The End of Generic Shopping: AI’s Next Move in Retail

In this episode of The Ultra Commerce Podcast, Jamie sits down with AI expert Marco van Herpen to explore how artificial intelligence is reshaping the future of ecommerce. From hyper-personalized shopping experiences to redefining digital product advice, this is a must-listen for anyone navigating the new AI-powered retail landscape.

What Marco is working on:
👉 www.newtonic.nl
👉 https://www.eyewear.stylecode.ai/

Discover what Ultra Commerce is doing straight from the source:
👉 https://ultracommerce.co/platform/order-management
👉 https://ultracommerce.co/platform/product-information-management

Creators and Guests

Host

Jamie Schouren

CMO at Ultra Commerce

Editor

Constance Jamotte

Guest

Marco van Herpen

Founder & Managing Director at PTTRNS.ai

What is The Ultra Commerce Podcast?

The Ultra Commerce Podcast offers practical insights tailored for both B2B and B2C eCommerce operators and developers. Each episode features expert guests who break down real-world strategies and share success stories from the digital commerce space. The show places a strong emphasis on the operational side of eCommerce, focusing on systems architecture, fulfillment, and data orchestration, going beyond just marketing and storefront design. Hosted by Jamie Schouren, she invites guests to deep-dive into the complex world of digital commerce, exploring the strategies that power successful eCommerce operations. If you're interested in the nuts and bolts of enterprise eCommerce from systems architecture to fulfillment and data orchestration; this podcast offers smart and actionable content in an accessible format.

Jamie (00:01.154)
Hello and welcome everyone. I'm back to already the third episode of our podcast. This one is going to be an amazing one. We're going to talk about, you know, the most talked subject today in tech and that is of course AI. But we're not just talking with someone out there who thinks he's an AI expert because they can write some prompts, but we're actually going to talk with a real AI expert who's been in this field for years, even before.

majority knew what was going on. Marco, come and help. I'm very, very happy that you're here. Would you mind, would you introduce yourself to us?

Thels, thank you for that introduction already. Well, and thank you for the invite invitation. Yeah, as you mentioned, I've been working in this field since I would say 2017 roughly. So a little bit of my background is that I come from a different field. actually got a PhD in business in economics. So it's from a different area. Was in strategy consulting for about 10 years, then in e-commerce for about five years and started in 2005.

15, I was getting more involved in new tech, very interested in what was going on with visualization techniques such as virtual reality and augmented reality. And a friend of mine, an old colleague who knows a lot about this new technology was very interested in all the breakthroughs that happened in 2015 in the AI domain. So we started a company in 2017, trading neural networks, pioneering, doing both

projects and creating some products. And recently we started a new company together with other IT experts called Newtonic, trying to build on our past experience, but specifically also focused on combining AI and e-commerce to make sure that companies that are maybe struggling a bit with the AI strategy, struggling also with how should we

Marco (01:58.808)
Where should we do things ourselves? Where should we use off the shelf AI agents, et cetera, and making it very pragmatic and starting helping them to start work on it. So that's what I do.

Wow, that's super cool, Marco. And I've heard a lot about cool things. I heard augmented reality, virtual reality, AI, and I heard you say even more of those kind of words. And these have been evolving in the last, I would say, months, months that have reached the big public. And of course, before that, also other experts in the field. In terms of for e-commerce, we've seen

things happening in AI and I see mostly what people are doing is using chat bots, using them for bots, for chats and customer service kind of things when you are like in the B2C sector, for example. And of course we have the content creation part, right? We see so many content creation being done by AI. But AI is much more than that and it's called artificial intelligence. And I know that you have a different view on what intelligence is and I would like to hear more about that. But what is it that you can really do with this?

Of course, it's nice to optimize our customer service and to downgrade our content production work, or actually upgrade the work and downgrade the time that we spend on it. where is that real value? What do you see? Tell us about your journey, your products, your vision.

Thanks. Well, I actually, think to best explain it, I would like to explain a little bit how I look at the e-commerce industry and what is really happening in retail e-commerce domain over the past, I would say 15 years and how AI can basically take that one step further. So what you see is I tend to think of things along the lines of what the consumer experiences are and where the value added at is of a specific business or an industry.

Jamie (03:48.078)
Traditionally, if you look at a store, the functionality of a store is that they are showroom, they give advice, you can buy, so you have the purchase functionality there. You have the curated assortment that was offered. So you also have a logistic function there. And these functions were combined in one physical building. And what e-commerce has done is, to some extent, is take some of these value edits.

ads and take them out of the normal physical store and bring them to online. I don't know if you remember this, but a lot of time people were talking about, well, a store is just a showroom now and where you get some advice. And if you want to buy, you can go online and buy the product. And the reason why I go back this far is that if you look at this taking out different value ad steps and you can actually build businesses or build modules on that. And I think that

what AI is doing right now is taking some of these value adds and making them really come to the consumer in digital way. So let me explain what I mean with that. And I'm gonna give the example of a product that we built actually already, I think five years ago, something like that, along those lines was we were involved in a project with Hans Anders. It's a Dutch optical chain and we helped them give a

an advice function on to sell, for example, sunglasses online. So again, if you go back to my explanation of the physical store and the online, if you buy sunglasses, you would rather go maybe to a physical store to see which sunglasses are out there, but to also try them on and see what fits your face. And still, at least for me, maybe it's easier for other people and maybe for you, but for me, it's still quite difficult to.

then if I am in a store and I see 200 sunglasses in front of me to know which one would actually fit me. I have no clue. And a salesperson would come up to me and say, okay, you should really try on these glasses. And what a good salesperson would do, he or she would be trained to have a look at my personal style, maybe the clothing I'm wearing, et cetera. But specifically also look at my jawline, the shape of my eyebrows, the undertone, the warmth of my skin.

Marco (06:07.016)
the contrast in the face and those type of elements would determine what she would then show me and because there are some harmony rules about which type of frame would fit my face. That's the advice function I'm talking about. So again, this physical store is a showroom and these functions bringing them online is very difficult. So but with AI it is really possible. So what we've built was, okay, take one picture of your face when you're online shopping for new sunglasses and we get

think it would 28 features be distilled from this picture. So your jawline, your eyebrow line, all these elements that a good salesperson would look at to come up with which frames would fit your specific needs. And within a couple of seconds, we then can suggest to the consumer, you should really try one of these five frames. These are the five that would fit your face. And still you can say, I want something that's a bit more bold or outspoken or actually something a bit more reserved.

So you can actually also put your style into it. And this is something that AI is bringing right now to the digital space, so to e-commerce platforms. But actually what it's doing now is taking one of these really core elements that was still in the physical store, so get good advice, helping you in your shopping journey, that's being brought online now. And I think this is the biggest change we are seeing emerging right now.

that is still up for grabs and is still a bit uncertain. I think if we talk about it a bit more, we can maybe get to a sort of vision on it, but it's still uncertain who's going to take this domain. Is it a retailer like Hans Anders? Is it going to take that domain or is maybe Google or JetDBT or Perplexity going to take this domain? Who's going to give you this good advice?

And the results with Hans Anders, we're still working together with them, but we are also working with some other e-commerce, eyewear sellers, big companies all over the world on this and with the solution. But this solution is really something I think that is a good example of how AI is changing the shopping experience.

Jamie (08:22.166)
Is it something that we can call, and I would like to make it like an umbrella work for that because this is around eyewear, but about advice and really fitting you. I think this is something that I would like to call real hyper-personalization because it's about you. I see, I read these reports about what people are because I'm very interested in content as a marketeer and how can you profile people and what can you do with that?

think that one of the interesting things that I see is that people really trust AIs. Of course, they have a mistrust around, my data being used, but they trust them. And you can see that there's a lot of conversations in AI going about psychology, about things like, can you help like therapeutic kind of things, or doctor advice and medical advice and really psychological help advice and things that you sometimes don't even trust your partner with.

These are things that you, but people do trust a machine because they don't feel judged. They don't feel somebody has an opinion about it and can change their opinion about you when you would say something about that, for example, to your partner or to someone you like or you love. And this hyper-personalization, I'm not talking about, you know, use depression for selling them, I don't know, sunscreens or something like that, but it does show how we can, I think, in the future, really understand much better the people that we're working with.

And the people that we are trying to sell to, or the people that we're trying to do it. And I think one of the simple things that people do is allow access to your photo album and we can see your style, we can see what you like, and we can offer things to that. But there's always a GDPR thing there, but as long as it's not, as long as that can be contained, that can be worked around, if people can work there. Hyperpersonalization is obviously, I would say a front end thing, right? It's with the customer, but it also happens.

a lot on the backend because I know you're working also with products for optimization, for assortments, for product categories, product management, et cetera, with this AI, with this new intelligence.

Marco (10:26.19)
You raised a lot of interesting points. So I think that first of all, think the term is spot on. This is really hyper personalization, helping consumers in their shopping journey. And I think AI is very well suited for it. The second point you raised was, do we trust AIs? And I think you're spot on. We trust AIs with all these that you gave the examples. And

If you can make a comparison, I think it's the same way that Google made a breakthrough in 2000-2001, was being a search engine that is objective or clearly indicating when they have paid search. And that's how Google conquered this market. think JetGPT and Perplexity and some others are doing the exact same thing. You trust the advice, you can give some examples, you can give some questions to this AI.

And they come up with suggestions what would fit you or what would help you or what would suit you or what product would fulfill this need. And I think that we already see this being announced and being introduced in the United States to a much less extent right now in Europe is that they actually say, okay, we can help you find a solution and a good product, but we actually are becoming a shopping platform by ourselves because, and there's a big...

threat and opportunity coming from that. The reason why they can do that and take that position, I think they will take the position in the future, is because we trust them. Just like the example you gave. And this is where you go to the e-commerce platform and what a retailer or an e-commerce company can do with this new technology and not being threatened by the JGPTs and perplexities of this world. They still have the knowledge of a specific segment and all the

data of the specific segment. So if you go into how intelligent is an LLM really in the core, I would still argue that at the current state, the intelligence is limited. It's still a prediction engine with very good understanding of our grammar rules and linguistic rules. But it still relies on data to summarize it and present it in a coherent way to us. And so giving it

Marco (12:49.144)
good instructions and directions will still make a large difference. So going back to the case with the eyewear, if you would ask right now to a Chattie PT or perplexity which frames would fit me, you will get some generic response to the extent of if you have a round face, you'll probably need glasses eyewear that's also a bit more round. Fine. But if you want to really implement it in a way that we have developed a solution,

you would actually go much deeper, on the understanding the consumer, but specifically also on understanding the product. So you need to know of each frame, how round is this? And what is really the undertone of the color? And for instance, those type of elements, you need to specifically train and help an AI to do it in the right way. So there's still a big opportunity to distinguish yourself from this, what a generic platform can offer, but you need an

e-commerce environment where you can get all this detailed product information in and that is flexible in offering modules to the consumers to have this interaction with this really detailed product information. So then you can see the competitive difference.

But I think it's the same as in real life. mean, if I go to a sports shop, where there's sports goods and stuff are being sold, and they also have sunglasses, there's obviously someone who can tell me, yeah, this one fits you. But they have some, because these people have seen them. And then there is going to a really optical, I don't know, optician, I think it's the name in English, going to an optician who actually are specifically trained for that. And if we train people or we train this AI, I don't think, because they do both do the same thing.

They somehow use intelligence, of course, but they use the information that we put into them, coming from books, coming from internet sources, coming from courses, trainings, whatever, other people, seniors, that give them that information. And then they do something with that. They make, let's say, the calculation as a computer does, or as a person in their head to give that advice. And I think you're definitely right. So the specialist, that is something that you can add and that you can offer to...

Jamie (15:04.866)
give that, you know, instead of a sports shop selling sunglasses or optician selling sunglasses. And that goes for everything. Of course, you know, that's not just sunglasses and that even goes, you know, for medical advice. And, you know, we talked about this yesterday that I had a friend who went to a doctor and, and she was a bit like, yeah, I had this kind of weird rash on my hand and she wasn't sure about it. So she made an appointment with the doctor and the doctor

took a picture of it, General Prexler took a picture and asked Chad about it as well. So he his own opinion and he asked them and they were like a bit weird. They were like, hey, but that's weird. I could have done that myself. And the answer basically is of course is yes, you could do that yourself. But as you mentioned yesterday, there is this thing, know, Chad is not always right and the doctor is not always right. But if they have like the same percentage of wrong margin and they then combine it and they get other insights, that is a very valuable asset to add to that.

And there's even like something like a pre-selection, know, things you can just rule out if you use this kind of technology or people you can help who don't need that, you know, that deeper help, but just you can go with that more generic advice there. But then you also mentioned, but Jamie, there's a difference between real intelligence. And I would love to share that because I love that statement, how you said that. And I really like your vision on that. And I would like to share that with the audience as well, because I think that's very valuable.

Yeah, so the thing I'm pondering on this is right now is what is real intelligence? What is so? And to what I'm seeing right now that where we are as a state of the industry is that a lot of the LLMs are currently not capable of doing reasoning on their own. can, you can give them really good input. You can give them really good directions and they will based on that

give really good output, but that's something different as being able to reason on their own. And so it is a very powerful tool to help you digest large quantities of data. It's also a very good tool to create large quantities of text, et cetera. You already mentioned it yourself. Those things, it's very capable of doing that. But still, that doesn't mean that there is true intelligence to the sense that the way we are

Marco (17:22.369)
just by having this conversation come up with maybe new ideas, being creative on that. And that is something, step that we haven't taken yet as a society or as an industry or as how you want to call it. And that is something that we're still discovering. Having said that, it is brilliant to see what is capable right now. And I think we're actually on the other end. Maybe are overestimating the amount of intelligence, but we are underutilizing

the intelligence that is in there. So the way I sometimes describe it is like somebody came up with this way of, it feels like somebody put a lot of Lego blocks on the table and we're still looking at it and someone put four blocks together and said, I created a house. And then everyone's looking at it like, we can create houses now. Yeah, but you can make everything with Lego. And it's the same here. We can still use it even if it doesn't have this reasoning in it. If we do a little bit more of the thinking,

and let AI do a lot of the work, you can get to very nice steps. But that is the way I look at it. And it is a technology. don't think I can give a lot of examples in the past where new technology was adopted by the audience in such a rapid way as what is happening here. So people have adopted it. They are using it. They are trusting it. They might be overestimating the reasoning and the intelligence that's really in there, but...

There are so many new opportunities and you just have to be have a flexible infrastructure and guide it. And then you can get to very huge results that we've been discussing thus far. So that's my thing.

Yeah, we need more creativity then. I think, you know, we're not going to be outrun by AI as people. think as long as we have that creativity of thinking of more than building a house with four blocks, but you know, thinking of building other things as well. And I think there is, there's a big future there for the creative people. And as long as I create podcasts with intelligent people, AI is never going to be more intelligent. So I'm very happy. But besides that, so, you know, talking about future, what is the future? What is coming?

Marco (19:28.788)
It's so difficult to predict. think one of the things that is going to happen is that, and I'm going to limit myself a little bit to the e-commerce domain if that's okay. So one of the things that I really see happening is that a lot of the shopping functionality will go to the LLM platforms. So JetTPT and Perplexity is going to help consumers. So you want to buy.

Stupid example, you want to buy a lawnmower. So then if you want to buy a lawnmower, can just ask ChatGPT, I've got a lawn of 200 square meters. There are some trees in there, which lawnmower should I buy? Or what should I look for? And then they can give you functionality you should look for. You can say, okay, based on this functionality, which products would you recommend? Where can I buy them right now and have them installed even tomorrow? And what is the, you give me some advice where I want to buy it? So you can have that.

the entire interaction in one platform. But the next thing that's going to come is that basically you're going to say, okay, then use my account to buy it and have it delivered tomorrow. And the platform will take that role. So that means that to some extent, not for everything, ChatGPT will then place that order in such a platform. And the role of the e-commerce platforms will there diminish at that point. I think that there are some spaces where this will not happen.

It will not happen in marketplaces, my expectation, because marketplaces by themselves are combination of where people and products come together. So marketplace will probably use LLMs on top of it to have the same type of interaction within the marketplace. And it will not happen in really specialized environments, as we talked about before. So there will be a push from it to my expectation is that AI will.

These platforms will take a huge role in diminishing some of the aspects in e-commerce, but the capability of AI will enhance the quality of the shopping experience online, making it, and that's something that will happen, making it even harder for physical stores to survive and to have a real added value in this new environment. So that's.

Jamie (21:39.97)
And I think the example that you give with the lawn mower, you know, it's easier to take a picture. This is my lawn. What do I need? These are practical items. And I think for practical items, that's going to work. And that's already works. mean, that's fine. That's something you can do today. There is the emotional part, obviously, of buying and selling. And as a woman, and as a mother even, I'm even more aware of that. You know, I like to watch clothes and jewelry and, you know, shoes and baby stuff and all that kind of things. These are emotional products because you don't...

need them, cannot ask, chat, I'm in need of something emotional because I don't feel well. And now I want some to buy something for my child. You know, that's, that's not something you can for now, but I do think, you know, when we, where we started with, with hyper personalization that is going to happen. Even that one is then my doom scrolling that satisfaction that I get from, know, putting things in a card and not buying them or, just looking around that can even come from there. And there was, think a big interesting gap there even now to.

to fulfill an element that you're not going to chat GPT directly, know, the chat GPT app for that, you are going for the lawnmower, but I think there is some opportunity for platforms to offer that to use these engines, but to offer that doom scroll. I don't know, doom scrolling sounds so bad, but like that emotional shopping, you know, fashion, that kind of things, you know, or home decoration.

You know, just looking around and getting inspired. And I think in combination with my social profiles, my images, my pictures, pictures of my house, that can be such a nice one where I can build my profile on the things that I like and have my, let's say, personal shopping assistant there. But not only shops practical, but also shops or shows at least that satisfaction.

And what is going to happen? I agree. what's going to happen? So for instance, right now you will, if you go shopping for a dress, you will go to a store, an e-commerce platform, you say, I go to category dresses, and then you do put some filters on and maybe a price filter on and you get to a product. If you change that shopping experience within the platform, so forget a little bit about chat, you PT, but if you change the platform, basically saying that in the search engine, you can just type in, I'm looking for something nice to wear.

Marco (23:48.846)
Yeah, for a wedding. And then you come up with suggestions and you start an interaction with this consumer. The consumer will share more about themselves and it's then also easier to then say, but if you go to a wedding and you have this dress, you should really try to combine it with this type of purse or with this type of belt or whatever you want to do. And it becomes much more natural. And actually what's going to happen is that's my, it's the same, what a good salesperson in the store would do. You would have this dialogue, this value exchange of information and

It's easier than to upsell and cross sell. And this is much better than buying a dress online and then going to the shopping basket and then seeing other people's also bought and then you see some advice that's not really helpful that doesn't convert that well. this is what's going to happen online.

And they can even combine it, know, reverse that influence engineering. So, you know, now influencers promote you, follow them and they show you a dress, but they can do it the other way around because then they can show you like, hey, and Jamie, look at this, this is the dress. And here is actually some pictures of girls, know, or women your age or whatever, you know, your category who are wearing it. Girls that I might not know that are not, you know, more famous influencers, but I can see, you know, the real people that are using it because they can use.

Obviously there are engines to search the images out there on Instagram to find the song dress.

or even create an AI image of a person with a dress, with a belt, with a purse and just show what it would look together.

Jamie (25:21.502)
But Marco, let me tell you one thing. I'm using AI pictures of myself and I look good on them. Well, that doesn't mean that it's reality.

That's reality, that's reality.

Okay. Well, Marco, we are half an hour already. I think you can talk for hours and hours and but I would like to know one thing and then we're going to close it and then we're going to do follow-up podcasts on this one because it's a really interesting topic. So, and I think we should do things like, you know, AI and now we talk more about consumer, but let's, you know, talk about how for B2B and how this works. And we do that in follow-ups, but Marco, one last thing. So if somebody wants to try this, wants to brainstorm this, can I call you? Of course. Can I send you a message? Yes.

course, it would be really nice to have a conversation with anyone who wants. It's such an interesting domain and there are so many new developments. And as you said, over the past months or even weeks, you get new input. And I think that just having these types of conversations, everyone learned from it. So yes, please. It's been a pleasure having this conversation with you as well. It's always a pleasure. so thank you very much for that.

Thank you. We will share in the show notes, will share your email address. We will share some links, websites, some examples of what you've been working on, the ones that you shared. And it's been really, really nice to have you on our podcast, Marco. I am very looking forward to the reactions, everyone. Let us know what you think of it. Share your thoughts with us, share your thoughts on AI. Maybe you disagree fully. I mean, let us know as well. And of course you're welcome to be a guest in the show if you want to share another opinion on this. Marco.

Jamie (26:54.978)
Thank you very much. And I'm gonna see you in a follow up!

Thank you, my pleasure.