There's also been a lot of free stuff on the internet, and one of the reasons it's been free is because of traffic and advertising. And so if you know, that's not going to get as much advertising or traffic anymore because everything is going through an LLM platform and people just aren't linking back directly. The temptation would be to try to make up that revenue by charging for the actual content.
Yeah, but that assumes that what I want to browse and what the humans want to browse will be the same thing. And I'm not sure that'll necessarily be the case where, you know, it'll still be the case that, say, for media properties, you know, that is aimed at human consumption. And so they'll do fine with the advertising revenue they have. Maybe they have some incremental revenue that they can get from the AI stuff. And then I mean, maybe you're right that like the really crummy, you know, aggregator pages for it is like top ten list. You know, you Google something and you get some SEO page and there's like a million Taboola ads on it that you're trying to click through, and the pop-over takes over your screen. Maybe those sites have a harder time, but I don't think anyone's going to really be too sad about that.
Hello and welcome to another episode of the Odd Lots Podcast. I'm Joe Weisenthal and I'm Tracy Alloway. So Tracy, another episode where I can mention that I've been vibe coding. But, you know, here's something that I may have mentioned before.
I'm sure you have, Joe, but I'm not 100% sure.
Okay. I think I've mentioned it. You don't like the horror. It's actually very easy. You just come up with an idea and you type it in English and then it's there. You know, the hardest part, I think, is registering a domain.
Yeah. You complain about this like everything else is so smooth, right? You just type it in English and then you like have your little app or whatever.
And then it's like, oh, I want to put this on, we're on the web. But then you go to like a domain name registrar. I had been using GoDaddy for a long time, and then it's like then set this field for this field, this field for this field and toggle this and this worked or this pages and the field in the app field of the B field. And I swear that's like it's so maddening. I just hated it so much. It's so annoying. It's the hardest part of the whole aspect.
I have terrible memories as well about dealing with Bluehost and things like that. Is this the first time you've ever set up a domain?
No. There's the thing. This is the other thing. Yeah, I've done it before, like I've done it before, and it's still incredibly annoying.
Progress comes slow to domain registration.
I mean, there's probably good reason for it to be complicated. You don't want it to switch in the wrong direction. Anyway, I was very excited. It's a couple of weeks ago. It was actually. Well, the two of us were in Madrid. The company Stripe, the payment company, announced that there was going to be some agreement with Cloudflare, which does a lot of web hosting stuff, where you could just have the agent. Right, just buy a domain name for you and show you think of one, and you just type in and buy this domain name and host it there. And I mean, it was like, oh my God, I wrote about it. The newsletter was like, this would be so this would have been so nice a month ago.
Yet another thing that you will never have to do again.
Exactly. Let another. So I mean, I saw the Stripe news and it generated a lot of buzz and it all comes under the umbrella, I guess, of agentic commerce. So this idea that you can have agents that are actually directly transacting on your behalf with other agents or other websites, and I have to say, as someone who shops online a lot, yeah, this is the first time I feel disintermediated by AI. And I was actually like, I'm actually particularly excited about this episode because I'm very curious. Like basically what you specifically will ask as someone as a professional shopper, I was I was going to say as a person of taste, I think you as a person of taste, I have seen you say scrolling a very nice paper, like websites, like wallpapers and things like that. And I'm curious, like, what does it mean for that if like, you know, the AI I bought two years from now knows you so well. There's like, Tracy, I thought you needed this new wallpaper because you're doing this. I just went ahead and ordered it, it will be delivered tomorrow.
This changes the world. It brings up so many interesting questions about discoverability and taste, as you mentioned, and also just the nature of the internet, because when I think about the internet, you know, like a lot of other things, it's basically built to sell you things, whether that's products or information content. And we know that it also runs on those sales in the sense that, you know, a lot of the internet is driven by advertising, which is related to traffic. Obviously, we actually did that episode way back about how the internet runs on millions.
Yeah, it's like, yeah, yeah. Ad auctions. That's right. And so this topic of agentic commerce and the payment system, obviously, it's of interest to anyone who shops and does stuff online, but it's also really relevant to the actual evolution of the internet.
Couldn't agree more. I think it's really and I would go further and say it's really important for like and the, you know, culture itself, advertising, creation of culture like ads are influential and do all this stuff. What if there's a changing advertising, all this kind of stuff. Anyway, let's stop hearing from ourselves. We really do have the perfect guest, someone that we've wanted to talk to for a long time. Someone who is right in the thick of it and all this stuff. With the agentic commerce and payments. We're going to be speaking with John Collison, the president and co-founder of Stripe. So, John, thank you so much for coming on Odd Lots.
I'm very excited. What how do you. Yeah, buzzwords etc. What is quote agentic commerce unquote.
Yeah. So it is a bit of a definitional question, but broadly is is AI buying something for you? Okay. And we think about it in two categories. One is on the consumer side of things, you know, this is where you were skeptical whether, you know, Tracy will ever be an adopter, where you, you know, research something in ChatGPT or Claude or what have you, and then go buy it there or have it buy it for you. But then the second is your Cloudflare example. And so it's B2B or a developer layer to commerce where you're just trying to have, you know, Claude Code do something for you and as part of it needs to buy resources. And we can talk about that because they're quite different.
Yeah Tracy. And we'll get to this. But the resources is another thing I've learned about, which is that when you do an API key, you're like, you have an idea and you're like, oh, you have to go to the web to buy the API to enter credit card number. That's really annoying, etc. But we'll get into that. But that is another big element of it that I feel like is getting closer to being solved.
Okay, well, on this credit card point, I mean, the history of online payments is basically like a continuous elimination of the friction, right? And going from I have to enter my credit card details to I have to log in to PayPal to now I can just basically click a button, whether it's Apple Pay or something else. Is agentic commerce just like another leg of the frictionless process, or is something fundamental in your view, changing here?
Both. So part of agentic commerce is just reducing the friction. And you're right that the general direction of travel over time, you're back to the, you know, introduction of credit cards when the earliest adopters have computer networks and then, you know, the original e-commerce. And now, like you're saying, wallets like Apple Pay and Google Pay, and we have a product called Link and things like that. And our view is that at a very minimum, we're going to see. And Tracy, I think this is where you can maybe get behind us is leave aside any agent decision making that's happening. Just the final step of buying something. You know, you might have researched a product in an AI app before where you're going back and forth. I was trying to buy a bike case recently, like, you know, to travel with my bike, and you're asking all these questions hard versus soft, and you need to take off the handlebars. And what's the right one? It's finding all these cool little niche brands. I think it's actually quite good for smaller merchants, by the way, because you have the ability for them to be discovered that they might not have had otherwise. But anyway, when you find the product at the very end, do you really then want to go and be filling out all these web form fields and things like that? Or do you want to just say, yeah, that sounds good, buy it for me in this size. And our bet is that people will want the lower friction option. I think in the history of technology, the lower friction option tends to win out. And so we think that clearly will happen and is happening to some degree already. It's just the agent filling out the forms for you. I think where you get this interesting question is and when people hear agentic commerce, I think their mind jumps to the agent actually having some autonomy in some decision making for you. I think the problem is people somehow jump to the examples, like anyone who talks about this problem, they pick terrible examples. They're like, oh, the agent will book a vacation for you. You know, all the activities on your vacation for you. It's like people think it, but they daydream about their vacation all the time. Researching and booking is the fun part. That's the fun part. Or similarly, you would like, you know, women's fashion. It's like the whole point is the scrolling, you know, not just the buying. And so people pick these, you know, super enjoy it like it's the robots taking the jobs. And, you know, people want to keep those scrolling jobs. And so when you think about robots having autonomy, I would instead proffer some examples where you really don't need to be, you know, you give the AI a recipe and say, buy the thing this size. You know, we need to make this tonight. And yeah, I can probably guess that you have salt and you have garlic at home, but you need the ground pork. Or again, you know, then migrating to the B2B example, no one needs to configure yet another domain and go through all those forms. If Claude Code can work with Cloudflare to buy the domain for you and set it up, I think Joe will be much happier not feel like he's missed out on a valuable life experience.
I want to talk more about the B2B stuff, but let's for a moment stick around on the B2C element because I think there are gradations of this. So one is just, you know, another elimination of the friction layer at the far end of the spectrum. Conceivably, there will be things where the AI truly does anticipatory buy you things that you didn't know you need, etc. And then there might be like some middle. I saw this screenshot recently and I couldn't find it before the episode. Someone tweeted it and I'm it was but someone had like some retailer, I think it was a shoe retailer, had put a bunch of shoes on their website, but instead of like putting any brand names on the shoe, like we would say like a Nike Pegasus or Nike Air Jordan, all of the shoes were like, hiking boot. That's good for semi-serious hiker. Do you see this currently where there's already a change in how retailers are presenting or displaying or promoting their wares such that they're more tuned for what the chatbot would identify or the AI would identify as cool versus what a human influenced to influenced by brands would pick up as cool.
Yes and no. I mean, there's definitely changes. I think, you know, some of the stuff that's most notable is often not the most relevant. It's a little bit it sounds like what you're describing is a little bit like the, you know, the Thai restaurant calls Thai food. That's right. To, you know, do best in the, in the SEO. But I think what we see, as like, kind of maybe a more interesting direction of travel is, again, what's happening already some amount of, you know, the form fields being filled out for you and kind of the friction reduction and a huge amount of, again, I presume you guys have both done this is, research in AI apps. And, you know, the way I would frame that is just keyword search is ridiculous. Like, it's ridiculous that we got to the year 2026 relying on keyword search where that makes sense for buying a book or a DVD, where you know the title of the book or DVD that you're trying to buy. But that's about the limit of keyword search. And so the fact that anyone was trying to buy, you know, furniture or clothes or anything like that, and there was this text box at the top of the website where you put in your keywords. It's like, that's just that's not how I know in shops in the real world, things like that. And to receive this AI-powered research where people can explore a product space and start to give constraints, and you have this textual search interface, but you end up with kind of a much higher quality search where it's like, okay, I have a spot in this room where I'm looking for a piece of furniture. That's this one. Why is Max and this high? Because I need to fit it in there. That's the kind of thing that's just much better suited to the AI models. So you see different kinds of product research possible. And like we said, it's very helpful to smaller brands because people aren't necessarily stuck behind the traditional aggregators anymore. They can break out. And again, I have found in my little research projects that I've done find of how do you often get these small brands surfaced that you might not have heard of before? And so we think it's actually probably quite good for the dynamism.
Well, let me ask Joe's question in a slightly different way, which is if we assume that in the future, if I'm a business selling something online, previously I was selling to humans more or less like maybe I'd optimize for Google search results or whatever. But like mostly I'm trying to appeal to humans, and now I'm trying to appeal more to agents. What does optimizing for agents actually look like?
Part of it is similar to traditional SEO, right? So you want to be legible to those agents. You want to be discovered by them. And but but part of what at least I've seen is that the AI have pretty good taste, like, Joe, I know if you've found this in your vibe coding, by the way, Joe, I'm reminded of. It's sort of like the old joke. How do you know if if someone vibe codes, they'll tell you about it, and. Exactly. But, Joe, I'm I'm guessing you found that these are very tasteful at picking libraries and picking, you know, specific subcomponents to use. And similarly, in our kind of early experience of the agentic commerce, they're quite tasteful in finding tools that suit the job pretty well. And so I think you may get this more efficient market phenomenon where at least in picking coding libraries or something like that, the AI have like I find, you know, when I do stuff, they're discovering libraries that are perfect for the job that I hadn't even heard of or something like that. I think you're starting to see the same phenomenon with products where again, I have had the experience of, you know, I was trying to find a travel adapter and I found it from this, like, tiny company that I had never heard of, but doing exactly what I wanted. It's like an integrated travel adapter and power brick. It's really good. And I would never have come across that that company otherwise.
This is incredibly important because, as I mentioned at the beginning, advertising creates culture. It's part of culture, especially iconic advertising. And maybe it'll be a little while before, you know, the Nikes of the world. There are the Nikes and Apples of the world. They're going to be advertising for a long time. But like, do you see this eventually as sort of like, you know, one of the things I think that the internet appreciates about Stripe as a company is you and your brothers' commitment to interesting discussions of economics and so forth. And you talk about econ history. There are a great debate about whether ad spending is a huge deadweight loss on the economy, or whether it's something positive, etc. And I'm curious, like, could you see a sort of post-advertising future? Yeah. You know, at some point on the horizon?
