he says. “For me, it was like a treasure hunt.”

Fast forward 20 years, and Ruber, a former Apple software engineer now based in San Francisco, is the cofounder of a new AI-powered search engine platform designed to replicate the thrill of thrifting, but online. The site, called Encore, aggregates items from hundreds of resale websites and helps shoppers find esoteric and unique items—the proverbial needles in the haystack. What makes Encore different is that the site doesn’t just search for terms on Facebook Marketplace or eBay, but rather it asks the user to describe what they’re looking for in the same way they would describe it to a friend.

Thanks to its large language model technology, Encore lets shoppers run really specific searches, with queries such as: “dress like the one Carrie Bradshaw wore in season 6, episode 12, in a size 0 or 2.” Or “mid-century modern dining table in walnut finish but it has to have leaves to accommodate eight guests or more.” Shoppers can edit their search and type a follow-up prompt like “rectangular table only” or “under $1,500.” And if the site draws a blank, the user can toggle a button and search for new items.

The ultimate goal? “To become the Perplexity of online shopping,” says Ruber, who cofounded Encore with former Twitter and Asana engineer, Parth Chopra.

Shopping Spree

Everyone who loves to buy things secondhand has a reason for doing so. Some are looking for a bargain, others want to reduce the carbon footprint associated with big polluters like the fast-fashion and fast-furniture industries. Others yet enjoy the lower barrier to entry for luxury items. As a result, the global resale market is booming.

Encore launched in June and has 50,000 searches per month, with 25 percent month-on-month growth for searches. It is one of many companies trying to make secondhand shopping easier and more fun by providing a more refined user experience than the search aggregators that have come before. The Beni app, for example, lets you type “checkbeni.com/” in front of any product URL to see whether a secondhand version exists on various resale marketplace websites. Meanwhile, the Berlin-based Faircado has built a browser extension that lets you browse for items as you normally would, and pops up with “pre-loved” alternatives when they’re available. (The Encore team started with a website so anyone could access the site from any device, but they are also launching an app in the next few months.)

Encore is using a blend of GPT-4 and its own computer model, which is a fine-tuned version of GPT that the company trained on some fashion and ecommerce datasets so it could recognize various brands, styles, and aesthetics. People using the free version get 30 to 40 results per search; chronic shoppers willing to pay $36 a year (there are currently a few hundred of them) get twice as many results per search and a few other perks. But unless your query is overly convoluted (think “boxy bomber jacket, with elastics on sleeves, and make it like the one Tom Cruise wore in Top Gun 2), Ruber says free users will get the same—albeit fewer—results as paying customers.

So far, Encore shoppers are most interested in high-end fashion, furniture, and kitchen appliances, but users can also search for watches, books, power tools, games, and anything that could be conceived as pre-loved. Fashionistas can use Encore to search through Poshmark, The RealReal, DePop, and other apparel resale websites to see all the results in one place. Those looking for furniture will see aggregate results from AptDeco, Chairish, and Kaiyo, among others. Encore also aggregates results from the popular Japanese site Mercari and the French company Vestiaire Collective. Shoppers can untick some sites if they don’t want to see results from them.

Encore earns anywhere from 2 to 10 percent commission (depending on the partner site) on every final sale generated through the platform. But unlike Google Shopping and Amazon, which use various product-listing ads and sponsored ads to bump certain brands to the top of the list, Encore makes money only through affiliate links and paying customers. The search results aren’t dictated by SEO, so the only guiding factor is the quality of your prompt and that of the underlying algorithm.

Ask Me for Assistance

By leaning into conversational AI technology, Encore is making a bet that detailed natural language searches are the future of online shopping.

Signs are already pointing in this direction: In February, Walmart introduced a conversational AI called Ask Sam that can assist its staff with queries like “What aisle is the cinnamon located in?” or “How much is this hand soap?” Recently, Amazon launched a conversational shopping assistant named Rufus, which is trained on Amazon’s product catalog and can assist shoppers with questions, recommendations, and of course, purchases.

Elsewhere on the internet, you can get travel advice by phrasing your request the way you would to a (human) travel guide, or ask a search engine a question the way you would text it to a friend. Some experts project that, by the end of the decade, the conversational AI market will grow from $13.2 billion in 2024 to almost $50 billion in 2030.

Serious online bargain hunters can already do much of what Encore is achieving by downloading Chrome extensions and checking a barrage of advanced search filters on the popular resale sites. But Encore’s team believes its AI-powered platform will help remove that kind of friction and open up these marketplaces to more people. The team hopes that conversational AI can upgrade the shopping experience to one that facilitates discovery and allows people to browse more freely, just like they would at a thrift store. “Secondhand shoppers are really unique in that they love browsing a lot, and going down different rabbit holes until they find that gem,” says Ruber.

The team has other hopes for its platform as the AI tools it’s using become more powerful. For example, you could ask Encore to send you a text when that dress becomes available in your size, or even to buy it for you as soon as the listing appears. The Encore team also plans to double down on a “discovery channel” that is currently pretty sparse but will end up including inspiration boards, staff picks, and curated collections by micro-influencers and other tastemakers. “It’s as if you had your own personal product assistant,” Ruber says. “And not the annoying kind.”