AI organizes large swaths of data—but platforms are avoiding an AI-forward customer experience.

Morning Brew Inc, Photos: Adobe Stock
Online resale platforms helped the secondhand market boom by digitizing thrifting, a previously IRL-only practice, and the incorporation of AI has allowed secondhand platforms to make online thrifting even more convenient for customers by increasing the information they’re able to provide, and—in some cases—keeping prices down.
Although AI is the backbone of many online resale platforms, customers aren’t necessarily looking for an AI-forward experience. Morning Brew spoke with three companies in the secondhand space about how they’re using AI behind the scenes while keeping the user experience human.
Kate Sanner, the founder and CEO of Beni, a resale search engine that can comb through more than 40 secondhand platforms, told Morning Brew that resale shopping traditionally swaps lower prices for a lengthier shopping experience. Beni’s model offers a twist: In exchange for saving users time, it takes a cut of the distribution fee it gets from resale marketplaces.
The tool, which can be accessed via an app or web browser extension, quickly locates listings “across 800+ dimensions” using an AI-powered ingestion engine, which can organize a million listings per hour, including those that are incomplete and unclear.
“How do we take what people are searching for, and figure out what the right match is, and then what we’re layering in for resales? How do we hold a real-time catalog of available inventory?” Sanner said. “That just is not possible at scale without AI.”
“Our goal really is to show up wherever you’re already shopping,” Sanner said. “We want to be the pipes into the world of resale to show you all your options before you make a purchase.”
Pickle, a platform that allows users to rent out and sell pieces from their closets, relies on AI for its SKU matching system. Julia O’Mara, Pickle’s co-founder and COO, told Morning Brew that the platform is able to connect users who are interested in a specific item from one person’s closet to the same thing in someone else’s. In exchange for playing garment concierge, Pickle keeps 20% of the rental cost.
“If you’re seeing a really cute red Reformation dress that you want to rent, but maybe it’s not in your right size, or it’s not something that’s available tomorrow,” O’Mara said, “we can recommend the same SKU that is maybe in someone else’s closet, even within your same city.”
AI has helped Pickle increase the company’s product velocity by “more than 10x year over year,” O’Mara said. An upcoming feature on the app will feed customer sizing data into AI to give users insight into how items will fit. Through user profiles, Pickle will gauge how an item will fit one person using data on how it will fit a different user, and compare sizes between brands.
“That is going to give our end consumer much more confidence in being able to pick an item that is really going to fit them,” O’Mara said.
And Pickle gets inspiration for new features from its users, so much so that O’Mara said she feels Pickle is “building directly with the consumer” and “adding continuous value” for them.
“They are massively driving the roadmap for our company and what we’re actually building,” O’Mara said. “[They] have often been a step ahead of the technology that we end up rolling out.”
Managed secondhand marketplace ThredUp couldn’t list 100,000 items daily without leaning on AI. Alon Rotem, the company’s chief strategy officer and chief legal officer, told Morning Brew that AI helps human employees with two key aspects of listing items: taking photos and pricing.
In the interest of speed—which Rotem said keeps ThredUp’s prices low—employees take quick, potentially imperfect listing photos, which are then cleaned up by AI. Then, ThredUp’s AI calculates the item’s measurements while the employee fills in details like color, brand, and clothing type.
“There’s a relationship [between] the human and the machine to make it right,” Rotem said. “Because at the end of the day, it doesn’t actually serve our business to misrepresent an item.”
ThredUp’s pricing agent then pulls from other resale marketplaces, retail sites, and other publicly available information to assign the item a price. The agent, Rotem said, ensures that items are priced accurately and fairly so they can sell quickly and give customers a “consistent experience.”
Automating parts of the listing process has allowed ThredUp to expand into more analog-style avenues for selling. In September, the company started selling on Whatnot, a live selling platform where buyers can watch a livestream of sellers auctioning off their items.
“It’s ironic that there’s so much AI being pushed and, at the same time, in the retail landscape generally, and now in secondhand, live selling is the thing,” Rotem said. “That’s so human.”