2026.10.01 | Substack | Internet | Stratechery Interview: Del Rey on Muse
原标题：An Interview with Jason Del Rey About Muse, Amazon, and Walmart (Stratechery Interview 10-1-2026)
An Interview with Jason Del Rey About Muse, Amazon, and Walmart
Thursday, October 1, 2026
Good morning,
This week’s Stratechery Interview is with The Aisle author Jason Del Rey. Del Rey has been covering the e-commerce space for years, including a decade at AllThingsD and Recode. In 2023 Del Rey published Winner Sells All: Amazon, Walmart, and the Battle for Our Wallets, tracing the multi-decade battle between Walmart and Amazon. Then, late last year, Del Rey went independent, starting his own e-commerce focused publication called The Aisle.
In this interview we cover Del Rey’s background, including a brief history of Recode, before diving into his area of expertise and one of my areas of fascination: Walmart and Amazon. I loved exploring the history of these companies, and you can make the case that the news of the last week — Amazon blocking Muse, while Walmart signed up — is actually just a continuation of a battle Del Rey has been on the front lines for. We also discuss agents generally, groceries, Google, and Shopify, and why Del Rey is optimistic The Aisle will be his last job.
As a reminder, all Stratechery content, including interviews, is available as a podcast; click the link at the top of this email to add Stratechery to your podcast player.
On to the Interview:
An Interview with Jason Del Rey About Muse, Amazon, and Walmart
This interview is lightly edited for clarity.
Topics:
Background | Recode | Walmart | Groceries | Agents | Meta vs. Amazon | Shopify and The Aisle
Background
Jason Del Rey, welcome to Stratechery.
JDR: Great to be here, Ben.
It’s honestly surprising I haven’t had you on before, I think our careers writing about tech have largely run in parallel. What I love about you is that I immediately think about you when any e-commerce story comes up — I go straight to your X account to see what you have to say. You have a good brand, good job by you!
JDR: Well, I sort of landed here by accident, but as I got deeper and deeper into e-commerce, I fully became an e-commerce nerd. I’ve had smarter journalists than me over the years tell me I should branch out and cover different things, and I kind of found that with the intersection of AI and e-commerce. But not really — I’m still here.
One of my longstanding takes about — an issue I have with journalism is this idea that you could just be a generalist and parachute in anywhere. I think particularly on the Internet, there’s massive returns to specializing and just going deeper and deeper. When the market gets very, very large, you can sort of make your own pond, and that’s a key to success.
I want to get to what you’re doing now at The Aisle, but I’m interested in your background — before you ended up doing what you’re doing. Where’d you grow up? Did you just want to be a journalist? What’s the path to 2026 and sitting here on this podcast, getting ready to talk about Amazon versus Meta?
JDR: So I grew up in the forgotten borough of New York City called Staten Island. I grew up wanting to be — I think my earliest memory is a SportsCenter anchor on ESPN.
Oh, nice.
JDR: No one told me at the time that my accent probably would be the first thing that would kibosh that dream. But I sort of knew journalism, and probably sports journalism, was where I wanted to end up and honestly, for about a year-and-a-half I was a sports journalist covering high school and college sports in New Jersey, which is where I live now.
Yep, that’s how everyone gets started.
JDR: I think the scariest person I’ve ever interviewed — it’s not Jeff Bezos, it’s Bob Hurley, the legendary high school basketball coach and father of Dan and Bobby Hurley, the college coaches.
So I did that a few years and I applied to journalism graduate school, which a lot of people told me was a mistake. But I applied to one school and said, “If I get in, I’ll go, and if I don’t, I’ll work up my career as a sports journalist”. I got in, I came out, and there were no sports journalism jobs, there were only business journalism jobs. I grew up with a CPA as a father, a brother in business, and a sister as a lawyer, so business was not a stranger to me, but it was kind of an accident that I began covering business journalism initially at Inc. Magazine. And then there’s a long road in between leading to today.
Recode
Well, you spent many, many years at Recode. Recode started about the same time as Stratechery did — I think maybe six months before or something along those lines. Recode had an amazing team, there’s a lot of folks that are still around, but it had an amazing team that was there for a long time. Why did you have such an amazing team? Why did it succeed when it did? And then why did it fail ultimately? What was your takeaway from there?
JDR: Wow, it’s too early for a drink, but I kind of feel like I need one going down this lane.
I do think about it quite a bit, especially as I got started with my own publication last year. I landed there a few months before it became Recode, when it was still called AllThingsD and Rupert Murdoch had his hands on it at Dow Jones. Peter Kafka, a great media journalist, got me there, I was basically competing with him when I was at Advertising Age, I started going to his events and I wanted to be noticed, honestly. Every speaker that was up there, I would raise my hand in the audience and try to ask a question until Peter noticed me.
Wow, so that strategy actually works!
JDR: I tell myself it did. Eventually we became friendly and he introduced me to Kara Swisher, one of the founders. And there was one opening at the time, it was covering e-commerce there. So I jumped at it, frankly, to work with Kara and Peter and Walt Mossberg and the rest of the great journalists — Mike Isaac and others.
I think it worked for a couple of reasons, we were a trade publication covering the technology industry, but with a voice and with some reported analysis and I think over time, the mistake we made was, frankly, never monetizing that via subscription. We got bought by Vox Media — great CEO there, Jim Bankoff — and there were attempts to maybe launch subscriptions, it just never made sense for Vox back then as a consumer-facing news publication, or a collection of them.
I think you put your finger on an important thing. The consumer market is different than the trade market. There actually is very much an opportunity for trade publications, but the business model isn’t always an advertising business model — unless you’re Politico and you can get every lobbyist putting ads in your publications.
JDR: Right. And I think the events business was very lucrative for a while, and then it became crowded, and we didn’t have the subscription business to fall back on. To be clear, I wasn’t super privy at the time to the business-side operations, but I think it could have been different with an early subscription launch, kind of like what [Jessica [Lessin] did at The Information.
Yep.
JDR: But what Recode did for me — Kara specifically really pushed her journalists to, I hate to say it this way, but build a brand. She encouraged me to launch my own conference series, which I ran for four years, called Code Commerce, focused on e-commerce. She encouraged me, and Peter too, to help launch this franchise of podcasts called Land of the Giants. It was a narrative podcast — I did the first season about Amazon. So really my entrepreneurial bent, other than my family, was inspired by my time there.
Recode had a great run. The consistent quality filled a real niche in the space, and people like you and Peter and other folks that were there remain, I think, the best in your space. And events — that was the one company I regularly did events for, and they loomed large in that space. I think you don’t need a drink, I think maybe a celebratory one, all good things come to an end, as it were.
JDR: Appreciate that.
Walmart
Well, you wrote a book in 2023 entitled Winner Sells All: Amazon, Walmart, and the Battle for Our Wallets. What was the overall story you were trying to tell, and why did you feel that was the right time to tell it?
JDR: I told myself I would only ever write a book if I felt like I was the best, or one of the best, people to tell it. Which seems obvious, but I don’t know that it always is to authors.
I really was one of the few — maybe the only — at the time diving deep into both companies. I started diving deep into Walmart once they acquired Jet.com and really did a big acqui-hire of the entrepreneur Marc Lore. I had followed him since I started covering e-commerce, and just thought it was fascinating to see them try to reinvent themselves.
But what I was trying to get at in the book was a real explanation of why they missed the opportunity for so many years and, “allowed Amazon to become Amazon”. It’s obviously more nuanced than that, they really just sort of passed on it.
What is this story? Actually, I think Walmart is super interesting. I’ve covered them, not a massive amount, but more than you would expect for a technology site, because I do think it’s super interesting. Both their many years of failures, and then also the way I think they’ve changed the business. But you’re the expert, walk me through the whole thing, I think Walmart looms very large going forward, so it’s very pertinent.
JDR: They do indeed. I found one of the earliest Walmart software engineers who was basically the leader of the skunkworks team back in the mid-1990s that was investing in e-commerce. He hadn’t talked to the press, I think, ever, so I was very excited to talk to him.
He basically told me they saw the future, they were experimenting — they were selling gift baskets online for Sam’s Club in the 1990s and other experiments. He went into the office of the then-CEO, David Glass), and basically said, “Listen, I don’t need a huge investment, I just need the other divisions, when I ask them for help, to not laugh in my face about what I’m doing”. And the story he told — David Glass has long since passed — was that Glass looked him in the eye and basically said, “This online thing is cool, but it’ll never be bigger than one year of sales at a Sam’s Club store”. This guy, Robert Davis, ended up leaving and going to work for Amazon for about a dozen years — not in super high-profile roles, but in critical internal roles.
So part of it was just not taking the opportunity seriously enough, but once they did, there were a lot of internal turf wars over who would get credit for sales. I think the internal incentives, until really the pandemic, for executives leading e-commerce versus physical stores were not aligned in a way that made them want to work together. Those two things for sure. And did they have the technical talent? Maybe not at certain parts of the 2000s, but in the early and mid-90s they really did.
How much of it was the assumption that, “We’re already the best in retail logistics in the world, this will be easy” — and it turned out the logistics for stores was way different than the logistics for e-commerce?
JDR: That’s absolutely right. I talked to a bunch of old-time Amazon folks and they kept saying, “We were waiting and waiting for Walmart to turn their stores into an e-commerce advantage, and they just never did”. It is difficult — even now today, real-time store inventory, someone ordering online for pickup and showing up when someone just walked out the door with the last one, there were all these complications. But Amazon was truthfully amazed for the better part of two decades that Walmart didn’t — until the pandemic forced them — turn the stores into the real advantage they could be for online shopping.
It’s interesting because — correct me if I’m wrong, you’re much deeper in the space than I am — my view is all the things you listed make total sense, they didn’t take it seriously, once they did, you had all the internal turf wars and those issues, and I think part of the Jet.com acquisition was “Let’s build an e-commerce entity completely independent from Walmart because it won’t work internally”, but it also feels to me — and I don’t know if this is right or wrong — that stores didn’t work for e-commerce because delivering and distributing goods to consumers individually versus to stores are similar but very different problems, that was my take for a long time. But to your point, today part of why Walmart is a great e-commerce experience is they are leveraging the stores. Did that only work because they built e-commerce on the side under Marc Lore and then glued that on? Or did COVID come along and save them from themselves, and they could have gotten there much earlier?
JDR: I mean, the answer is yes — there’s a little of all of that. Mark coming in — a lot of people criticized, and still criticize, that acquisition, $3.3 billion at the time, mainly to get Marc Lore and a few of his executives in. But what he did was change the narrative. He really did a good job changing the narrative, both on Wall Street and in hiring circles.
But he also brought the metabolism and the launch speed of new things — really, expanding the marketplace. Yes, following in the footsteps of Amazon, but expanding the marketplace that they would ship from their warehouses or their partner warehouses. I think that in combination with the pandemic forcing them to launch curbside and just focus on it and invest in it in a real way — I think there’s part of the answer in those different areas.
I think people don’t realize that Walmart has a marketplace like Amazon does, almost every item — well, not almost, but a lot of the items you find on Amazon that you feel you can only get on Amazon are actually also on Walmart, usually sold by the exact same sellers. It’s almost like that gave the foundation, and now you can layer on the stores, which are fundamentally limited. Walmart stores have an unbelievable number of SKUs, but it’s always going to be a limited number relative to what Amazon has, or what an entire marketplace is going to have, and then you get the experience of, “Walmart has everything, and some of the stuff you can get in an hour”, it’s a surprisingly large amount of stuff you can get in an hour. I tell people this, and I think the current advertising campaign for Walmart — “Who knew?” — is perfect.
JDR: Yeah.
People don’t know — this is actually an incredible e-commerce experience for a whole bunch of products.
JDR: Yeah. We in the Del Rey household we mainly use the curbside product, and it’s just very well done. Part of the problem I have with our local one — which is actually next to Teterboro Airport in New Jersey, which is funny enough where Walmart executives fly in, and it’s usually the first store they visit — is there’s honestly not enough parking spaces for all the pickup orders, because you also have their delivery network picking up in the same space.
Amazing.
JDR: I haven’t been there in a few months, so maybe they’ve fixed that. But no, it shocks a lot of people who live on the coasts. There’s a reason they’ve done very well on the market over the last five, six years, and part of it is that they’ve figured out the pickup business.
Groceries
You mentioned the Walmart angle in the book. But what did you learn while writing the book about Amazon, despite the fact you’ve been covering Amazon for years and years? Was there a new revelation that came along as you were putting together this narrative, comparing and contrasting the two companies?
JDR: Amazon folks like to talk, of course, famously about customer obsession and not competitor obsession.
(laughing) Have you used Amazon.com lately? But yes, anyhow, continue.
JDR: I mean, we could have a whole discussion there. I guess how much they looked over their shoulder for the longest period of time really was a little surprising for me, having only started covering them in 2013, that was one piece. Another was just how many failures they went through on the grocery and consumable side of the business. I knew a lot of them, and I covered them in real time.
Once you actually put a list together, it’s actually even longer than you realize.
JDR: It is incredible. I remember talking to some Amazon executives who were like, “Oh, forgot about that one, forgot about that one too, oh, and that one as well”. But then to see them today — they’re boasting, I think, on maybe every earnings call, about what they call their Everyday Essentials business now, which is basically everything you’d find in the middle of a grocery store. Maybe not the fresh stuff — they sell the fresh stuff, of course — but all the consumable CPG products.
Going deep on that surprised me a bit and I guess the last thing is on the Walmart side, I knew how many people Amazon poached from Walmart over the years — there was famously a lawsuit back in the day — but I didn’t realize, until I started digging for the book, quite how many Amazon folks had gone over to Walmart, taking a really big paycheck to do so and moving themselves to Bentonville, Arkansas. Even now, one of the most senior folks at Walmart today — a young guy in his early 40s, I think — David Guggina, was a long-time Amazonian.
That is interesting. It makes sense if you think about it, but you wouldn’t think about the bi-directional nature of it. Was the Whole Foods acquisition a success for Amazon? Or does that fall in the failure category — still functional, but not what they expected?
JDR: Yeah, I mean, we’re a decade — almost a decade — out now, I think.
Yeah, I think it was 2016.
JDR: Maybe 2017. I only remember because I was off that day it was announced — I was going shopping for my daughter’s one-year birthday party.
You’re right, 2017.
JDR: I had a tip the night before from someone who would know, saying “Hey, you need to watch out: Amazon is making a big splash in physical retail”. I told myself after I ran these errands Friday morning I was going to make a ton of calls. Well, it was too late.
I think for many years it’s been a failure, the stores kind of lost some of the soul that made them Whole Foods. But I will say in the last couple of years, under new leadership there, and with a little more independence, they seem to have found a stride a bit, so we’ll see.
But a stride that doesn’t really have anything to do with Amazon’s core business.
JDR: I mean, it’s a return hub now, I guess, and I think they’re trying to bring pricing down there. Honestly, I’ll be candid that in the last year or two I’ve not paid as much attention to Whole Foods as I did.
That’s the answer — you weren’t ready for my question because it doesn’t really matter. Well, actually, I want to — you mentioned grocery, which I think is really interesting. To go back to the Walmart question, how important was grocery for Walmart in terms of figuring this out and actually mounting a credible challenge to Amazon?
JDR: Hugely important, I think.
Amazon had it right, they were right to be worried, and they were right to realize they needed to get grocery, and here’s Walmart.
JDR: In my book I tell some stories about how Walmart used to own a UK grocer called Asda, the UK was very early to click-and-collect and online grocery delivery, so Walmart ended up importing a bunch of executives from that team to get online grocery off the ground here. They started with delivery, and it was like, “Oh, this is really expensive, we’re not sure this will work”, Walmart wasn’t a big fan of money-losing expensive things at the time. And I think, again, not to give too much credit to the pandemic, but it was a big factor, they then saw the need for grocery pickup.
I’ve talked to [[Walmart CEO] Doug McMillon, and I know you talked to him before, he was very transparent that COVID changed everything, the good thing was they had the foundation in place. And the other thing with bringing in Marc Lore was that $3.3 billion mattered not because they acquired assets, but because it gave him internal credibility inside the company to actually build up this foundation. So when COVID hit, they were ready to take advantage, but it also was tremendously fortuitous for them.
JDR: Yes, absolutely, and I got lucky to spend some time with Doug for the book, and I thought he was pretty transparent and candid about where they were ahead of the curve, where they weren’t, and where they got a little bit lucky.
Actually, one more question on this, I think the grocery thing is fascinating. What is so interesting about Amazon succeeding with Everyday Essentials is it’s so counter to Amazon — where the whole premise is the everything store, and Everyday Essentials, they’re succeeding by having a dramatically limited amount of SKUs, just close to you. It’s like they’re out-Walmarting Walmart. Because Walmart, their retail store, as you said, is limited — but they’re also pretty good for a retail location in how many SKUs they have. Everyday Essentials is a very small number, and it turned out that was the key, it’s really interesting.
JDR: Yeah. If you look at the history, I spoke to Doug Herrington, who is now Amazon’s — he’s basically the guy who runs the retail business, CEO of Worldwide Amazon Stores or whatever they call it. He was a CPG guy, early online grocery — I’m forgetting whether he was at Kozmo or Webvan or one of those. And he basically said from the earliest days, “If we don’t figure out grocery or Everyday Essentials, we can never fulfill our vision of being there for a customer for every buying need”.
Right.
JDR: You buy a TV every couple years, you buy a camera whatever, but this stuff you buy several times a week.
They’ve just been relentless, decade over decade, trying different paths. We’ll see if this is the end state or not, but investing in the warehousing network, plus trying again and again different ways to make the small incremental purchases make sense — it feels like they’ve gotten there in the last year or two.
What’s interesting about this Everyday Essentials, and you sort of drew this distinction between buying a TV — buying a TV is a fun experience, you do research, you look at all the reviews, it’s an entertainment experience beyond just being an entertainment device. And this applies to a lot of things you might buy. It applies to booking a trip — for most people, they’re not flying every single week, they want to go out there and plan, the anticipation is part of the experience. That’s one category that tech demos tend to focus on and think about. On the other hand, there’s buying milk and eggs and paper towels — things you just do that are not fun, they just need to be done and to your point, Amazon realized these are a continual touchpoint with consumers. If we don’t own that touchpoint, someone else is going to, and they can expand into these other areas.
Agents
Did I just explain the entire agent debate by talking about this specific issue?
JDR: (laughing) Maybe! That’s sort of how I’ve been thinking about it recently — trying to figure out the boundaries for where a personal shopping agent does and doesn’t make sense. Not in a VC’s life, or a super AI-pilled researcher in Silicon Valley’s life, but in an everyday person’s life.
Because it feels like the agent thing — there’s the chatbot angle, where it’s great for planning trips and all those sorts of things and at the end of the day the conversion is almost beside the point, because there’s so much else that goes into it. I’ve made the case that shopping is fun — people do it because it’s enjoyable, it’s not enjoyable to buy paper towels. But to your point, it’s probably underrated that the thing people buy the most is actually everyday essentials, and why not have your agent buy it? Particularly the whole CPG category is predicated on you getting attached to one brand and buying it until you die. Why not have your agent do it? It’s not like it’s fun to buy Old Spice deodorant.
JDR: Yeah, it’s interesting. We have a mutual industry friend in Michael Morton, the analyst over at MoffettNathanson. He had some report — I don’t know when exactly — looking at the Subscribe & Save penetration in Amazon’s business, it was actually a little smaller than you might think, which kind of complicates this discussion.
No, it’s great. It’s smaller than you think, and the Dash Button — which I thought was awesome — totally failed. Are there takeaways from that?
JDR: Well, I think about my own life as a consumer. My wife isn’t in tech or media, so I have, like, a normie living with me, which is helpful for what I cover. I’m sure she would love that’s what I’m saying she’s helpful for.
Oh that’s fine, my wife ends up as fodder for these takes all the time, it’s very useful.
JDR: I’ve thought long and hard about why we aren’t a Subscribe & Save family. For us, at least, little things come up — a trip here, something changes in our kids’ lives and we don’t need that thing as much. I just don’t want to end up with too much or not enough of something. And Amazon has made it, without AI, easy enough that I can get that thing later today or tomorrow anyway. So what are the takeaways for AI agents from that?
Well, no, just to double down on that point. I was in Taiwan for a long time, there’s no Subscribe & Save there, but e-commerce — everything was 24 hours, because it’s small and dense. There’s just a lot of advantage that comes from that, I think it’s underappreciated what delivery in the U.S. means when it’s very large and very spread out. So I moved back, and I’m like, “Oh, Subscribe & Save, that’s going to be great”, so I did it for a bunch of items, and I lasted maybe a month because it was too hard to calibrate. Suddenly I’m in my garage and I have like 13 cases of sparkling water, and I’m like, “This is super annoying”. I think, to your point, it’s because it’s too dumb. It’s not necessarily a counter to say AI agents are going to fail — it might be the case for why they’re going to be compelling.
JDR: Yeah, it might be. And I actually just published a newsletter looking back at Walmart and this little internal startup they had called Jetblack.
Yeah, I was going to ask you about that. I’ve never heard of that before.
JDR: Well, great, I’m glad I wrote about it. It was, I think, partially Marc Lore’s idea for a long time — a personal concierge service. They spun it up in New York, they started actually by targeting wealthy moms, which didn’t make a lot of sense for Walmart proper, but I think they saw consumers who would actually pay for such a service.
It was money-losing, but not more than it should have been. It was basically “text something and get it delivered to you that same day.” You know, “I need size three diapers”, to talk about something in the everyday essentials space, and it would show up later that day. They started building some technology around it, but this is late 2010s, so this is pre-everything we’re talking about today.
People loved the service — a small customer base in New York City — but they were losing gobs of money. Walmart essentially was like, “We’re not going to keep dumping money into this thing”, so I talk about this in my book, Marc Lore went out and found $300 million of outside capital from, I think, FedEx, Visa, and Microsoft, and they were going to spin it off, but they needed Walmart’s blessing, Walmart would have owned 50% of it. They thought it was a done deal, Lore and team, then at the last minute, Doug McMillon decided against it.
The view from the other side was that he was fearful of funding something that might end up being a real competitor down the line. He told me, when I asked him about it, that he wanted to keep the technology for internal use, and that eventually you’d see Jetblack bear fruit in services that Walmart would launch, that was about four years ago when I talked to him.
It all came up in my mind because last week Amazon blocks Muse, and then Walmart is one of the partners of Muse. I had to reach out to the former CEO of Jetblack, a woman named Jenny Fleiss, and I had to say, “You must be thinking about the what-ifs”, and she’s like, “Oh, absolutely, all the customers keep reaching out saying ‘what could have been'”. Obviously a lot could have changed between then and now.
I think though this is a broader point. In the dot-com era, a lot of the ideas were right, but they depended on everyone needing to be connected — not just connected at home, but connected on the go. Once everyone had phones and was always connected to the Internet, all these ideas that seemed ridiculous in the late 90s made sense. And that’s what I’m hearing about this Jetblack service — it made total sense, but it didn’t make sense until you had an AI agent so you could actually do it profitably and to your point, people complained about token costs or inference costs, they had people on the other side, right? It’s a lot cheaper to do it this way.
JDR: Right. And the prime example of “the 90s idea until the technology was there” — and there are many — is Instacart a great one. Until the mobile phone and the gig-worker economy, it didn’t work.
Which depends on the mobile phone, to your point.
JDR: Correct. And so the Webvan and Kozmo jokes of yesteryear — is this another one that we’ll be looking back on? I thought it was interesting to look back from now, but we’ll see.
Meta vs. Amazon
Well, let’s get to the Meta versus Amazon, which prompted me finally having you on — which clearly we should have done ages ago. There are a lot of angles to this — it seems like just a small thing on the surface, but there are lots of layers to unpack. One I sort of hinted at before, how optimistic are you about agents? This is kind of the most important question — is this actually going to be a category or not?
JDR: So I’m playing around a lot with mostly Muse right now, I started with Instinct, and now I’ve kind of invested in Muse heavily the last few days, as Instinct keeps nudging me to pay attention to it again — texting me.
Yep, I have it too, I think it’s kind of given up a little bit.
JDR: And so I will say I’m having a lot of fun, but what I’m also realizing is Muse is failing at most things I’m trying to have it do. I spend too much time on X, the former Twitter, and I see all these Meta execs reposting these what appear to be real use cases. I’m like, “Wow, that sounds awesome, maybe I could get it to do something like that”, and then I’m running into a lot of roadblocks.
I’ll give a couple examples — I made a bunch of notes, and I had Muse grade itself on everything I’ve tried to have it do.
Nice.
JDR: Last night, my son collects sports cards, and after the Yankees were beating the Red Sox, I said, “You know what, we should find the best rookie cards for these two players who had big games”, I asked Muse to go to eBay and compare prices and pricing history and find me a couple of deals on these two players. It ran into a CAPTCHA, and then said, “Can you log in?”, then I took over the browser, which seems like it would be easy, but it’s actually not super intuitive — you have to drag what looks like a cursor on the screen. I’m trusting this is early days, right? But eventually the login didn’t work — it got blocked.
On the shopping side, I similarly tried to reorder my favorites from Nespresso, and it got blocked and shut down trying to log into Nespresso for me too. Then I said, “Can you create an email address and just order these things I want from Nespresso?”, it told me it couldn’t spin up an email address for me. So there’s endless opportunity, it’s just not nailing a lot of the things I wanted to do right now, and I’m going to keep going at it.
There’s endless paper cuts.
JDR: What I’m very curious about is, I’m not the average consumer, right?
Yep. You are literally paid to continue going at it.
JDR: And again I am also doing it because I find this fun, which says something about me maybe. But how many chances do you get with the average consumer? Maybe I’m wrong — maybe people are using it right away to find money that states owe you, one of the early use cases, maybe for some people that’s all they need. But there’s just a lot of work to be done.
I think you put your finger on it. You can see how this would be awesome, but do you burn customer goodwill and willingness to try because it’s not good enough yet? But then if it’s not good enough yet, when is it ever going to be good enough? There’s some aspect of you just have to go for it and figure it out, I think that’s a very good observation. To that point, how essential do you think shopping is going to be to agents? Can agents be compelling without nailing shopping?
JDR: I’ve been at the intersection of retail and tech for a long time. A lot of people feel like shopping will be a nice-to-have, and maybe a vertical shopping agent will nail it. I don’t know if this is a correct comparison, but Amazon’s the everything store, and yet you have vertical players that figured out a specialty — you think about Chewy and their customer service, Wayfair and their wide selection.
Selfishly, for the story, I’d love for some vertical players — whether the retailers like Amazon and Walmart, or AI-native companies, and there are tons of startups — will figure out that the data they sit on, and the taste or merchandising they bring, can be a differentiator. To answer your question head-on, I don’t think shopping is critical for these to work, but I think they need to do a few of the most-wanted shopping requests decently enough, and I’m still mixed on what those are going to be. But for now, Amazon doesn’t need Muse.
Well it’s kind of the question, are agents going to be like an assistant, where you really just want one because it’s too complicated to keep track of different assistants? Or are agents a user interface, where every service has its own agent and that’s just the best way to interact with them? Amazon is sort of making a bet on the latter — shoving Alexa in your face whether you like it or not, the formerly-Rufus, now Alexa, I guess that’s another way to look at what an agent even is going to be.
JDR: I think that’s right. One other point on Amazon, which I may have tweeted about, is something I feel is a little overlooked. When Amazon CEO Andy Jassy got on stage — I think maybe last year, when they reimagined and released Alexa+, the voice assistant part — there are too many Alexa names now, it’s kind of confusing to me, it must be confusing internally.
Right. It’s a Seattle thing, I think.
JDR: He got on stage and he basically said the difference between this Alexa and the old one is, yes, the conversational nature is going to be a lot better, but this is going to take action on your behalf, this is going to actually be an agent. I haven’t gotten a good read yet on what’s gone on the last week, but it feels like maybe they don’t care, because they have it built — their play is now within the Amazon website and app, but the new voice assistant was supposed to be their agent, and it seems like they’ve gotten beaten to the punch in a couple of different ways.
To this point, I want to step back a year ago — OpenAI Instant Checkout. Why do you think that product failed? What was the issue there?
JDR: I think a couple of things, I was just talking about this with someone yesterday. He was saying the reasons everyone talks about have to do with, “They didn’t realize how complicated it was to get real-time inventory and clean product data from these retailers”, so the experience was just not good, you’d click on something and a different product would show up than what was described — I think that was the case.
Walmart publicly, in an interview that still somewhat confuses me because it was with Wired magazine — nothing against Wired, but I’m curious how that interview happened about retail and Instant Checkout — they basically said the conversion rate was terrible compared to our own website.
Cart size was small and conversion rate was terrible.
JDR: That’s believable too. But the other thing was internally — this person yesterday was telling me OpenAI, this was in the middle of, “Oh shit”, where Anthropic’s running away with enterprise, so I think they took some focus and resources away, and realized, “This is just not one of the big bets we should be making right now”. So it was likely a combination of all of those.
What I’m interested to see is Google has said they’re launching checkout across Gemini and Google AI Mode — this is through a protocol they’re building with Shopify called UCP. They announced it in January at a big retail event, and it’s now the end of September, and you basically have to be lucky to come across this checkout capability. I think it’s coming soon, and that’ll be a real test of whether people actually do want to check out on these AI surfaces at all — or whether this will just be a repeat of the buy buttons on social media years ago that failed.
Right. Because it failed on Meta too — they were going to have checkout inside the Meta properties, that was one of the responses to ATT, and it just doesn’t seem to work. Again, is it customer preference? Is it the difficulty of getting actual links to real data? Is it cart size? Is it conversion? Probably the answer is yes, all of the above.
JDR: Google has actually — again, we were talking about the list of Amazon failures in grocery over the years — Google has a not-quite-as-long, but pretty long, list of failures trying to make transaction shopping checkout possible. And I think they keep coming back to will it disrupt their ad business? And if it’s going to disrupt it too much, they don’t want to really invest seriously in it.
Is that the reason why Google is nowhere in the agent space? Are agents the actual threat that we thought chatbots were? It turned out with chatbots the human’s still there, so Google can build a chatbot — and that’s basically what Google Search has become in many respects — and they apparently seem to be monetizing it fine, but the whole thing with agents is there’s not a human to look at the ads.
JDR: One of the biggest questions I’ve been getting the last week is, “Where is Google?”, and my question with this agentic craze is, is this launch they have coming of, I’ll call it old-school checkout on their AI surfaces, going to end up being late, even though nine months ago it looked early?
It’s interesting, I mentioned before — sometimes it’s just like Subscribe & Save, where the issue was it was too dumb. There are ideas that sometimes make sense if you actually have sufficient intelligence and I wonder if this whole question of Instant Checkout on the property — there’s a bit where that’s easier to deliver on from a certain perspective and then to your point, you get in there like, “Holy cow, the back-end stuff is actually unbelievably complicated”. This idea almost applies broadly to API access — you sometimes build the API, and the API has to be kept up to date, and sometimes they just ship something that only shows up in the UI, “Where’s the API for it?”, you see this happen again and again.
Is computer use just the cheat code? Because if the AI can just go in a browser and use the UI, suddenly none of that matters. And if you think about the Amazon and Muse thing, the issue isn’t that Amazon can’t deliver the package Muse orders, they’re literally just blocking it, because Muse can go on Amazon.com without Amazon’s permission and make the order.
JDR: Correct. One piece is it’s not going to see the ads, which is the major profit pool other than AWS that Amazon leans on, the other piece is just touching that data Amazon has built up over what is it, three decades now.
Right. If Muse can log in, it can log into your purchase history. Amazon sends you these emails that say nothing in them so Google can’t grab the data, but if Muse has access to Amazon — what’s the most underrated use case for Amazon? When you get one of those stupid security questions and they ask, “What address did you previously live on?”, guess what — there’s one place on the Internet that has all my addresses since like 1995 or 1996. It’s amazing.
JDR: (laughing) Right. And Jassy, on every earnings call, keeps saying “We will eventually work with” — now granted, this is before the agent launches of the last week or two — but he keeps saying, “We will eventually work with the third-party horizontal AI labs, but only when the experience they’re offering is a good one”. What he’s not saying, “When it’s beneficial to us”. I’ve been thinking of the phrase, “Watch what people do, not what they say”, and they keep blocking the stuff, so I’ve been trying to think what a partnership could look like with direct access, it would have to be lucrative to Amazon in some way, whether in experience or monetarily, and I’m not quite sure what will make it work.
Is this actually the real Amazon counter? It’s not Alexa or Rufus. It’s the deal they announced with OpenAI just a couple weeks ago, which I think is very timely. If the issue is ads — and the ads thing is one of those where I always hate taking the total conventional-wisdom assumption about a thing — but at the end of the day for Amazon, the ad business drives all the profits of the stores. I think that’s probably the biggest issue and reason why. And I think this is also underappreciated as a second-order effect, if we get to a world where there are no ads, or they don’t matter, that means everything’s more expensive — be careful what you wish for when you’re shopping on Amazon. But what I see happening with Amazon ads in ChatGPT, where now ChatGPT is going to be a surface for Amazon ads to show up at, is they’ve sort of flipped this on its head. OpenAI was trying to build a checkout, and Amazon of course wasn’t interested. But if we flip it around, “Oh, you’re an ad surface, now we’re all in and we’ll help you figure it out”.
JDR: Yeah, that might just be it. I was talking to a retailer recently that launched something very similar — I think it was Target — and got none of the press or analyst coverage that Amazon just did but it’s because Amazon’s been standing off with these players every other way, and now this was the way that makes sense for them. So you’re right, next earnings call is the end of October, and it’ll be interesting to hear Andy Jassy talk about whether this was the mutually beneficial deal, or whether we’re just going to see them continue to block.
They block Muse, and they have this ongoing lawsuit with Perplexity over — and speaking of Perplexity, very early to the browser agent with Comet — just didn’t catch on for a lot of reasons. Maybe that it was Perplexity and not one of the big tech companies launching it, we’ll see if the courts force Amazon’s hand, but I think you’re right to think maybe this ad play is the answer.
It just speaks to — it actually re-emphasizes how important the ads are. Is that ultimately — I’ll get to Muse’s angles in a moment — but for Amazon, this is like another reason, they invested in OpenAI earlier this year. I talked to Matt Garman and Sam Altman together about having agents on Bedrock, but this is actually really important to the retail business too, their response to Muse is arguably to make sure ChatGPT wins.
JDR: I haven’t exactly thought about it that way, are you saying you think we’ll see them double down in some ways?
I’m saying any agent player who is willing to accommodate Amazon ads will be better, because they’ll have access to Amazon’s logistics to actually deliver on the agentic experience.
JDR: I think that could make a lot of sense. The other wrinkle here, I had some folks who pay a ton of attention to AWS pointed out to me on Twitter the other day, is that Meta does a lot of business with AWS. Just another little wrinkle, not that I think it’ll really play a role, but Amazon picking sides in this, whether indirectly or directly, could for sure be something I’m going to watch going forward.
Right. So how important was it to bring this full circle for Muse — that they get up on stage at Meta Connect and one of the partners is Walmart?
JDR: It’s nice to say, what remains to be seen to me is how heavily Walmart is actually leaning in and partnering. It’s not totally clear what the partnership looks like.
But they don’t have to do anything, right? As long as they don’t block it, Muse can just use Walmart.com, that’s the whole thing. Ideally API access is better, ideally access to inventory is better, but all that functionality is already surfaced on Walmart.com, as long as Muse can use Walmart.com, the end user could say, “Get some paper towels”, and Muse can actually deliver. I think Walmart’s the only company other than Amazon that could actually credibly fulfill that.
JDR: I think that’s right. I’m just not 100% confident it will be as free as that. Based on talking to some Walmart folks — they have a couple of different CTOs, a little confusing there too — and talking to them at an event last year, they talked a lot about wanting to be more open, but that it would have to make commercial sense. And maybe you’re right, maybe they’re just looking at it as incremental, especially with Amazon not involved and that’ll be it, but I was honestly trying to get an answer to this before I came on today, and I’m seeing a missed call right now from Walmart PR, maybe they were just a few minutes late.
But listen, I’ve written about this for years — Walmart in many ways had to just play catch-up with Amazon — with the marketplace, with their own fulfillment service, with their advertising business. But I have been waiting to see if there was any way they would try to get ahead and not just follow from behind.
Right. The analogy I go back to is the iPhone. The iPhone, in country after country, had to launch on the second-place carrier, sometimes even the third-place carrier — I think Japan is the canonical example, that was a brilliant play by Softbank. Yes, they ended up with relatively lower margins, yes they lost control of the customer experience, but they made it up in volume, both in the number of customers and these massive new data plans they ended up selling.
And from Walmart’s perspective, I initially was not sure, because of that interview you talked about where Walmart was very disappointed in the Instant Checkout reality, but it still matters they were on Instant Checkout early on. And you can see, of course Amazon’s not going to do that, which means it’s a chance for us, it’s a chance to reach customers who are the least likely to know how good Walmart e-commerce is, because they’re going to be more techie, probably more likely to be on the coast, all these sorts of things we’re talking about. In retrospect, I’m not surprised they jumped on. Just like Amazon will prop up any agent who wants to show our ads, Walmart is like, “We’ll jump on the train of anyone who’s getting customer traction”.
JDR: Yeah, I think that’s right. And I’m really excited to do more digging and reporting and finally get some execs on the phone from over there because I think the opportunity is super interesting for them. A few years ago, they almost became a significant minority owner of TikTok back when TikTok was first going to get banned.
That’s right.
JDR: And I wrote back then that I thought they were going to have a deep commercial partnership, I was like, “This is interesting, Amazon has sort of failed at social video and the influencer set, and this could be a different angle for Walmart”, and it ended up falling apart for a variety of reasons.
But the logic was there, even if the deal ended up not coming together.
JDR: I think it was. Listen, they’re a very impressive company, Doug McMillon is no longer the CEO, but John Furner, his replacement, is digitally savvy. He’s only been in the role about nine months, so we’ll see what his leadership looks like and his decisions over the years, but they are, in some ways, to be feared, and a great partner to have if you can’t have Amazon.
Well, the issue for Amazon in the very, very long run is they have this massive moat because of the huge amount of real-world investments they’ve made in fulfillment centers, distribution centers, robotics — all those sorts of things. That’s a reason to think they’ll be fine no matter what happens with AI. The problem is investments in the real world have to be paid for, they need to be used, this is what happened to Amazon in COVID, they overbuilt, and it was a very heavy weight on their earnings and stock price — because that’s the challenge with physical goods. If you do too much, you get in trouble, and the long-term issue is, yes, they don’t need Muse, but at some point, if Muse gets big enough, they might need Muse. It could just sort of tip.
JDR: Yeah, that’s exactly right. As someone who’s covered them — I guess 13 years now — it’s very rare, I’m trying to think of another example of them finding someone they really needed. Maybe ChatGPT, and them even pursuing the ad initiative there is a sign they kind of did need them in a way, but for someone who tells business narratives and loves good stories, it would be great for them to find themselves in a spot where they’re having to make really difficult decisions.
Actually vulnerable, because they’ve never seen vulnerable ever in a million years. Yeah, that’s true.
Shopify and The Aisle
Real quick, Shopify — where do they figure in this? I think it’s notable Shopify is, “We’re all in on Muse as long as you use Shop Pay”, is that the Shopify takeaway right there?
JDR: Yeah, I think they want to be able to tell their merchants and Wall Street and all the technology partners that, “We’re going to be anywhere a customer, an end customer, might want to be”. I think AI discovery has been a tailwind for them.
For sure. AI uncovers the long tail, and that’s where they’re good.
JDR: And this is a point I’ve heard Amazon execs make on a podcast, and I wrote about a bit in my newsletter today about Walmart, the endless aisle at some point is overwhelming unless you have something recommending and curating. The recommendation engines — ChatGPT and others — have been really good, I think, for some of the long tail.
Another angle for them is they want to get bigger and bigger brands to work with them, that’s midsize and enterprise. One angle is they’re saying, “Hey, you don’t have to be a Shopify customer historically for us to surface your product catalogs across all these new AI surfaces, just sign up for our agentic plan.” I’m curious how that will help them go upstream and get more of the enterprise customers they really desire, but that’s a big part of this too.
Yeah, for sure. Speaking of aisles, you said “the endless aisle”, if I want to read about the aisle, I can go to readtheaisle.com. How has it been as an independent entrepreneur writing a newsletter? Has it given you any surprising insights into commerce, even though you’re not necessarily selling paper towels?
JDR: Oh man. It’s been great, I’m growing each week, I don’t regret the decision — those feel like maybe low bars to some, but good bars to me.
That’s, by the way, a takeaway from this, it’s just steady growth, there’s no exponential curve.
JDR: Well, maybe once every six weeks there’ll be an influx of subscribers, either off a newsletter or an appearance like this, but otherwise that percentage growth, I’ve been shocked how steady it is. It’s just the same thing over and over.
I think I’ve done a good job getting into my long-term thinking, it’s like the best kind of slog. I’m getting immediate feedback that you don’t necessarily get from a big media company publication, I’m going out and meeting people — throwing parties for subscribers, I’m hosting dinners, that’s one of the ways I’m monetizing, getting one smart sponsor in, and helping them get the right people at the table. So I’m able to nerd out even more than I ever could on the stuff I love, and do it with people who are on the leading edge building and/or trying not to get toppled over by this new AI wave in retail. It’s lived up to everything I imagined, both the good and the bad.
Longer to maybe work than you might have expected?
JDR: I’m only 10 months in, so I think I have product-market fit just based on the subscribers I have, from all the companies and positions they’re in. And then meeting people in person and getting largely very good feedback, and some constructive criticism on what they want more of. I am a slow writer — I’m not offering up three or four times a week content like you are, but I’ll be coming out with other formats that I think will fill in some of the gaps I’m leaving open right now.
In the future, if I ask you “give me your work history”, is this the last one? Are you going to be The Aisle?
JDR: I really hope so. I think the timing was really good too.
I think it’s fine being a slow writer, by the way. People have asked for advice, and people see what — I started in 2013, that’s part of why I get the privilege of writing about the biggest, I cover the big five tech companies, dabble in media, dabble in things like Walmart, stuff like that. But that doesn’t mean the opportunity isn’t to try to do what I did, the take I’ve always had is if you want to be the biggest fish in the pond, find your own pond, that’s the angle. And like I said at the beginning, when I think retail — last week, when the Amazon/Muse thing happened, my immediate thought was, “What does Jason have to say about this?”, and I’m like, “Wait, why haven’t I had Jason on Stratechery before?”, so I think it’s fantastic. I think your product-market fit makes total sense to me, I’m thrilled to hear it’s succeeding. Everyone should go subscribe — readtheaisle.com. Jason, thanks for coming on.
JDR: Thanks so much for having me, Ben.
This Daily Update Interview is also available as a podcast. To receive it in your podcast player, visit Stratechery.
The Daily Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.
Thanks for being a supporter, and have a great day!