Etsy, the online marketplace known for selling one-of-a-kind handmade items, is hoping that artificial intelligence can boost sales of those crafty creations. The site has been selling less stuff and recently announced a plan to double down on high-quality and unique merchandise over cheap and mass-produced.

Now, it's launching AI-curated product collections, based on trends like island luxe or maximalism. They build on the work of human trendspotters, using AI to scan the site and tag thousands of matching products.

Nick Daniel, chief product officer at Etsy, explains what the company calls algotorial curation to Marketplace’s Meghan McCarty Carino. The following is an edited transcript of their conversation.

Nick Daniel: We have 130 million listings for sale on Etsy. That means that the most difficult thing for us to do is actually match a buyer with the thing that they're looking for. And leveraging technology, we're able to actually sift through all of those listings and really pull out those ones that are really inspiring for our customers and the ones that spark those new shopping journeys. One of the things that we learned in gifting last year was people love the ideas that we generated for our gift guides. They wanted thousands of listings, and doing that at scale is really difficult. We think we're there, finally being able to get thousands and thousands of topics in front of you with thousands of items that are high-quality.

Meghan McCarty Carino: Tell me more about how Etsy is integrating the large language model to kind of refine these collections.

Daniel: The thing that's the most interesting, I think, on that is how we've leveraged large language models just to know more about our inventory. That's been a core problem that we've been trying to solve for quite a long time. One of the things that makes us so different is that we are really seller-powered. We focus on our sellers. And where we've been able to leverage AI and LLMs is to help augment them, to help better identify the attributes that make the items in our inventory different or the attributes that our buyers would need to know in order to make a purchasing decision. And even just leverage LLMs to describe more accurately what this thing actually is. And I think that's how we're leveraging LLMs when we're thinking about discovery and curation, is with a more accurate and uniform description of what all of these things in the collection are — do these things make sense together?

McCarty Carino: So walk me through kind of what this will look like for the consumer because these are sort of collections that might have thousands of products in them, right? I mean, how is this being surfaced for consumers?

Daniel: What I would say is we're on a journey right now, and the journey is to become a starting place for special and the starting place for inspiration to really drive more consideration among our buyers. For now, where you'll see these collections most prominently is within our app. There are jumping-off points on the home screen currently. But our ambition is to ensure that we can seamlessly integrate these across all the touch points in your shopping journey, so that maybe you're looking at a listing and it's not exactly the thing that you're looking for. What we previously have done is just show you a lot of similar listings to the one you're looking at.

McCarty Carino: Yeah, you can kind of do a visual search and it shows you similar.

Daniel: Yeah, and that's great, and it helps people complete the mission that they're on right now, like they want to buy a thing like this. But what we want to do is make Etsy more of a destination for inspiration and broaden people's expectations in terms of what they can find on Etsy. So in addition to showing you similar things to what you're looking at, we can then expand that and say, hey, this thing relates to those trends that are hot right now, like you might want to consider shopping through these collections. And we found that people really love shopping through collections when they understand how the collection relates to what they're currently looking at and how the collection hangs together. So you should expect to see us leaning in much, much more, both in terms of this approach, where we're thinking about trends and topics and making it easy to shop these types of collections, but also just generally making it easier to discover great items across Etsy.

McCarty Carino: A lot of retailers are integrating AI into their online shopping experiences. How is Etsy's approach unique and kind of in support of Etsy’s unique brand?

Daniel: Where we've really focused is making it easier for our buyers and sellers to connect and secure a sale. So with the advent of better image-recognition tools and LLMs that actually can accurately describe images, we're able to deploy that against all of our images and fill in alt text for every single image, which creates a much better buying experience for our buyers on Etsy. But it also creates a better selling experience for our sellers because they don't have to worry about doing that anymore. So that's an example of how we're really thinking about, how do we leverage AI in a way that is directly applicable and beneficial to our buyers and our sellers in the here and now?

McCarty Carino: The thing that I am most often shopping for on Etsy these days is vintage lamps and vintage light fixtures. I'm obsessed with these, and you can find them from all over the world. How is this going to help me find more unique vintage lamps?

Daniel: So we've been talking about creating collections that are based on trends and interests. One of the other things that we've been investing in is a knowledge graph that is aimed directly at our buyers, trying to take all the data that we have about our buyers and ensuring that we connect those dots for them, and we are showing you things that we find are much more relevant for you. So in your case, I'm assuming you've been browsing around on Etsy. You've been looking at vintage lamps. You've been favoriting them. Maybe you've purchased some.

McCarty Carino: I have like a hundreds-long favorite list.

Daniel: That's perfect. So from that, we can now infer that you are really into vintage and lamps, and vintage lamps, like, up to varying degrees. One of the other interesting things, you should go check your favorites again because we recently released a new feature where for every list that you have, we autocurate those into different sections for you. So maybe thinking about different ways to segment your favorites and different topics that might relate to your favorites. And we also give a lot of recommendations of new items you might be interested in at the bottom of your favorites in collections.