EP355 - Seeing the Future vs. Timing It, with Matt Nichols of Commerce Ventures

0:00 Welcome to the Jason and scot Show, your source for the latest news and trends in the e-commerce industry. Featuring host Jason Retail Geek Goldberg, Chief Commerce Strategy Officer of Publicis, and scot Wingo, co-founder and CEO of ReFiBuy. Here are Jason and scot. Hey, everybody. Welcome to this joint episode of the Jason and scot Show and Retail Gentic. I'm one of your hosts, scot Wingo, and I'm joined by my always present co-host, Jason Goldberg. I feel like this is my first time on Retailgenic. I feel honored. I have a bucket list item.

0:40 We featured your NRF keynote, but it wasn't like an interview style. But you didn't tell me you were doing that. Yes. Yeah, I did. You were wearing a mic. You know what was going on. And then for the third time, we're excited to have Matt Nichols. He's a general partner, Commerce Ventures. And give us a little blurb about what Commerce Ventures does, and then we'll jump into it. So Commerce Ventures is a venture capital fund based out of San Francisco, been around for about 14 years, investing out of our fifth fund, which is about $160 million. We're called Commerce Ventures, which kind of says what we do. We invest in commerce tech, which we really think about as retail tech and fintech or infrastructure for retail and financial services. So that's what we do. I lead all of our retail tech investing. So most of the things I invest in are being used by retailers or brands selling anything from sweaters to travel to hotel rooms.

1:29 Yeah. And you actually come out of the retail business. It'll be a, I don't know if you and Jason ever compared notes to who's been in retail longer. You're way more youthful looking. So if you've been in it longer, I'm pissed. I have not. I have been, yeah, I've been investing in it since 2006 is when I first started investing in it. Yeah, but you were slinging diamonds before? I was slinging diamonds after that, actually. I invested and then started slinging diamonds for five years before we sold it. Oh, okay. That's right. Yeah. Yeah. Awesome. Well, a lot, it's kind of funny, you would almost, in the Jason and scot show, we kind of like wait two years to have people back. So some time has passed. But in our AI era, I feel like it's been six months and so much has happened. And this week has been, I think yesterday was the biggest news week I've ever seen. We had new model drops.

2:11 We've had Amazon blocking Muse. We had Muse announced like six connectors. There's a lot of rumors that Chachapi T's coming out with a personal agent. Let's do kind of a litmus test. How are you feeling about, you know, you guys are obviously an investor in us, and you've made a bunch of other bets on Agentec Commerce. How are you feeling about that whole thesis? And what are some of the new things y'all are thinking about there? Yeah, I mean, I think we think about what is AI doing to retail and I kind of bucket into two categories. Like, how is how is AI changing how consumers consume? There's like that thread. And then there's how is AI changing how retailers retail? So I think the first is more fundamental, right? It's like, how are people fundamentally shopping for stuff? And that's where a lot of this news has happened.

2:57 So we've got a bunch of theses in those areas, as well as some more tactical things about how retailers are using AI. I would say in the first category of how consumers are consuming, we've been pretty steady in our optimism around agentic commerce, whereas I feel like the market has gone in these waves. First, ChatGPT and Perplexity were really aggressively positive. I think we think they were overly positive, not because of the opportunity, but rather the complexity of actually making agentic orders work. I think people were too optimistic about how easy that would be. I think they realized it wasn't as easy as they thought it was going to be. They pulled back pretty aggressively, probably more aggressively than I think they meant to or that sort of indicative of the long term future.

3:40 And now I think we're coming back with a different model. Some of these Muse and some of these other platforms are bringing some of that excitement about the agentic ordering opportunity back to the forefront. So I think the market has been a little wave. Dave, we still have remained sort of super confident in the long term, but sort of pragmatic about what are the building blocks and hopefully companies we invest in, that are required to actually make this vision a reality, knowing that it's kind of harder than the hype cycle might have thought. So, yeah, just yeah, we obviously talked a lot about you're alluding to ChatGPT. Which, of course, aggressively launched Native Commerce and then pulled back my own personal read. Like everybody underestimates how hard this is for sure. And so I totally agree with your, your hypothesis there.

4:25 I actually don't even think they pulled back because they suddenly discovered how hard it is. I think they pulled back because other parts of their business were under stress and they had to write the ship when ChatGPT and, or I'm sorry, when, when Anthropic and Google started getting traction on them on the quality of the foundational models. I think that's probably true. And honestly, as bizarre, we all think about retail every day. I, I think for a while, and probably still, they may think that the retail category is too small. Relative to your point, relative to all of the other big opportunities, is this the one we want to go dinosaur it for today? Probably not. I think there's a piece of that. I totally agree with that. scot and I often talk about what's going to disrupt Amazon. And a weeding hypothesis is retail is just too small to be interesting.

5:11 Yeah. What do you think about, okay, so then on these personal agents, I think it's interesting. We spent all this time thinking about these protocols and making sure all the machine-to-machine worked. And then what's really caught fire with the consumer is basically browser use. Yep. So all these agents, what happened is one of the reasons we did these protocols is the models were not very good at figuring out checkouts, right? And also we didn't – I don't know why this has just happened, but the Stripe Link thing wasn't really wired the right way back in that first kind of era, like call it 2025. But now you have the Stripe link piece, which just makes it way more secure to pay. And then also you have the browser use just took a huge step up, you know, in like the last year. And that's unlocked things. Unfortunately, it makes it more easy to block.

5:54 Right. So that's why we have Amazon blocking Muse. What do you what do you guys think about that? Is that the right move by Amazon? And then does do we end up with this balkanized thing where, OK, if you use your personal agent this way, Now you can't really have it work on Amazon. And what does all that look like? I mean, I think it makes sense for Amazon. They have more to lose than they do to gain from everybody else. So as an initial reaction, you know, blocking it, I think makes sense. I think they will probably find a way to work with a lot of these personal agents and find a way for people to browse Amazon in a way that's beneficial to them, gets the data that they need, the money that they need, whatever those set of parameters are.

6:37 I think blocking is probably the strategic way to start to have those conversations. But honestly, I actually think Muse and Instinct, I think more of the traction in the agentic commerce world is actually not coming from browsing catalogs and buying sweaters. It's from booking your haircut and doing a bunch of services where actually that checkout piece isn't quite as core to the experience. So I think it's still going to be hard to really get a great agentic commerce transaction through the consumer checkout for a core like catalog based merchant. I don't think that they fundamentally solve that. I just think it's been really useful for some of the other.

7:17 Less complex, and honestly, even categories where there isn't an e-commerce transaction at all. The fact that they can interface with a human also is beneficial. Yeah, no, I 100% agree. I think agent-based commerce is a real thing. And in the long run, I'm excited about it. I don't think it's the biggest or first use case for agents. Yep. So I sort of think consumers are going to discover other quality of life improvements for the agent that are kind of the first movers. And I do think the browser as a tool for the agent, That's just an interim kludge. Like in the long run, this is going to be protocols or API to API. It's just too expensive. The tokens that they burn through for all these multimodal, you know, and it's fragile.

8:00 So I don't think that, to scot's original question, I don't blame Amazon for blocking them. Amazon's the incumbent in digital commerce that 40 to 50% share. If you have 2% share, you can let a thousand flowers bloom and try everything. But Amazon has an incumbent position to defend. And I don't think their plan is to keep any of those people out. It's to negotiate the most favorable terms and let them in in a compute efficient way when they have an agreement. Yeah. But also, I'm excited that Muse and all these guys have come out because I think agentic commerces need this kickstart. Right. I think the retailers have underinvested, probably rightfully so, because there wasn't any volume.

8:40 So if Muse can like kickstart the volume of agentic transactions, that will kickstart the retailers from investing in infrastructure to take the more structured data and then the right way can exist. Like it's been this chicken and egg problem. So maybe this is the unlock on that chicken and egg. I also think they have made the Kluge better, right? So we've actually had browsers with agents for a while. I mean, comment and all these things. And it just, the Muse browser is in order of magnitude faster and better. Yeah, I think I think the spark to your point is it lets the consumer imagine what a new experience would be like and how powerful it is, where you just say to this thing, go take care of a bunch of stuff for me. And it just kind of does it with you, you know, at the right point, it asks for permission.

9:23 That's really, really good in that way. It's not pedantic with questions like the Google type things are, but it's also, you know, instinct drives me a little crazy because it's like just a shade too aggressive. Like, it's like, you know, I went ahead, like, it's like right on the cusp of canceling something. I wasn't really quite ready for it to or something. But it's like, I think they're, they kind of, they're betting in the future. And probably that's what that, the next six months looks like is it's weird for it not to be so proactive. Yeah. But I actually think some of the categories that will happen even before buying sweaters and jackets is travel. I think travel is one of those use cases where you're actually seeing a higher and higher percentage of the consideration happening on AI, more so than buying stuff.

10:05 And I think there's a bigger demand for consumers to use that as a research and booking tool. I think that'll actually kickstart more quickly than buying stuff. I have a secret for you on an investment theory. Go to the big voice telco companies that provide all these call services and find the industries with the longest call times. Because I think everything that you had to spend half an hour on the phone to do is the most vulnerable thing to these agents. Totally. Yeah, I agree. Yeah. And a lot of people, I don't know if it's out of the box, but Meta can go make some phone calls for you. The Muse as well, which is well. Yeah. And I think today they're actually launching, they have an event where they're going to launch the glasses that have Meta Muse on them.

10:46 So you'll have your Ray-Ban glasses. And then if that works well, then you can imagine them coming out with a speaker format type device. And then it's going to get really interesting. And then back to Amazon, I think that's really bad for Alexa. If some of these companies can come out with a device that takes all the modern technology and puts it into a speaker and makes it even portable, that's going to be pretty interesting as well. Voice is honestly the unlock for... Building these networks that were historically not connected with software. So it's like it's using voice to go out. It's also for businesses that had voice that couldn't pick up all their calls, just using this AI that just works. And we're seeing companies today that like historically 60% of the calls could be handed by an agent effectively without the consumer really caring about it. Now it's like 95.

11:36 So it's just the technology has gotten better and the consumers or this is not the, you know, zero, zero, zero to get the operator thing anymore. Yeah. I also, I think another interesting thing about voice, when the first iteration of Alexa for shopping, which was voice shopping in the, on the Echo device came out, I was super skeptical, right? Like people aren't going to yell across their living room, like order a Lily Verbana size six, a leopard skin dress in a, you know, with this promo code, right? Like it just, that, that wasn't very realistic, but what's happened now with all these other use cases for multimodal AI is we are all learning to talk to the robots. And I think that's a game changer. We used to type two and a half words into Google. Now we type 28 words into the prompt and it's easier to say those 28 words than it is typo.

12:19 Yeah. And, and, and particularly for like a problem statement, right? Going, we've talked about this in the past, like before you had to take problem statement, translate it into four keywords into product. And then how do I use my product back to solving my problem, my, my, my problem statement. Now you're like, here's the problem statement, okay, here's the product you need, here's how you use the product, go, right? The marginal gains for giving it a better problem statement, for giving it more context, are huge. And so again, it encourages more of these long form inputs. My hypothesis is just it's making us all more comfortable talking to the robots. All the advanced AI users I know are using transcription. And I think that's going to unlock more common consumer behaviors down the road.

13:00 And it's changing how people are searching for things. Look back to the travel example. I was talking to someone the other day, and he was like, my criteria is I play basketball with my son. I want to find a hotel with a basketball court. Yeah. Historically, imagine trying to find that on Expedia. Yeah. No way. There's no structured data for that. You have to go to Chicago and these dates and then go into every single hotel and find out whether it has a basketball court. But now you can search on, find me a hotel in Chicago with a basketball court, and it works. And I think we're joining rooms because I have a family. There's all these use cases where now people can search. Or a room far from my family. Exactly. But it's also creating this challenge, I mean, to what you guys are doing and a bunch of customers. Like, this data didn't exist.

13:43 It wasn't a top-level search. So in this case, the hotels didn't care about it. But now they have to because consumers are searching against that long-tailed it. And it's got to be right. I think it's a big opportunity. The next vertical for a refi buy. Yeah. So let's tie that to investment. So your job is to take all this soup of things going on and find companies to invest in. What are some areas you guys are poking around there? So I think one area, we talked a lot just now, we talked about how consumers consume and how retailers retail. We have a thesis that honestly, I don't think is shared by many VCs. I think that there is going to be a next generation AI first ESP email.

14:24 So a next generation Klaviyo. Start from scratch. And I think this applies actually to most marketing categories. Our thesis has been every single marketing tool that retailers have used in the history of the planet has been a workflow tool. It allows a 26-year-old marketer to create a segment, to create a piece of content, maybe test those, maybe see the results, maybe do better. In a world where you can use AI to create this content, create these segments, create these flows, I think a brand new start from scratch marketing platform, I'm interested in email because everybody hates their email eventually. It's also been a great category for VCs to invest in. It's a category where if you were to start from scratch and not add AI to Klaviyo, but start from like first principles, how do I make this thing fully autonomous?

15:05 So you don't have consumers in the loop. And now you can treat every consumer differently, give them all content, which honestly we just was not possible for. So we have this thesis that there'll be that big opportunity for retailers. Yeah. One of the things, it's really interesting. We struggle with this at the user interface for the retailer, the operator at the retailer to use at ReFiBuy. So when we build agents, it's a lot like GrokBot in a way in that it's almost more like managing humans. It's like a Monday.com type interface because the agents are there doing their thing. And then we watch their state and if they need to talk to us, they send notifications and then they pause if they need human input.

15:46 And we found early on in our first design partner programs, when we exposed that, people were like, I don't know what to do with this. So then we put a SaaS interface on it, right? So the SaaS interface is just, basically between you and the agents. And it gives you that old style SaaS interface. Then we introduced a chat bot, and that's basically talking to the agents. And now customers adopted that pretty well. At some point, I think we're going to get to more. I think I'm excited by Muse and these things because if people use it in their personal life and GrokBot, then we can get back to that AI interface. Do you think retailers are ready for that or do you think we'll be stuck in this SaaS world for a while? I think we'll be in the SaaS world for a while. I still think there's opportunity for SaaS. Because in my mind, when I hear your email thing, that's what it would look like to me. I have a campaign bot.

16:33 I have an email analysis bot. I have a promotion, whatever it is. And then I have one that is managing my list and deduping, things like that, like a list-managed bot. That's my – if I kind of started from native principles. But I'm not really sure people are ready for that. I don't think – so I think that's why I think there's an opportunity for a company to go build something here. I mean, what you want is I want to replace my email. What's interesting is it's not just replacing Klaviyo. And I apologize for saying this. This is replacing Klaviyo and the agency that was running Klaviyo for me. It's a labor and technology arbitrage. You're not doing it because you want to save labor. You're doing it because you can increase your revenue per email.

17:14 So I think there's this weird opportunity to build both those together. But it's a complex system. And you could probably stitch something together to do some of this stuff. But there's so much opportunity. And there's also so much data you can gather. So the beauty about this opportunity is like there's data sets of emails that every single retailer has sent out for the past 20 years. You can get that data. Imagine sending an AI to like, here's the list of possibilities of how people will send email. And by the way, if everybody sends it in this format, that probably works. So it's one of those things like versus in historically you hire this great 26 year old email marketer. He or she, yes, here's some emails that I've sent my last two jobs and they're awesome. But like their wealth of knowledge is what they've done.

17:56 And now the wealth of knowledge is the Internet, which is the power of A.I. Probably. So, yeah, well, the good news for me, if you could call it that, I'm way more likely to say something career limiting and get fired before agencies become obsolete. So I don't lose too much sleep about that one. I I'm buying your hypothesis on, on the native retooling winning over the bolt on. Yep. And I think another reason that's true is the legacy solutions also still suck. They do. Like, like we, I mean, we talk about best of class, but the best of class solutions today, you know what they still do? They still like don't personalize on open. And so, you know, someone opens that email two weeks after the sale that it's promoting and it's, it's stale content and they bury all the most important attributes in an image that you can't. read with images turned.

18:43 There's so much opportunity to make it more efficient. And my view is the next gen solution is going to be hybrid. It's a new solution, but it's still going to cater to cases where there's a robot at the other end of that email and where there's a human at the other end of that reboot. And maybe there's even handshaking and a broker, right? Like maybe I go, all right, I'm sending this to someone. They've outsourced their inbox to a bot. So I'm sending them the feed and this other person hasn't and so i'm sending them a human readable email, yeah but even even if you never believe that the ai is going to consume the email which they will obviously yeah like it's i mean no one at this table is opening their own emails anymore.

19:24 And yeah and your point about like everybody hates their email if you looked at the nps for every retailer about their email, the slope over the course of time they have an email provider is always down yeah it's highest when they signed up and then it declines declines declines declines like they finally give up and like i gotta go someone new yeah so yeah so fatigue yeah the grass is always browner any other, the other the other area that's interesting is i think about categories where there is a i call it a high take rate intermediary in travel that's expedient booking in food delivery is doordash and instacart, and i think all of those platforms are at risk of being disintermediated.

20:04 Because if you think about what all of those do, they did customer acquisition, the golly's customers. They had a unified interface so that I didn't have to change how I bought from Kroger or Lucky's or whatever the thing was. And I had the delivery mechanism. You still need to, AI is going to be tougher on the delivery. But the first two are going to be done by the AI platform. And by the way, because of most of the value they provided was those first two, they were able to charge 15% to 30% of GMV. In a world where people are going to want to order their Chinese food that they normally get or book their Marriott hotel on ChatGPT. Is there enough dollars in the ecosystem for one of those intermediary players to pick up 20%? No.

20:48 But is there a need for aggregation? Like in all those cases for the AI experience to be great, you need to have all the hotels, the big ones and the small ones. You need all the restaurants. And you need to be able to inject an order into all of those. So I think we have this thesis, there's an ability for someone to create a new intermediary that's doing the aggregation. It's a lighter weight. A lighter weight for way less money. So, you know, call it 5%, 1%. Somewhere in that range where it does the aggregation because I don't believe that ChatGPT is going to, like, look up for every MCP server on every single restaurant in the world. Yeah. Because they're not going to have them. Same thing for hotels. Yeah. So. In that version, I mean, is Amazon a high take rate intermediary? They are. Yeah. But I would argue that they could still – they are also the aggregator.

21:32 They are also the aggregator. Yeah. But they also – I feel like the complexity of them doing the delivery part and how good they do delivery is probably more durable than – Yes, 100%. Interestingly, their biggest, most long-lasting mode is going to be the development center and the delivery system, the DSPs. So this thesis is really interesting. The frustrating part about this thesis is you go invest in one of these companies. We've done some of that. Time for Dyson. They build the demand, and then you're like, all right, ChatGPT, hit my server. It's crickets you kind of it's this waiting game you know what that'll have to happen because you know that this is the answer but there's nothing you can fundamentally do to get chat gpt or clod, to hit this mcp server first yeah i mean so it's a waiting game so we're confident but it's it's a frustrating game because you are at the mercy of them doing something and to your the muse point maybe muse and some of these other these other entrants accelerate that thing right so it's like you're not relying on just a couple models yeah i feel like that's the hardest part of your job is not like seeing the future, it's predicting the timing of the future. Oh, totally. Yes.

22:36 We've been certainly early, too early on some things. You need the agent to be so pro-consumer that it tries new things if it will save time and or money or both. And Matt, that's probably why you shouldn't take investment advice from an AI maximalist. Well, it does go to show distribution is the key. And that's why Muse is going to, I feel very bullish about Muse because they have, what, four properties that have billions of users, and they can, you know, they're apparently, I don't spend a lot of time on all of their properties, but they're maxing across all of them to, you know, they realize they're seeing something in the data that is very positive, and they're leaning into it hard.

23:16 I mean, I think they saw what Instinct was doing, you know, and it was working. So, yeah. But I think part of these agents will be like, how much are people willing to give over in terms of power control risk? Because the more you give over to them, the more magical the experience is, but the more downside you subject yourselves to. And I think Instinct was lucky in that they were a startup. They could go full beans, right? Yeah. Like, we have more to lose than we have to gain. I'm impressed by Facebook because they have a lot to lose. Sure. We have a lot to gain, but so we'll see. Do you think Claude and OpenAI come out with counters to all this? I can't imagine why they wouldn't. If the innovation is the text interface, just add a text interface.

24:01 And I think people would prefer to have one agent, like one primary agent. You spend all this time giving data to those things, which is why I'm probably more optimistic around in a bunch of these vertical categories where you've had vertically specific agents in travel, in fashion. Lots of fashion. I think, yes, those experiences today are more rich and more like an e-commerce experience, but they don't have the historical data. So they have this onus on them to get all of this data to be powerful. I think the fact that the big AI platforms that already have a lot of your data, I think they're more likely to build that interface than the other way around. So, A, 100% agree. I think there was a time when every time someone came out with a better foundational model I was jumping ship because they were but now, they're all great and I use whichever one has the most context about me because that's the the switching cost now is not the the better model it's the it's the context, do some of these interesting consumer players like Meta that have tons of context that wasn't originally for AI, but that they can apply to AI.

25:07 It feels like the Amazons and Meta with tons of consumer data about a billion consumers is a leverageable asset. It definitely is. But I think also even individual retailers, I think that they have data about consumers, right? It's not as broad as those things. But for a specific journey, they have very relevant data about what you or I bought. And I think there's going to have to be, I talked about this on stage yesterday, I think there's going to have to be some sort of brokering between the big models and the retailers, because each of them has interesting information about a consumer for a specific consumer journey. And how do you bring those two together to deliver a really powerful journey? But the rules of how that will happen, how does that compensate, I think are still early.

25:51 And honestly, it's so early that we haven't even seen startups doing it. But I think we will, particularly with the guys who have lots of data, all of the folks that have big loyalty programs. They're trying to figure out how to make that loyalty program and loads of data, work inside these AI models. And we're early, but I think there's exciting stuff there. And then another, so you guys have a couple legs of areas you invest in. The other one is more on the fintech and payment side. Anything that you're poking around on that? Are you big believers in stable coins or all these different things? We are definitely big believers in stable coins, right? So we've got quite a few investments in that category. So it's one that we're super focused on. But, you know, I think a lot of the things that are happening with agents and some of these questions about security are probably even more problematic and like opportunities inside of financial institutions, right?

26:38 How do I get agents to work on my behalf in financial institutions? Or how do I create, you know, thinking about financial planning and budgeting? Is there an agent-based thing versus a mint or, you know, like that category of folks that changes how consumers interact with their money using AI? I think the answer is yes. What that looks like is still early, but yeah, it's a bunch of opportunities there. And then you're sitting in the heart of Silicon Valley there. Has the number of startups that are kind of approaching you guys, like what's that looked like over the arc of the last two years? Where are we at 11, a fever pitch of 11, or has it calmed down? It's still a fever pitch of 11. Honestly, I don't think in this category in retail, it's just a more interesting time to be investing.

27:24 I think the landscape is now shifting, which is creating opportunities for startups that didn't exist before. Like, could there be a new Shopify? Three years ago, I'm like, I don't know. Like, I'm not probably going to spend too much time thinking about that. But in a world where the fundamental consumer journey is changing, okay, that opens up a wedge. So I just think that there's more, as things change, there's more opportunities for early stage startups to navigate those wedges in a way that it's hard if the market is a little more stable like it was three or four years ago. Yeah, we have a lot of on Retail Gentic, we have a lot of folks on Wall Street on the buy and sell side that kind of follow our content. And the two questions I get are about Amazon. You know, will someone attack the fortress of Amazon and Shopify?

28:07 Well, let's spitball that. Let's say, what is Shopify's weakness? You know, I've pointed out several times that if you look at Shopify, it's not a software company. It's a fintech company. 80% of our revenue comes from ShopPay. So I was very cognizant when I was using Muse on a Shopify store that it offered me the Stripe link. Now, I think that Shopify has done some deal where that will now be ShopPay. But you know that that's pretty interesting because if it's stripe link my i'm pretty sure that shopify's cash register didn't ring yep and if i'm stripe at some point do i flip on them and i'm kind of like, i want that margin and then yeah so what is all that you know or is it some startup that just comes out of left field and solves things so different we can't even see it yet i think there's for me shopify there's two fundamental risks that they have one is does Does agentic, commerce just take off in a place so like.

29:01 Scale so high that the orders aren't happening on the e-commerce site? They're happening on ChatGPT or Claude. Yep. And yes, they could be a part of that transaction, but they don't have to be. So like that's risk number one. Does that scale? And are they not able to stay in front of it? Like they have some advantages in that like they can enable a large group of merchants for agent to commerce. So that's the value, but it's not clear that they'll be able to extract all of the payment processing dollars that they've had. So I think that's risk number one. And I think risk number two is that the e-commerce site of the future looks fundamentally different than what Shopify has. So right now, every e-commerce site is a catalog, right? And consumers kind of navigate their way through the catalog.

29:43 In a way, we're all now being trained to have a conversation and to have a journey through ChatGPT or wherever our AI of choice is. I think the e-commerce site of the future will, one, be more journey-like, right? Take some of those characteristics with the visual things that e-commerce sites do really well. So we'll be one more journey-like. And two, it's going to be more and more important to be able to start that journey based on wherever the consumer left off in their AI journey. So if I did all of my consideration on ChatGPT and then jumped over to the e-commerce site, I need a much different journey to get to conclusion than if I did just a little bit, found out about the brand, need to start from scratch. So I think the e-commerce site of the future is journey-based and needs to be.

30:27 Picked up from different places rather than everybody goes to the homepage or hits something. The little part of your, the part of your journey that's intent and, and context needs to kind of like be transportable for those pieces. So I think that's the risk that it like that e-commerce site of the future is just so fundamentally different. Someone has an angle to beat shopify which again like we're saying before three years ago i i didn't see an angle how you beat them like that's just that good i agree i think there's a we're in an inflection point which opens the door to more of these disruptions i also think shopify does a bunch of stuff great and i admire a bunch of things, they you mentioned one thing that's super important and durable in this which is aggregation and i actually think aggregation is shopify's achilles heel like they, they have all this aggregate volume but they don't have rights to aggregate it like and so, like they don't try to the shop app but it's never gotten any traction yeah yeah you can go you can go in the shop app and actually see the the total catalog and stuff but i don't think any consumers even aware that although the shop app is a but even that it's like it's not good it's dicey, they they sold web stores to all of these brands under the premise that they own their own brand like it's not clear that they have the rights to put all those brand opt those brands in to the shop app. They did it. They did it.

31:41 But there's so little volume on it that nobody with any that can afford a lawyer said anything, but it is possible a lot of Shopify's GMV is churn. It's small businesses that join, get some GMV for a year and then go away and then read some repeat. And so it's unclear what percentage of the important durable GMV in the market they really own. Yep. How about Amazon? So one theory I have is, you know, like you talked about the basketball example. So you start to get all this context outside of e-commerce that's interesting, like what you do for travel, what car do you drive, and these things. And then if you want something e-commerce-y, the competitors will be able to pick that up.

32:23 So the horizontal agents, like a muse, will have a little bit of an edge on Amazon because they'll know a little bit more about the consumer than Amazon ever will because they do more things for them. And then, you know, I don't know if you pick off that much because people just have such a habit with Amazon, but that's that is an entry point on that one. Yeah, I think that Amazon has is both the catalog, all like the breadth of products, and they've got this fabulous delivery infrastructure. And I think neither of them, regardless of how one consumes, finds those things, those still are lasting advantages. So I think they have a longer term, like they're gonna be less exposed here. And I think they're also so big that they are one of the few players that has a legitimate shot at keeping a large percentage of their consumers on their sites using AI.

33:11 I'm starting to use it already, but they're going to be good. And if I'm doing a shopping journey, Amazement does know enough about me, especially my shopping habits, that they're one of the few that has a shot of being the starting place for AI. Yeah, to that point, Jesse said 350 million active users on AFS, which is Alexa for shopping. I prefer Rufus, but here we are. And then the people that do rattle through there, their basket size is 40% higher. Jason will correctly point out, is that causation or correlation? Have they just picked off? The top spenders, they don't, in a very Amazon fashion, they don't tell you any of that. But the $350 million, they definitely, and the fact to me is the fact they're changing the search engine to very frequently call into AFS when, they know it can't really deal with whatever the consumer is putting in there, that means that they feel very comfortable that the conversion is good and that it's working.

34:06 I just think also you just can't, it's hard to bet against Amazon. Hasn't gone well so far. No, I, one of my best investments, I, when I was in 1998, I was a junior in college. I invested a thousand in Amazon and I sold it at the peak of 15,000. So I made a 15 X and I thought I was crazy smart. Yep. And I did hit, I hit a peak, clearly not the peaks. So they're just, they're great operators and they have a thing. Their physical infrastructure, I think is, is, is going to be more resistant to AI than most of the other folks out there. Awesome. I think we're going to wrap because I know you need to get to your next appointment. Matt, thanks for coming by our mobile recording studio and putting up with all the fun noise and activity here at the trade show.

34:52 It's been a fun experience. Thanks for coming by and we'll have you on in another six months. Sounds good. Thanks for having me. You've been listening to The Jason and scot Show. For all the latest news and trends on e-commerce and shopper marketing, subscribe to us on iTunes or visit jasonandscot.com.