- The RealReal is one of the world's largest online marketplaces for luxury secondhand goods.
- It developed an AI program to write clothing descriptions and uses AI to detect fake luxury items.
For "CXO AI Playbook," Business Insider takes a look at mini case studies about AI adoption across industries, company sizes, and technology DNA. We've asked each of the featured companies to tell us about the problems they're trying to solve with AI, who's making these decisions internally, and their vision for using AI in the future.
The RealReal is a fashion website that lets customers buy and sell luxury secondhand goods. It's one of the biggest names in the online, resale sector, with almost 20,000 items uploaded to its website every day.
Writing copy and noting attributes of each item take a lot of "human horsepower," Samantha McCandless, The RealReal's chief merchandising officer, told Business Insider.
"We had to unlock how we could scale the business, ensure it was cost-effective," McCandless said.
The company turned to artificial intelligence to streamline the upload process and address one of the biggest issues for high-value resellers: authentication.
Every piece on the website needs to be authenticated, as fake goods proliferate the secondhand-luxury market. The RealReal has previously come under scrutiny for its authentication processes. In 2019, following news stories about counterfeits appearing on the website, investors filed a class-action lawsuit against the company, claiming that the authentication process "fell far short of its description." The lawsuit was settled in 2021.
The new technology helps "ensure that we keep our brand promise and things are 100% authentic," McCandless said
Key staff and partners
McCandless said the project was developed internally. The company's software engineers built the AI solutions to work alongside preexisting teams of copywriters and authenticators.
AI in action
The RealReal started using artificial intelligence in 2018 with a program designed to speed up the writing of descriptions for newly uploaded items. It built an AI model that could scan images of clothing and note attributes, such as patterns and colors, to accompany the listings.
McCandless said the process involved a lot of testing, as the model could miss nuances. For instance, it initially struggled to correctly identify fur and faux fur. "We had to tinker," she said. "It's really an iterative process."
Now, The RealReal is focusing on developing the next stage of its automation process: authentication. The company developed two AI programs. First, a machine-learning model called Shield determines which items should be prioritized for human authenticators. Second, its Vision tool examines images of items and predicts the likelihood that they're counterfeit.
"It helps filter out things that we need to pay extra special attention to, or things that are an easy, 'Nope, that's not authentic,'" McCandless said.
Humans are still part of the authentication process, which helps to manage issues such as false positives.
"It just shifts your workforce so you lose some of the mundane tasks, but you move to higher-level tasks," McCandless said.
Having a strong database of clothing items and descriptions was crucial to the implementation of the projects. "The more items that we've pulled into our platform, the smarter and smarter it's going to get," she said.
Did it work, and how did leaders know?
McCandless said implementing an AI process was different from other product development because there's no clear end and the company had to constantly evolve the model. The systems are continually being accuracy-tested by The RealReal's quality-control team, and the authentication process will become more refined as testing continues, she added.
According to the company, The RealReal has identified more than 200,000 counterfeits since its inception in 2011, using a combination of human authenticators and artificial-intelligence technology. The RealReal has said its Vision AI is used on 50% of the handbags uploaded to the website.
What's next?
Looking ahead, McCandless said she's excited about AI's potential to create a personalized shopping experience. The end goal is to get the right products and offers in front of the right people.
"With the generative AI that exists today, there's so much more we can do to really, truly one-to-one personalize that experience," she said. Previously, The RealReal used an algorithm that predicted preferences based on shopper demographics, and McCandless said that there wasn't much behavioral information.
"The new AI really allows you that behavioral stuff and the learning, and I think we'll get better and better at that."
We want to hear from you. If you are interested in sharing your company's AI journey, email jhood@businessinsider.com.