Can AI Spot Fake Watches?

Today AI can analyze luxury handbags, sneakers and other collectibles in seconds, if not minutes. So what actually change and how trustworthy is it? I am so excited about this episode cause today we welcome Kate Stewart, Chief Revenue Officer of Entropy.Entropy is the global leader in AI powered authentication for luxury goods, using advanced technology to verify handbag, sneakers, and apparel with 99.86 accuracy across more than 90 countries.So very excited to get into authentication and building trust on a global scale with AI with Jake. So hello, Jake.Hi, Danny, thank you so much for having me excited to be here. Of course, let's just start straight into it. So just to understand, serve the context, what has happened in the past five years? [...1.6s]Yeah, there's, there's lots of things, right, that have happened over the last five years, but just to maybe highlight a couple of really key things that have changed, especially relative to authentication and technology and its, and its acceptance within authentication.First and foremost, [...0.8s] authentication in the demand for authenticity, I think has shifted over the last five years.Something that used to be very back room and almost behind the scenes has become, because of consumer demand, something of the forefront, right? And that transactional trust and the request for authenticity from the consumer through their businesses has really increased as a whole.On top of that, in the background, companies like entropy have been building technologies that help connect those dots and really build out that ecosystem so that everyone from [...0.6s] social resellers to market places to pawn chops and everyone in between can drive that transactional trust through third party relationships and partnerships and partners like entropy.And [...0.6s] for us, [...0.5s] over the last five years, the evolution of data access, the improvements of [...0.6s] and the speed and efficiency of technologies have really allowed us to rapidly expand support and coverage and model refinements and adjustments at a much quicker pace than ever before. [...1.5s] So that's, it's really interesting.So I think first and foremost, it's sort of two different side. What you're describing is like kind of two different sides of it.First is in the past five years, [...0.6s] look, handbags, sneakers, everybody read about it during covid had huge booms, um, of involve, of market involvement, both of new entrance values increasing and quite frankly businesses, more small businesses, more businesses on different platforms coming into the space.Whereby I think to your point, [...0.5s] the, an older authentication needed to exist on a globalized scale to service those markets actually even becoming what they are. Cause without it, like how could you signal trust with, how could you signal global trust?But then on the other side of it, I think from what you're defining is sort of the, the technology side which take one part I've always been really fascinated about is, um, with AI and talking about the education, like a lot of the data of images and things like that existed, but they didn't have anywhere to be applied to. Am I understanding that correctly?And that's of what AI, AI is sort of like the first frontier of actually taking a lot of the data that previously existed and turning it into an outcome.Yeah, I think that's, I think that's a good summary and statement of that, really. What we are from an AI perspective is machine learning and that is existed for a long time.But what's really critical is that all of those images are that data, so to speak, right? It comes together and it gets built on a, on a platform and with a, a basis that can then take those images and do something with them.Yeah okay, it's really helpful. Um, so in that, in that vein, is AI replacing expert opinion or augmenting it?I, I think the best way to think about it is really augmentation as much as anything, um you know, I think, I think there's a misconception [...0.7s] around lots of AI, and it's different functionalities and, um, where and how it replaces humans.Does it in some cases, possibly when you're talking about authenticity though, and just in general, the AI is only as good as the humans behind it.Right? And so at the core of what we do, there are still lots of humans involved in the mix of things to train and refine and take machine learning and AI and what we do from a technology standpoint and ensure that our value propositions of speed and accuracy in those things, um, all happen. Right?And so for us, I think augmentation is the much better way to think about it. We are authentication layers for businesses, but then we are second and third layers for other businesses that have that human authentication layer. And so I think, I think for us, it's all about collaboration and augmentation more than anything [...1.6s] interesting.And I think that's sort of similar to what, um, exists in kind of the trading, what sort of the news and exists in kind of the trading card market is exactly that, that, um, to your point, like stand alone [...0.6s] AI isn't as much a value prop because it continuously like it has from involvement on both the refinement side and your point that being a additional provider of verification side.It's not like an and, [...0.6s] and that's the combination for us, you know, we're AI first in everything that we do, but we have humans behind everything that we're doing to ensure that the AI is doing things correctly.And if the AI, um, doesn't, if it sees something that is an oddity, it goes directly to robust and really big human layers of experts to be able to still deliver on our technology and on our promises and organization to our, our partners. [...1.1s]So to that next question, um, [...0.7s] my next question on sort of this is [...0.6s] you, a lot of times with sort of AI providers, what happens is you'll get sort of this notification of instant verification of authenticity.And if you're a reseller, a retailer, marketplace, or really anyone that's a buyer, anyone that's receiving this information or using this information, um, how do you know that it's, that it's trustworthy? [...1.6s]Um, for us, there's a couple of things that we've done [...0.6s] beyond the speed and accuracy that I mentioned. What we do really uniquely is that we financially guarantee our results out of the gate to help overcome that fear that you're, that you're mentioning or that maybe, [...0.6s] uh, [...1.1s] gap.We've, we've built a layer of trust with our customers and then hopefully that can be pushed through to their customers. So if we're ever wrong, we, we pay for it, and it's not the cost of the authentication, it's the actual value of the item.So we wanna make sure we stand behind our product and that our customers know and that ultimately the end consumer knows that if we give a result. And it's within that speed and accuracy side of things that we're gonna stand behind it and sort of be that insurance layer, so to speak. [...1.4s]Is there any other, um, if you're new to this and sort of looking into it, that's, I mean, that's a huge, [...0.6s] that's a, you cannot get a bigger trust signal than sort of guaranteeing your authenticity, but is there another, is there anything else you would advise someone who sort of looking at this of what they should look for from providers?I, I think accuracy is a huge part of it, right, depending on their needs. Speed can be a component of it. Um, more than anything, I would say, try it, test it. See see, see what works for your case, your situation in your scenario and your different use cases. And that's what we love to do. We love to put our technology, um, in, in customers hands and prove it all out.Cause I think, you know, proof is in the pudding as they say, and, um, there's no better way than to actually apply it. And we'll always happily do that for our customers and anyone to ask questions for that. [...1.7s]Okay, so then now on to, um, [...0.7s] condition grading. Um, is it, is condition grading in your experience ready to go?Is condition grading the next frontier? And for those that sit reminder condition grading, I mean, not trust whether or not something is authentic, but actually [...1.0s] giving, um, identifier, giving the, a numeric rating or identifiers of this is very, this items very worn or some kind of context of what the condition of the work is. [...1.2s]We've been thinking about a number of questions over the last handful of years beyond the question of, uh, is this real or fake?Right, that's the core of what we do. But by nature of how much data we have, how many images we receive, how much knowledge that exists on the products themselves that we answer that question for. We've also been thinking about the questions of, what is this? After knowing that, what is the condition of it?And based on that, then what is the value of it? And all of those sequential questions [...0.6s] have been a really big focus of ours over the last year.So in the background [...0.6s] and [...0.7s] to sort of in a long winded way answer your question, we recently launched a whole feature set that answers those questions called Market Edge wherein post knowing if something is real, we will also then tell you what it is, what its condition is, and in turn what its value is.We don't ever do anything if we don't have extreme high confidence levels and extreme accuracy levels in that.And so for us and the deployment of market edge in that feature set, the answer is absolutely condition is ready. It's available. It is going to continue to improve and be enhanced, but we feel really confident in that whole market edge side of features. And now those questions being able to be answered through technology. [...2.1s]It's one of the things I think is most I'm so like, personally the ability to do that with three 60 objects blows my mind that you're doing this with handbags and sneakers, but also for as a collector, like what you're effectively walking through [...0.6s] previously [...0.5s] had about [...0.6s] anywhere 4+ different touch points to it of physical receipt of item, view of item price guide tooling to then try.There's not a price guide tooling for your market, meaning it consolidate a place of historic results.You then go on [...0.5s] a very interesting journey of trying to look at sold results in different pieces, then trying more about doing that to then get context of whether or not those results because you don't, you only have flat images and no condition grading on those results to then see whether or not you actually have a true comp.So this is like, I am always fascinated by where this is going, cause it's not, it's not sort of seconds one back for collectors. It's hours, days.It's, it's just, and, and for resellers, obviously, um, resellers, literally you go from for a lot of them sort of having their hand side of how I even start my business, then having business.But it's it's not, it's not seconds. It's truly like hours, days and so much mind space that can be freed up. And that's why, that's why we built that went on that journey to do it beyond having all the data, was all about the efficiency that we could help our customers gain.Um, I still think there's lots of collaboration to be had globally within that feature set condition here means something very different, uh, when you talk about Japan, for example, as a really unique condition market.And so there will be a lot of, I think customization even within our market edge feature set and within grading specifically that will all come together, though I think really beautifully to just continue to enhance what that experience looks like, and hopefully drive confidence and efficiency throughout people's processes and systems across all those questions and how to answer them. [...1.7s]So [...0.8s] as far as, okay, so that's where condition greeting is going as far as categories go. Are there some categories that are further along and more developed, and others which are still difficult?Like, for example, let's just take, I know you're not talking about, you know um, [...0.6s] right now, it seems like if you have [...0.7s] memorabilia that sort of, let's say music memorabilia that was in a concert, or it's autograph memorabilia from sports memorabilia, that's autograph from certain game.Right now, sort of the verification process, all really great businesses, but separate tends to, it can have often separate businesses that do certain parts of that verification.Um, so to that point, I'm just curious, like, what categories in your mind, [...0.5s] um, have you seen a sort of more developed, what are less or what's driving that?Yeah, I, I think it's a whole the categories where there's lots of data, right, [...0.8s] those are the [...0.7s] call it easiest or at least the most accessible to go and deliver these different feature sets around.So things like bags and shoes and apparel and even jewelry, um, watches, somewhat [...0.7s] trading cards without autographs very much so, right.But when you start to get into really limited subsets of, of goods and really small amounts of data wherein [...0.5s] you can't or don't have comparables, that's where things get really challenging.And a lot of that expertise isn't anywhere but in someone's head, right. Um, and, and absolutely things like memorabilia things that look at autographs and really one off sort of scenarios and situations, those are still difficult because there's limited amounts of data and systems need data [...0.6s] to be able to really produce the [...0.7s] efficient results through technology.Now I think there still can be signals and there still can be ways to combine those two things, right, both the AI side and the human side for some of those struggle categories, but it, but it all comes down to the amount of data available [...1.1s] fascinating. Um.So [...1.1s] I think to, [...0.8s] I think you kind of answered this in that last question, but for those is for those sort of think of AI as a threat to traditional authentication jobs and opportunities and [...0.6s] what would you sort of say to them? [...1.3s]I, I would say that it's, it's an opportunity, not a deterrent. And when I say that, what I mean is AI and the combination of AI and humans, I think creates even [...0.6s] stronger trust layers and more efficient and improved results.And [...0.7s] ultimately the consumer is starting to demand that in lots of ways, right? They want, they want authenticity and in lots of categories that's extremely valued.You don't want bias in that, they want third party results. And so I think more than anything, there needs to be collaboration and augmentation of those things.And there needs to be a view of how do I, how do I improve, [...0.8s] you know, a, B, C, and d in this technology help me do that versus ever, ever looking at replacing something. [...1.3s] So last question is, um, in 10 years, um, what is authentication look like for you? [...1.1s]I think, I think for me, what that looks like is [...1.3s] more broad appeal to authentication, [...1.0s] more authentication across virtual ecosystems and in businesses and even consumers hands [...0.5s] to be able to really drive that.I mean, there are wild, wild Wests of goods being traded across places like Facebook Marketplace and even Instagram and classified all over the world. And authenticity isn't a component at all to those scenarios, right?For the most part. And so I think I think beyond the integration of technology across all these different ecosystems, there will be the consumer demand that marries up to that.And, and there will be consistency across, [...0.8s] hopefully any channel [...0.6s] where and you can have those questions answered through technology.I think that's, I think that's the direction it's headed. And I think there's probably lots of unknowns about what more we can provide from an information perspective and how and where Providence becomes a component to it in a much bigger way. But I think the stories can now be told more about the products in their lives.As, as this becomes deeper and more embedded into all of these different touch points for our, for our shopping experiences. [...2.9s] The great answer, um, I think we'll leave it, we'll leave it with that.Thank, um, thank you so much, Jake. And as a reminder, Jake is the Chief Revenue Officer, entropy will put Entropy's website in the notes and yeah, I can't thank you enough.This is fascinating to sort of [...0.8s] dive into AI and authentication and get a real picture, sort of where it's going and exactly to your point, how it's a collaborative process with the ecosystem not, not a frictioned one.And, um, [...0.7s] like I, as a user and serve bystander, I'm excited, just so excited to see where this is going.So thank you, thank you, Danny. Appreciate the conversation. Thank you and thank you to everybody for watching built to collect. And if you wanna learn more about, uh, conversations like these where trust and technology intersect with collecting, follow the show. And thank youyou