AI Wants to Shop for You. What Could Go Wrong?

16 Sep 2026 · 41 min · 12 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

The episode debates whether “agentic shopping” (AI agents that can research and potentially buy items for you) will soon change retail, and what could go wrong—especially for Amazon’s marketplace role, pricing access, and advertising.

Guest backgrounds

Jason Del Rey, longtime tech/e-commerce reporter (previously at All Things D, Recode, Vox; later Fortune), now independent. He runs The Aisle newsletter focused on how AI is and isn’t changing retail.

Key claims

Early “agentic commerce” is mostly limited to chat-based product Q&A and comparisons, not reliable end-to-end purchasing. Progress is constrained by real-time pricing/data access, retailer bot defenses, and conversion issues (e.g., “Instant Checkout” performed poorly). Even if agents improve, Amazon may resist disintermediation by blocking crawling, withholding up-to-date data, and suing competitors (Perplexity). Retailers may lean in if Amazon won’t.

Notable examples

Instinct (text/voice agent; can attempt actions like reservations/refunds) and Meta’s Muse (travel shopping agent; users report errors like refund mistakes and account bans from Resi). American Express’s Resi and its “fast lane” agent approval concept. Walmart’s earlier ChatGPT “Instant Checkout” integration. Amazon’s “dumb pipe” risk and internal AI experiments (e.g., Alexa for shopping, “Buy For Me”).

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Understanding Agentic Shopping

0:04 to 0:39

Jason explains the concept of agentic shopping and its evolution.

“They bring payments, inventory, staff, scheduling, and online orders into one easy-to-use system.”

Understanding Agentic Shopping

2:56 to 5:08

Jason explains the concept of agentic shopping and its evolution.

“Last year, he went off on his own and launched the Aisle newsletter.”

Challenges in Agentic Commerce

5:09 to 7:22

Discussion on the technical and business barriers hindering agentic commerce.

“That's the thing some people think is very exciting.”

Exploring AI Shopping Assistants

7:23 to 11:47

Talk about the potential and limitations of AI tools like Muse and Instinct.

“Is it biz dev stuff where different companies have different ideas about how this should work, if at all?”

Practical Applications and Trust Issues

11:48 to 14:00

Examining the practical uses of AI in shopping and the trust concerns surrounding it.

“So you can, using Stripe authentication, make a purchase today with this Instinct app or Muse.”

Complex Travel Planning and AI

14:00 to 18:40

Exploring how AI can assist in travel planning and the challenges involved.

“it to come from me, the response, I want you to draft something and then give me a second look and then send it away.”

Sponsorship and Business Solutions

18:40 to 19:52

Discussion on Square's business solutions and how they address operational challenges.

“Running it is where things get complicated.”

Amazon’s Approach to AI Shopping

19:52 to 25:11

Analyzing Amazon's reluctance to embrace AI shopping bots and its implications.

“And Amazon has been saying, basically, no.”

The Future of Online Marketplaces

25:11 to 28:07

Discussing the potential impact of AI on traditional online marketplaces and advertising.

“And that narrative, I really think a lot of it changed once people saw how much of a hard time OpenAI and ChatGPT, which was kind of on the edge, was having with just getting people to complete a purchase on that site.”

The Future of Advertising in AI

28:07 to 31:20

Exploring how AI advancements may disrupt traditional advertising.

“What happens, this is a media podcast, or one of the things we cover here is media.”
Show all 12 chapters

Grocery Shopping and AI Integration

32:49 to 38:50

Discussing the potential impact of AI on online grocery shopping.

“So things are, seems like they're going very fast, at least technologically.”

The Evolution of Retail Partnerships

38:59 to 40:56

Examining how retailers are adapting to AI technologies and competition.

“We too can, you know, do the Internet thing.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:01Peter Kafka:Support for this show comes from Square. We all know business owners need to wear a ton of hats, but it's about time you got a hat rack so you can simplify your operations. Square's got you covered. They bring payments, inventory, staff, scheduling, and online orders into one easy-to-use system. If you're starting a business or running one that deserves better tools, Square helps you sell, manage, and grow without slowing down. Right now, you can get up to$200 off Square hardware at square.com slash go slash channels. That's square.com slash go slash channels. Run your business smarter with Square.

0:39Peter Kafka:Get started today. It seems super obvious that if any of what you're talking about comes to pass, it hurts Amazon. And the worst case scenario is it turns Amazon into like a dumb pipe, just logistics center. And as a consumer, maybe you're 100 % okay with that. But if you're Amazon, you're freaking out.

1:03Peter Kafka:From the Vox Media Podcast Network, this is Channels with Peter Kafka. That's me. I'm also chief correspondent at Business Insider. And today, of course, we're talking about AI. But not, is AI going to end the world? And do you think maybe we should do something about that before that happens? That's a pretty good conversation to have, but that's not what's on the agenda here. Instead, we're going to go with before AI maybe ends the world, is it going to change shopping? And actually, I think these two things are related. Okay, bear with me. AI shopping bots aren't going to end the world. But the idea of telling a bot to go buy you a plane ticket or groceries or land you a great restaurant reservation has gone from pipe dream AI guys talk about to actually maybe this is going to work quite quickly.

1:54Peter Kafka:Perhaps you heard of Instinct or Meta's Muse, which are very new AI agents that specifically promise they can do that kind of stuff. And if that really does work, it's going to have all kinds of knock-on effects. But also, AI in the real world isn't just about AI. It's not just about how powerful the LLMs are. It's about how people and companies and systems all fit together, how sometimes they are quite resistant to change, and how plopping new tech in there isn't necessarily enough to change things. All kinds of stuff has to get rethought. Shorter, AI is getting much better, much faster than some of us thought, but we still don't really know what kind of effect it's going to have.

2:37Okay, enough theorizing.

2:39Peter Kafka:Let's get to reality. Which is why I brought my pal Jason Del Rey on to tell us what's actually happening. Jason is an expert on what we used to call e-commerce, and now we just call shopping. I worked with Jason for years at what was called All Things D and Recode and Vox. Then he went to Fortune. Last year, he went off on his own and launched the Aisle newsletter. Get it? AI. Dedicated entirely to the way AI is and isn't changing retail for businesses and consumers. So listen to our chat and then go sign up. That's what I ask. Okay, here's me talking to Jason Del Rey.

3:25Peter Kafka:I'm with my old friend, Jason Del Rey. Jason, I do not understand how AI is going to affect shopping. Luckily, you are literally spending all of your time writing, reporting, talking about this over at The Aisle, your great newsletter. More on that later. For now, you're going to make me smarter about all of this for free. Thank you, Jason. Pleasure, as always, Peter. You say you're covering AI commerce? I call it the AI era of commerce. That's right. And the phrase that you hear all the time now is agentic shopping? That's the sort of equivalent Now, I mean, they're different things, but that's the thing that has everyone excited.

4:06Right. The phrase a lot of people in the industry use is agentic commerce. But yes.

4:12Peter Kafka:We'll call it agentic shopping. We can do that. What do they mean when they say agentic shopping? Well, that's funny you ask that, because I would say nine months ago, 10 months ago, when I started my newsletter, they meant, you're going to tell some AI agent, I want this thing below this price, and it's going to go out on the internet and buy it for you. Well, they realized that wasn't going to happen very easily for a variety of reasons we'll get into. And so now when people say agentic commerce, they typically mean at some point in the shopping journey, an online shopper went to an AI chat and mentioned something about a product or asked it a question about a product or to compare two products.

4:59Peter Kafka:So that exists today, right? And I understand why that might be interesting and important in your world, in the tech world, in normal person world, that first version. That's the thing some people think is very exciting. But we're nowhere near there yet. Well, it's a good time to have me on because just in the last couple of weeks, I think we're starting to see what at least people in the tech industry and in retail could consider the beginning of what could become like real agentic commerce. The bots go do the thing for me. Correct. Now, do people want the bots to do the thing is a much bigger question.

5:37Peter Kafka:If we get to that future where a bot goes and buys something for you instead of you going on to Amazon or wherever and doing it, other than the novelty, why is that a big deal? Is that better for us, the consumer? Is it because it generates new opportunity for new companies? Is it because it screws over Amazon? We'll break it all down, but what is supposed to be exciting about this? Well, I think you hear a lot of people who are building these things, both at big tech companies and startups, saying, we're going to take the mundane stuff. You hate to shop. You may not hate to shop, but you hate to do the boring shopping.

6:23You hate grocery shopping, even if it's online. You hate clicking 37 times to make your shopping cart on Instacart. Or you want to go really deep. You're a hobbyist in photography, and you want to go really deep and find the absolute best camera with these specs. And you don't want to browse 20 different sites or product reviews or read everything. You want to just ask a super intelligent being to do all that comparison shopping for you and lay out on a plate the best option.

7:01Peter Kafka:It's supposed to be a better experience for you and me, the average consumer. Yes. You're skeptical. All right, we'll talk about why that might not be a good thing. But what is holding it back today? You said it doesn't really exist today. There's some edge cases people are excited about. You said nine months ago they thought this was the future. Now I'm like, not so much. Is that technical on the part of AI? Is it technical on the part of everything outside of AI? Is it biz dev stuff where different companies have different ideas about how this should work, if at all? Yes. Okay. Let me try to unpack that a little.

7:36I referenced nine or 10 months ago when I started writing this newsletter. And at the time or shortly after, OpenAI and ChatGPT had launched something called Instant Checkout, which was basically the ability for you to buy something without leaving ChatGPT. So you're researching and it becomes a shopping site. essentially. Well, a couple of big companies integrated, Walmart being one of them. And then Walmart came out a few months later and was like, the conversion on this on ChatGPT was awful. People don't want to buy one item at a time. They want to build a shopping cart. So that part's technical.

8:11Another piece of this has been access to pricing and product data that's really detailed. And real time, right? In real time. Which is one of the things AI is particularly bad at at the moment. Correct. And in some cases, you have retailers who don't want to make it easy to get access to that data, Amazon. In other cases, they may check, you know, a bot may check every 15 minutes or a thing goes out of stock. So there are all these ways it may break. And I think the other piece of it is companies have been building stuff without really knowing what problem they're solving. Which we could go back, I mean...

8:53Seen this one before.

8:54Peter Kafka:Yeah. I mean, Metaverse is like one a couple years ago that comes up a lot in this space. I mean, they know that shopping and commerce is a big, big, big, right? Correct. Jeff Bezos can buy super yachts and rocket ships in the Washington Post because of that, right? So I understand why that would just be an attractive place if you're a builder. Like, oh, let's go attack that space. The problem has been technical on one hand, business deals. Amazon has not, you know, biggest online store on the planet, has not agreed to really do business with any of the LLMs. They want to build it all internally on their own site.

9:31And then the other piece is what are the use cases? I named a couple, but if it was really valuable to the masses, I should be able to rattle off.

9:40Peter Kafka:Well, every LLM that pitches ever says, and you are going to one-click and we're going to plan and buy your trip for you. That's the one they always go with. That one seems the farthest away just because I'm the least likely to entrust an AI to book me something that involves my money and my travel. I guess my question is because I've been testing out some of these messaging agents, and I'm trying to plan this complex trip with my wife's family where people are coming from different airports, different budgets, and we're trying to find something that works for everyone. That part's great, right?

10:15The planning, just any version of better Google, I'm all for.

10:20Peter Kafka:And you still got to fact check it. We're going to keep talking about these without talking about it. But I was playing with Muse. This is Meta's version of it. I had Muse make me a sample trip for something the other day, big expensive one, about someplace I don't know anything about. And I said, oh, I'm making a lot of money on this podcast. And then I just moved it over to ChatGPT and said, fact check Muse's work. And it found some problems. Which gets me to one core question I have. is separate from the limitations outside the AI. Do we think that the LLMs are just good enough at this, that the error rate is low enough, again, leaving aside trust, that they can be counted on to do this right most of the time?

11:00I mean, I think there is a small percentage of early adopters that will tell you, yes. And they like to be excited by living on the edge of whatever transformation they're foreseeing. I can name drop a couple of VCs, some well-known, some not, who are out there telling you everything that they're using.

11:22Peter Kafka:Yeah, and these guys were the first buyers of Tesla, too. So that doesn't mean they were wrong, just that they were buying stuff that was not ready for prime time. Right. I am not handing over my credit card or my email inbox, for that matter, to an AI agent right now. Now, at the same time, you have companies like Stripe, big payments company, a lot of trust there, at least on the merchant side, and some of the card networks that are out there signing up to help make these agents' purchases possible. So you can, using Stripe authentication, make a purchase today with this Instinct app or Muse.

12:00Peter Kafka:Of course you would if you're the payments company, right? All you want is to make the payment happen. I mean, if it becomes very problematic for the consumer side, then maybe you don't want to. But I think there are points on both sides. We're going this direction. Our job is to make payments happen. Of course, we're going to set this up. So we've been talking around this. Let's talk about a couple examples that seem like they're promising to do what you said doesn't really exist yet. One is called instinct. No normal people have heard of it. But if you live in San Francisco, if you live on Twitter and that kind of AI world, people are over the moon about it.

12:41Peter Kafka:And then Meta just released something called Muse a couple days ago, which basically is promising to do the same thing. I'm going to go book your travel. I'm going to go shop for you. I played a little bit with Muse, and I do not trust it to – I'm not giving Mark Zuckerberg any more information yet. So I'm not going to let it do that. Is there any left? You know, it's funny, because as soon as I started using Muse, and I'm just using it as a chat GPT competitor, all of a sudden, like, well, here I've got a difficult problem to solve with my finances or some paperwork. And it's not doing it for me, but I'm like, I'm giving it a lot of information.

13:13Peter Kafka:So maybe I'm just being... The difference is I'm okay sort of like showing it my ass a little bit, but handing it my credit card seems way scarier. So first of all, tell me about Instinct and Muse and what they can do and what they can't. Yeah, so Instinct, to me, maybe not to people really, really building the stuff in Silicon Valley, seemed to come out of nowhere a couple of weeks ago. And apparently, the parent company, it's a startup. It's raised hundreds of millions of dollars. And you can either WhatsApp it, you text or WhatsApp it or send it voice messages. And you basically say, give it access to your inbox and tell it, go through and highlight, you know, action items or answer, you know, every, unless I flag this email as I want it to come from me, the response, I want you to draft something and then give me a second look and then send it away.

14:13to I can't get into Chisiyama in Manhattan, and I want you to check for the first opening you can find at 6 p.m. Just hammer the reservation site over and over. I mean, we'll talk about that in a second. Yes. To what I was talking about earlier, which is I have this pretty complex travel planning I'm trying to do, and I'm going to give you all the details, and I want you during this 10-day period to return to me the cheapest options on each day. And I think I also maybe let it log into my United to show me miles too. I know, living on the edge.

14:55Peter Kafka:You're not having it buy a ticket though yet. But it says it could. And theoretically it can, right? Well, that is the big question. Theoretically it can, But a lot of sites have said a lot of retailers and other sites. I mean, they've long had sort of bot fighting technologies. Now, there are ways to get around it. We can get into Amazon, you know, has a lawsuit going on about this. But theoretically, if you give login information to a site that needs to be logged in instinct, a muse is saying we can go ahead and do this for you. Yes. and Muse, I don't know if there's a Muse horror story yet that's circulating on social media, but there were a bunch for instinct.

15:43Peter Kafka:And it's funny, depending on how you look at them, they're either a horror story, like look how exciting it is and it's the bot actually not doing what it's supposed to do. But imagine when this works the right way. And one was someone complaining that the bot had actually screwed up a refund and said they lost hundreds of dollars. Another was someone using it to get into a hard restaurant and they were using Resi, the reservation site. And Resi, after at some point, said, actually, we're cutting you off because you've been hammering our accounts over and over. And it turned out to be a temporary ban, but banned the account.

16:18Peter Kafka:And so when you see those, are you like, yeah, this is why this stuff isn't working, can't work, it's a long way from working, or are you in the boat of like, oh, this stuff's really close to being working for real people? You know, all these commerce companies, and that can mean a retailer or it can mean a resi, a lot of them now have teams. It may be two people. It may be 50 or 100 that are paying attention to everything we're talking about and no one wants to be left behind. And so trying to plan ahead for what this world looks like. So I'll give you an example. Resi is owned by American Express.

16:55I forget when that acquisition happened, but it happened at some point. And I did an interview with the chief innovation officer or something at American Express months ago. And they were planning for shopping agents at the time. And the announcement was, if you're making an AI agent and you register with us and we approve of you, we will back our cardholders if they use your agent.

17:21Peter Kafka:We're going to allow that internet fast lane. We're going to allow it to happen and we're going to actually cover the cardholder. So my only point here is whether it's happening or not right now, there are a lot of smart people at big companies planning for it. When you see these stories where it's not working today, do you go, yeah, it's not going to work, or it's actually one problem to solve, and then this does work? So I have two thoughts the last two weeks when I've seen Instinct and Muse. One is, wow, they're getting the attention of all these people at all these companies. Do those companies, when and if do Muse and Instinct have enough leverage to get those people to the table to actually kind of make this easy?

18:06But the part I'm leaning toward is actually a lot of these companies are just going to try building this stuff essentially internally and making their sites smarter and be able to do complex things more easily. Right. And because there is the disintermediation question. Yeah, so that leads us to Amazon.

18:27Peter Kafka:We'll be right back with Jason Del Rey, but first this.

18:39Peter Kafka:Support for the show comes from Square, the system that fixes one of the biggest headaches in running a business. Starting a business is exciting. Running it is where things get complicated. You end up juggling payments, inventory, staff, scheduling online orders, reports, and somehow you're still supposed to serve customers. Square makes it easier. Whether you're selling coffee, tech accessories, or maybe a podcast, that's right, this podcast is a small business, Square helps you run your business without running yourself into the ground. It does that by bringing payments, point of sales, inventory, staffing, and online sales together in one easy to use system.

19:15Peter Kafka:You get sleek hardware, intuitive software, a host of AI powered tools that help automate routine tasks, giving you more time back as a seller. With Square, you could sell online, in-store, or both and customize your setup as your business evolves. If you're starting a business or running one that deserves better tools, Square helps you sell, manage, and grow without slowing down. Right now, you can get up to$200 off Square hardware at square.com slash go slash channels. That's square.com slash go slash channels. Run your business smarter with Square. Get started today.

19:59Peter Kafka:and we're back you wrote a whole book about amazon and walmart you tracking all this stuff it seems like most of the big companies target walmart lots of big retailers have at least said we are open to this idea of the future where people come from somewhere else using a bot to shop us. And Amazon has been saying, basically, no. We want you or your bot, whatever, to come to Amazon and interact with us in the way we want to interact, which makes plenty of sense, right? Amazon owns this space. It seems super obvious that if any of what you're talking about comes to pass, it hurts Amazon. And the worst case scenario is it turns Amazon into a dumb pipe, just logistics center.

20:41Peter Kafka:And they can't do any of the things Amazon is good at, which is showing you other stuff that maybe you want to buy, any of the ads. In advertising or whatever that revenue number is. And all that goes away, and they're just a warehouse, and they're the same as somebody else's warehouse. And even worse for them, maybe you have no idea you bought through Amazon, you bought through ChatGPT or Instinct or Muse. And as a consumer, maybe you're 100 % okay with that. But if you're Amazon, you're freaking out. Yeah, it's becoming harder and harder to envision them completely getting disintermediated. And one reason is, as you said, they've basically said, so Andy Jassy repeatedly has asked about this.

21:23I think every earnings call the last six quarters. What's going to happen when the bots come for your business? And he actually has been saying, we will eventually do business with some of the third-party AI companies. But right now, the experience they give shoppers for shopping is shit. And he may use a different term, but that's what he's trying to get across. And I don't think he's been wrong.

21:50Peter Kafka:But let's say it's really good. Let's say they get really good. Isn't the core problem that if it works, it's bad for Amazon? If I say, go buy me a dishwasher, and it may end up buying through Amazon, but it may go through Target or may go through Walmart, and Amazon doesn't own that interaction with me, the consumer, that's just undeniably bad for them, right? That would be bad, and that's why I think they're doing a couple of things. They're not cooperating, really, in terms of letting a bot or these AI, the apps, easily crawl their sites. They're not giving up up-to-date pricing info. They're suing a company like Perplexity.

Read the full transcript

22:31Which is trying to do that. Which actually before Muse and Instinct had come out with their own AI agent that uses a browser called Comet. And basically, humans were using the Comet browser to shop on Amazon through their own account. And Amazon told them, Perplexity, please stop. and then told them, you have to stop, and perplexity was not. And so they went to court, and that's ongoing. So, yes, this would be bad. And that's why you see Amazon, I think, doubling down internally on their own AI use cases within the Amazon ecosystem so that plus experimenting with their own shopping agent that would buy stuff off of Amazon as well.

23:19Sure. Rufus, is that the one? Or am I confusing it? Walmarts. No. The names leave a lot to be described.

23:26Peter Kafka:They have their own AI that will go - They rebranded it to Alexa for shopping, which - Ooh. Yes. Maybe a third rebrand is coming. That one lives on Amazon. Right. That makes sense. I'm on Amazon. I want to shop here, or Amazon wants me to shop here. I'm going to use a better version of search. Yes. Fine. They have some live features. I don't know how many people are using it where you tell Amazon or Alexa for shopping. That, I can't say it with a straight face. You tell it, buy this product when it drops below$20. And it will do that. And like, maybe that's useful to some people. But they are experimenting with something else, which is called Buy For Me, which they're surfacing products on Amazon that Amazon actually doesn't sell.

24:13They're not actually asking for permission from these brands. And then they're saying - Amazon is helping you buy something that Amazon doesn't sell, even through their third parties. There's other website over here that we have no connection to. We can buy it for you from there. Correct. And again, I think there are some people doing it. I think it's a fringe use case. But the world you're talking about, it would be very bad for Amazon if they just become the dumb pipes.

24:39Peter Kafka:And it's bad for other companies, too, right? Anyone who's sort of in that middle clearinghouse, all the travel companies, right? The Expedia's of the world, in theory, the DoorDash's and Uber Eats's of the world, right? Where if all they were doing was connecting you and the thing that you were buying, someone else can do that connection now. Yes. The narrative about a year ago was, or the question was, are marketplaces dead? These online marketplaces, which all of them are that you just mentioned to a certain extent. And that narrative, I really think a lot of it changed once people saw how much of a hard time OpenAI and ChatGPT, which was kind of on the edge, was having with just getting people to complete a purchase on that site.

25:26And so that narrative has kind of gone away.

25:30Peter Kafka:Because it's not ready yet. Correct. But now? Now, for me, this is an exciting time. And whatever happens with Muse and Instinct, I mean, Instinct has hundreds of millions of dollars. So how many tokens can that buy? I don't know. They have some runway to give this a go. And Meta being out there, not that, I mean, Meta's track record of building its own, you know, sort of cloning what's out there. Sometimes it works, sometimes it doesn't. Correct. But it is a, I don't know, you've played around with the app. I mean, it's a decent looking app. I think the biggest obstacle is quickly getting someone from what do I do with this thing to sort of a, I'm going to use, here's a magical experience.

26:22And I haven't quite gotten there yet, and I've been trying for about a week or two. People on X tell me I'm wrong, so I don't know.

26:31Peter Kafka:Who does the agent work for? In theory, it should work for you, the consumer, right? But I'm not sure that's the case. That's one of the big questions. I mean, the first question is like, do people want to shop this way or buy this way? The next question is, I think, yeah, who does it work for? And so I haven't seen the language yet from Instinct, but I would be shocked if part of their messaging, once they start actually talking a lot. Our job is to get you the best price regardless of where it is. We don't owe anyone anything, blah, blah, blah. Right. It is unclear how they will monetize their service.

27:11So if it's free for the user, the user's product typically is what we've seen in the consumer internet. And so once you involve advertisers, marketing, then I think that question is even a tougher one to answer with. If you're going to answer with a straight face, it's working on behalf of the human. I just don't know how you get there. Unless you're paying for it. Unless you're paying for it. Yeah, like a broker, basically. And I listened to parts of Mark Zuckerberg's interview with independent journalist turned VC, Alex Heath. And that question came up, essentially, who's paying for this? And he said in the future, he can imagine taking a small cut from whoever's getting transactions funneled to them via Muse.

28:07Peter Kafka:Interesting. What happens, this is a media podcast, or one of the things we cover here is media. What happens to advertising in this world if Muse comes through, if Instinct comes through, if this world comes through, right? An obvious question is, well, these companies spend a ton of time putting stuff on billboards, so I'll pay attention to them. And they spend a ton of money trying to get me to transact very specifically at different points on the Internet. And you have stuff like retail media, which is essentially people paying Amazon and Walmart for primary placement. So instead of walking down the aisle, you see them when you're scrolling.

28:46Peter Kafka:Correct. Possibility that a lot of this goes away. There is that possibility, and it would be an incredible story to tell. I am doubtful how much disruption we're going to see in the next call rate 12 to 24 months in that regard. Advertising does not disappear in a year, but? There's been some people experimenting with sponsored prompts. And you have bots advertising the bots, right? Like, how can we get this bot to pay attention to us? But it seems like inevitably that's a race down to zero. Yeah, then you get into the whole discussion, people talking about auctions between a human's agent and a company's agent.

29:33When the Instinct user got banned by Rezzy, I read all these very sure people on Twitter talking about how, well, obviously, all the inventory is just going to have to be auctioned off because the agents are taking over on one side and the merchants are going to have to build agent on their side. And, yeah, where do you influence? I covered advertising back in the day when I competed with you and media. I just trust there will be ways for companies to insert themselves into these conversations, but at what cost to their existing business. And they are absolutely worrying about it.

30:20Peter Kafka:I mean, I imagine there's a version where they go, no, no, no, the brand becomes even more important because it's the thing that distinguishes. And so we are going to convince you that a J.Crew polo shirt is actually worth paying for X percent more than a quince polo shirt. Maybe they're right, by the way. But and so that actually that brand is in a world where like you can't see any of the mechanisms. You're going to rely on the brand even more. I've heard that argument. I think it's probably true for one percent. This is the same argument that all the publishers make about the AI world. It's going to disrupt everyone but them because their brand is so valuable, a customer will come to them.

30:56Right. And I don't think many of those are looking at themselves with a straight face when they say that, but they have no other choice. I think there are brands where, yeah, the brand matters. People care about it. I think everyone in the middle could be fucked. but I think we're still some ways off from that. We'll be right back with Jason Del Rey,

31:21Peter Kafka:but first a word from a sponsor.

31:27Peter Kafka:Support for the show comes from Square, the system that fixes one of the biggest headaches in running a business. Starting a business is exciting. Running it is where things get complicated. You end up juggling payments, inventory, staff, scheduling online orders, reports, and somehow you're still supposed to serve customers. Square makes it easier. Whether you're selling coffee, tech accessories, or maybe a podcast, that's right, this podcast is a small business. Square helps you run your business without running yourself into the ground. It does that by bringing payments, point of sales, inventory, staffing, and online sales together in one easy to use system.

32:03Peter Kafka:You get sleek hardware, intuitive software, a host of AI powered tools that help automate routine tasks, giving you more time back as a seller. With Square, you could sell online, in-store, or both and customize your setup as your business evolves. If you're starting a business or running one that deserves better tools, Square helps you sell, manage, and grow without slowing down. Right now, you can get up to$200 off Square hardware at square.com slash go slash channels. That's square.com slash go slash channels. Run your business smarter with Square. Get started today.

32:48Peter Kafka:And we're back. Where are we in a couple years? So things are, seems like they're going very fast, at least technologically. We don't know about adoption. prognosticate for me. Oof. I'm not going to name some bad prognosticators that I don't want to be like. You're the only one I care about, Jason. We've kind of glossed over grocery. I do think there is a world where online grocery shopping, you know, there are people who say, I just want these 12 things. You know, I don't really care that much about brand. I want the best price on them. And I've written about a couple of startups that are basically trying to be this, which is you tell us what you want, and we're going to find you the best prices.

33:38And then we're going to just get it shipped to your door from an Instacart or whoever. I think grocery is ripe for that. Because it's commodity-ish? It's repeated. It's commodity. And I think there's some people that care a lot about my wife. We don't order online groceries.

33:57Peter Kafka:Who knows what the tomatoes are going to look like. Right. But I think for certain sort of demographics and maybe it's the, I don't know. Could be all kinds of people. Could be all kinds of people. Or they just know that a box of Cheerios is a box of Cheerios. So who cares how that gets there? I do think in the next two years, there's going to be some type of personal concierge that is a business, I'll say. And whether text messaging it is really the best way. I mean, I think we started this sort of hype cycle the last two years at Agent of Commerce, thinking that, calling this kind of conversational, where you're going to type back and forth.

34:36No one wants to do that, right? So I think the interface is what needs to change. You need to be able to ask a question and then have it be smart enough to give you multiple choice, and you're tapping through.

34:47Peter Kafka:Well, and OpenAI, when they're not saying we should slow this down because maybe the world will end, they're saying voice, voice, voice, voice, voice, where the future is voice. Right, which is really funny because we're talking about AI agents and the word Alexa, do we have any? I'm not setting up any Alexas in here. Sorry for everyone who's got one in their home. Sorry about that. I think it's supposed to be smart enough not to happen anymore. But Alexa, you know, I've covered Amazon very closely a long time. A lot of people thought they should have been first to sort of the AI boom of the last three, four years because of Alexa.

35:23They were not. They were caught off guard. It was really embarrassing. And so they spent a couple years rebuilding Alexa. And when Andy Jassy got on stage in New York City to announce it, I think that's where it was, he said, the difference with this Alexa, it's going to take actions for you. It's going to be an agent. And we're having all these conversations or the tech universe is the last few weeks. And Alexa really hasn't come up. And my only point there is, A, I guess I wouldn't count them out, but B, I think it's a little bit embarrassing again for them.

35:58Peter Kafka:There is, I don't know if it's a nice symmetry or romantic or whatever, it's interesting symmetry, right, that you have Amazon going after the big booksellers who disrupted the little booksellers, Amazon. That's why there's, you know, does Barnes & Noble still exist? I guess it still exists, yeah. You haven't read about their resurgence? I know you can buy Legos there. Okay. They still exist. Anyway, Amazon took over that business. Now there's this slew of energy and money and technology basically aimed at Amazon. And we're sort of, again, a certain kind of person is used to the narrative like, no, no, no, the tech, tech plus innovation plus what consumers want always wins.

36:42Peter Kafka:and whoever is the institution eventually gets eroded, disrupted. Amazon's thinking about this a lot. Do you think they have the capacity to hold this back? I do, and I think a couple of reasons. Whether or not they have the leading sort of AI researchers and LLM developers and fine-tuners, you can argue they likely don't have the leading edge there. What they do have is I don't know how many warehouses within one mile from your home that can get you something in an hour. They have fleets of drivers, fleets of planes. We just saw what happened in one of them. They have Prime, which is super, super sticky to use sort of an Amazon word.

37:30And the thing we didn't talk about, when an agent is acting on your behalf and purchasing something that is a physical product, a lot of people do care how quickly they can get that. And are they sure it's going to show up? Consumers are used to so much convenience, and it exists all the way to your front door and not just on a computer screen.

37:54Peter Kafka:Why do you think the Walmarts and Targets and big retailers of the world who are saying, at least, no, no, we're up for this, do you think that they don't actually mean it? Because it seems like they have all that same risk, right? They have all that same disintermediation risk. Or is it just that they feel like, oh, this is one advantage they could have over it? As long as Amazon is going to push this away, if they're the place you can go, that's their benefit. I think Walmart, for as well as they've done the last decade, they are still, I don't know, Amazon's six or seven times the size in online commerce in the U.S.

38:29Them and Target, I think they're looking at, A, we sort of need to lean in if Amazon isn't. B, they've learned very quickly. The AI labs know very little about the retail industry. And so if we can be the one kind of partner with them, maybe we'll be a beneficiary if and when these platforms take off as real sort of shopping adjacent platforms. And, yeah, I think if you look back in different cycles, you had Walmart and Target at different times saying, you know, trying to be as loud as they could be that, hey, we too are, you know, tech enabled. We too can, you know, do the Internet thing. We have a website.

39:16And so I think that's part of that. I'm actually going to be on stage next week with Target CIO. And, you know, they were very early in terms of partnering. And I think a lot of it is if Amazon's not there, we absolutely should be. You have this newsletter. It's called The Aisle. It's 10 months, you said, old? Nine months old? Just about 10 months, yeah. I trust having covered e-commerce for 13 years now that what's going on now in the next couple years will lead to some sort of massive disruption. and so to try to be out in front chronically. You kind of have it to yourself, right?

39:55Peter Kafka:I'm sure there's other people doing good work, but I don't know who they are. I just saw a post today. There's a retail industry trade that is, you know, someone changed jobs and they said their new role is covering social platforms and agentic commerce. So there are smart people paying attention, but it's going to turn out to either be the best decision of my career or the worst decision to focus entirely on it. Okay, the value exchanges, Jason took his time to come on this podcast for free and let you guys get smarter for free. And what I'm going to ask some of you guys to do, because some of you need to pay for this, is go to Jason's The Aisle.

40:32Readtheaisle.com.

40:33Peter Kafka:Readtheaisle.com. Maybe we're going to have a discount in the show notes. As a bonus, I'm going to give an extended free trial. There is no free trial right now on the site to just the ordinary people. But for listeners of channels, there will be one, I think, in the show notes. I'll wait to get that. Okay. Go get them, channels people. Thank you, Jason. Thanks so much, Peter. Thanks again to Jason. It's a bummer when you see your friend more in a podcast studio than you do in real life. We've got to fix that. Thanks to Rishi, who is editing and producing this week's show. So is Charlotte Silver.

41:10Peter Kafka:Thanks to our sponsors who bring this show to you for free, even if you don't subscribe to Jason's newsletter. but you should. Thanks to you guys for listening. See you soon.

41:46Peter Kafka:software. If you're starting a business or running one that deserves better tools, Square helps you sell, manage, and grow without slowing down. Right now, you can get up to$200 off Square hardware at square.com slash go slash channels. That's square.com slash go slash channels. Run your business smarter with Square. You get started today.

From the publisher

AI companies have spent years telling us that we’ll be able to use them to go shopping for us, or book us airplane tickets, or get us awesome restaurant reservations. None of that ever materialized.But in the last few weeks, services like Instinct and Meta’s Muse have made that promise feel much less theoretical.  Are they actually ready? What’s holding them back? And, crucially: Do we even want them doing this stuff for us?

My pal Jason Del Rey, a longtime journalist who now writes The Aisle newsletter, spends all his time learning and thinking about this stuff. We talk about what “agentic shopping” really means, why it’s been harder to build than Silicon Valley expected, and what happens if it finally works: Who wins, who gets cut out, what happens to advertising — and why Amazon may have more to lose than almost anyone.

And since you’re here: Channels listeners can start a 30-day free trial of The Aisle here. Or if Jason already sold you, subscribe today.
Learn more about your ad choices. Visit podcastchoices.com/adchoices

More from Channels with Peter Kafka

All 70 episodes
AI Wants to Shop for You. What Could Go Wrong?Channels with Peter Kafka · 41 min
Listen in VO