Retail Media Breakfast Club

AI shopping agents are starting to change how people discover products, compare prices, and actually make purchases. But as these agents take on more of the shopping journey, there’s a much bigger question we need to be asking: who’s going to pay for them, and will they still be working in the shopper’s best interest when they do?

In this episode, I’m sharing some of my favorite moments from Jason Goldberg and Scot Wingo’s discussion on The Jason & Scot Show, including a fascinating story about how Walmart approached AI adoption, why Amazon is building its own shopping agent ecosystem, and what happens when AI agents start handling the actual execution of a purchase. I also share details about my new podcast/newsletter AdGentic, dig into why advertising could ultimately become the business model behind personal AI agents, and what that means for retail media, brands, and consumer trust.

This episode is sponsored by GrowthLoop

Timeline

[01:00] – The Walmart AI Wake-Up Call: How a conversation with former Walmart CEO Doug McMillon illustrates why AI adoption has to start at the top.
[03:23] – Why Amazon Is Building Its Own AI Shopping Ecosystem: Could AI agents become another shopping and media surface for Amazon to dominate?
[05:05] – What Actually Counts as Agentic Commerce? Scot Wingo’s broader definition of agentic commerce goes beyond autonomous shopping to include any AI-assisted part of the purchase journey.
[06:11] – The Shift From AI Discovery to AI Checkout: Why shoppers handing over price comparisons, shipping research, and checkout to AI agents could disrupt retail media in unexpected ways.
[08:50] – Who Pays for AI Shopping Agents? Jason Goldberg and Scot Wingo explore monetization, affiliate commissions, subscriptions, and the challenge of keeping agents on the shopper’s side.
[11:34] – Will Advertising Fund the Future of AI Agents? Why I think ads could become the dominant business model for personal AI agents — and what we should be asking about relevance, transparency, and trust.
[14:55] – Is Shopping Fun or Just a Chore? Why shoppers may want to keep the enjoyable parts of shopping while handing over the tedious work to AI agents.

Links & Resources

What is Retail Media Breakfast Club?

10 minutes of expert insights every weekday. Your morning ritual for staying ahead in retail media.

Who’s Going To Pay For Your Shopping Agent? [Snips]
===

[00:00:00] kiri: Good morning and welcome back to Retail Media Breakfast Club. I'm Kerry Masters, and if you listen to the very popular podcast, The Jason and Scott Show, you might have heard my voice at the very top of yesterday's episode, and that's because I've launched a [00:00:15] new show. I'll tell you about it in just a minute.

[00:00:18] The episode itself is Jason and Scott live on stage debating agentic commerce while the audience fires some questions off at them. In case you're not familiar with the [00:00:30] show, Jason Goldberg is Chief Commerce Strategy Officer at Publicis, and Scott Wingo is CEO of RefyBuy.

[00:00:37] And full disclosure, my new show is part of the RefyBuy media network alongside theirs.

[00:00:43]

[00:00:44] kiri: So today I'm [00:00:45] gonna share a few snippets from that episode that I loved and add some thoughts of my own in there

[00:00:51] And the episode begins with a story that Jason tells about Walmart's recently retired CEO, Doug McMillon, and it is the [00:01:00] best illustration I've heard of how a giant company like Walmart actually decides to change

[00:01:06] jason: But before he retired, he... I got an opportunity to go to a meeting with him, and this is about two and a half years ago. And we come to this meeting, it's a- Need a mic runner? He's [00:01:15] the 38th most powerful man in the world according to Fortune Magazine.

[00:01:18] It's a big deal to get a meeting with Doug, and people super prepare for the meeting, right? And so we come to this meeting. Everybody's ready. They're super excited. And Doug stands up and says, "Hey, thank you, you guys, very much for coming to the [00:01:30] meeting. I know you worked really hard to get ready for this.

[00:01:31] But before we start, I have a question for you. How'd y'all use AI to prepare for this meeting?" And you could literally see people's faces just deflate. Like, they were so bummed, because nobody had used AI two and a half [00:01:45] years ago to prepare for a meeting at Walmart. And there was this awkward silence. It felt like eight minutes.

[00:01:49] It was probably, like, eight seconds. And then some super brave soul chimes in- "Gosh, Doug, I'm really sorry. We didn't use AI to prepare for this meeting," which, not shocking, right? [00:02:00] And then he pauses, and then he drops the bomb. "AI is illegal at Walmart." And everyone's like, "What? What are you talking about?

[00:02:07] AI's not illegal at Walmart." And D- and this guy go- explains to Doug McMillon. He's like, "Oh, I get a email from IT every week. You're [00:02:15] not licensed to use Claude. Don't use Claude. And ChatGPT hallucinates, and don't you dare put any of our corporate data on there where it might leak." And it suddenly occurred to Doug McMillon, the CEO of Walmart, that like, "Oh my God, this is what's [00:02:30] gonna disrupt us," right?

[00:02:31] That this is what's gonna knock us off of the top of the, the retailer list, is that we have 1.8 million associates that are all doing their job today the way they did it yesterday, and they, they don't feel [00:02:45] like they have permission to do it different. And so he went on this, this whole pilgrimage to make AI innate at, at Walmart, and they, they bought ChatGPT for 1.8 million associates, and then they bought everything.

[00:02:57]

[00:02:58] kiri: AI is [00:03:00] illegal at Walmart. Imagine hearing that from your own team. Maybe you actually have. Here's the thing, when we think about innovation, we tend to picture startups and small teams that have less to lose, more appetite for [00:03:15] risk. They're not public companies. And then you hear a story like this, and the expectation resets.

[00:03:23] a Fortune one company and the push for change came from the very top. Now Amazon [00:03:30] is the other example I keep coming back to because of any retailer, it probably has the most to lose if AI agents pull shopping away from its search results page. But it got in early with [00:03:45] Rufus and started building its own agent ecosystem.

[00:03:49] Now, whether or not their posture around rejecting all AI assistants and agents, regardless of how that pans out, [00:04:00] it's not a one-way door, by the way. They could decide to change that at any minute if it suited them.

[00:04:05] But anyway, they've built their own agent ecosystem, and if that plays out, that moment that looks like an existential crisis for Amazon could become [00:04:15] one more shopping and media surface that Amazon dominates. So I love this illustration from Jason from the stage, just showing how innovation continues to happen at these huge companies.

[00:04:29] And [00:04:30] just because you're big and things move kind of slowly

[00:04:33] That's not really an excuse when you hear a story like that. Um, then the rest of the episode gets into this, like, kind of debate, uh, they call it. It's, it's not so much, um, [00:04:45] a, a debate as a discussion with the audience. But a good chunk of the initial setup is spent on talking about what agentic commerce even means.

[00:04:55] They talk about three different definitions. The first two are pretty cut and dry, [00:05:00] fully autonomous buying or a journey that has to end in an AI-driven checkout.

[00:05:05] But there's another one with a fuzzier definition, and I think this is where the real threat lies. Let's listen to Scott

[00:05:12] scot: And then number three- Yeah ... and spoiler alert, [00:05:15] this is kind of what I believe, I think any time an AI assists you in the journey, that that should be called agentic commerce. So I count when you're using, uh, uh, uh, you know, Rufus, which is now called Alexa for Shopping, or AFS, when you're using that, when you're using [00:05:30] Sparky, you may actually use it on the discovery part and not want to do the checkout.

[00:05:34] You may want it to go do... I use these new personal agents, I use them for the checkout a lot, because I'll use it, I'll do the discovery myself and I'll say to it... We ha- we do a lot of tailgating, I'm in North [00:05:45] Carolina, and I found this really cool cooler, and I just said to the agent, "Hey, go see if you can find this for a better price and, uh, faster shipping."

[00:05:52] And it did, and I said, "Rip, rip it, rip and roll." Like, you know, "You do it." And it just gave me a Stripe link. I clicked that, and it bought it. [00:06:00] So I use it as kinda like a GPS. I kind of put it in there whenever I feel like I want it to help me out a little bit and make sure I'm getting a great deal, or I've got all the right parameters of the thing I'm doing.

[00:06:10]

[00:06:11] kiri: And so notice which half of the journey Scott actually handed [00:06:15] over. He found the cooler online himself. did all of his discovery himself. The agent did the legwork after that, finding the better price, finding the faster shipping, doing the actual checkout. and so much of the narrative around [00:06:30] AI shopping has been about discovery, that AI is the new search, that it's just about discovery, that people are going to wanna check out themselves.

[00:06:39] And that's where a lot of the anxiety in retail media sits as well. A few [00:06:45] months ago, I was talking about a hair tool purchase that I made, the Shark Glossie, and how ChatGPT did the research, and I landed on Best Buy already knowing what SKU, SKU I was going to buy, and I [00:07:00] finalized the purchase from there, and I argued, "Hey, this is n- now, like, sidestepping all of these ad units I would have previously seen."

[00:07:08] But Scott is using his agent for the actual execution of the purchase, [00:07:15] and honestly, that didn't really seem possible or palatable until pretty recently with the rollout of some of these more advanced AI agents like Meta Muse, Dots, [00:07:30] Grokbot and Instinct. So what if a lot of shoppers end up using agents his way? Then the disruption is gonna end up somewhere other than where a lot of us have been thinking about it

[00:07:48] Kiri Masters: Costco isn't scaling its RMN with legacy ad tech and manual solutions. It's leaning into a cloud-centric composable [00:08:00] stack. As part of the Costco Velocity Network, GrowthLoop empowers the retail media team to build audiences directly from the Data Cloud, enabling faster activation, greater [00:08:15] relevance, and better performance, all while maintaining privacy and governance standards.

[00:08:22] The result? Exceptional value and experience for Costco members and better ROI for brand [00:08:30] advertisers. Learn why the GrowthLoop Composable Commerce Media Solution was the right choice for Costco's Velocity Network. Visit go.growthloop.com/breakfast. [00:08:45] That's go.growthloop.com/breakfast

[00:08:50] kiri: Next up, an audience member asks the question that I think about a lot: How are these agents going to make money, and will they still be on [00:09:00] my side when they do?

[00:09:01] scot: I like having someone on my side." But you know, they're free right now. So I would, I would like to pay them a, a subsc- I want them... I'm happy as a consumer to pay for that to make sure it's always on my side.

[00:09:13] Because I, you know, [00:09:15] the other, you know, a lot of us in the industry, many people at a show like this, the most frequently said thing I say when, "Have you tried Muse?" They say, "I remember when Facebook walls were free," and they told us that would be great, and then the, you know, Zuckerberg will screw us over, is kind of like what everyone at this show [00:09:30] tends to feel many times.

[00:09:31] So that's real though, like he has a history of doing that. So it's gonna be inter- today the incentives are great on those two things, but they're giving it away free. It ain't cheap paying for all this imprints they're running. So at some point someone's gonna have to pay. Once the, [00:09:45] once the, if the merchant is paying through an affiliate link or something, then suddenly it's no longer gonna be incentive to fight for me.

[00:09:51] So it's gonna be interesting to see, you know, these early days I think there will be someone that decides, I'm gonna have a pay- the consumer pays for [00:10:00] this a bit, and that, but it's always on their side, and I think that will be a really pretty in- interesting winning combination on that one.

[00:10:05] jason: Yeah. And that, yeah, I think it's a wonderful question because I think those are two critical things are understanding the, the incentives and the monetization and how that [00:10:15] impacts trust.

[00:10:16] We're, later we're gonna talk about robots that the retailer owns, vertical robots versus horizontal robots that the, the LLMs own. And trust dynamics are very different there. Does anyone here trust A- Alexa for shopping [00:10:30] to point you at the most successful product for you? No. Right? Like, 'cause Jordan does, but he's a-

[00:10:36] he's kind of, he's very gullible. And Sam, Sam Altman very famously said like, "We get it wrong all the time," [00:10:45] but the customer believes we're trying to get it right. Like, when we get it wrong, it's not because we're trying to trick the consumer, it's because we just screwed up. And so at the moment that's the play out, but the incentives are likely shifting because, you know, Scott [00:11:00] is a very successful man and he has no qualms about paying for a service that then works for him.

[00:11:06] The vast majority of consumers don't do that, right? Correct. And Ben, Ben Thompson on Stratechery very famously talks about all the models that involve the consumer [00:11:15] paying for the service don't work. Yep. And he, he uses Dropbox as the example, and we're seeing this play out on OpenAI right now. Sam Altman started out very anti-ads, and they thought, "Hey, we'll get consumers to pay for this."

[00:11:27] And already ads are a very [00:11:30] robust monetization model on, on OpenAI.

[00:11:33]

[00:11:34] kiri: Now that point that Jason pulls from Ben Thompson at Stratechery, that consumers mostly don't wanna pay for this stuff and so ads end up [00:11:45] funding these ecosystems, that is pretty much the premise of the first episode of my new show. It is called Ad agentic, and it's about advertising in the age of AI agents.

[00:11:58] In the first episode, [00:12:00] which dropped yesterday, I look at how today's personal agents like Meta's Muse, OpenAI's Dots, Grokbot, a startup called Instinct, they all say that they're going to pay for themselves somehow, that they're gonna [00:12:15] save people so much money that they're gonna be willing to pay for it.

[00:12:19] Well, I'm not buying that. Subscriptions cap out at the number of people that are willing to pay, and most people won't. [00:12:30] Fees that you pay, incentives, commissions, affiliate fees, those are sneaky. And once an agent earns money on every purchase, merchants are gonna be paying to get to the front of the line.

[00:12:44] And you know what that sounds [00:12:45] like? An ad. So my bet is that these companies will put these creative fee schemes to the side ultimately and fund these AI agents with [00:13:00] ads. The open question is: what kind of ads? Are they going to be relevant? Are they gonna be clearly labeled, Or are they gonna be the kind that is subversive and annoying or weird?

[00:13:13] So that new show, Ad [00:13:15] agentic, is available wherever you listen to this podcast. It is on YouTube as well, and there is a newsletter on Substack and on LinkedIn also. So every way you could possibly want to consume it, that's where you can find Ad [00:13:30] agentic.

[00:13:30] But now back to the debate on The Jason and Scott Show. The next question from the audience was about how this looks across different categories, such as fashion and apparel.

[00:13:40] Surely a category like fashion that is so driven by [00:13:45] brand works differently. Let's listen to Scott's answer to that question

[00:13:49] scot: And a lady sat on the front lo- row, and she was very aggressive that, "I will never... I love fashion, and I will never use an agent." And that's fine. You know, I think in my definition, that's...

[00:13:59] If you [00:14:00] love shopping for dresses and you wanna go pick it out, touch it, make sure it fits, go do that. But I have two daughters, and they do that. They go to the store. They find the perfect dress. They love it. And then my 19-year-old, she gets great enjoyment from [00:14:15] battling it out on Depop, Posh, and then there's a whole...

[00:14:18] What's the... Vinted, Vinted? I don't know what Vinted's all about. Everyone talks about Vinted now. So she has an agent go do that. She says, "This is the dress I want in this size. Go hammer them. The retail price is, you know, [00:14:30] 150. Mom and Dad aren't paying that. Go find it for 75." And it just, like, sits there. So she...

[00:14:35] I count that as agent of commerce. So that's one use case I can... My, my j- I bet I can make it fit almost any argument, because you would, [00:14:45] you can use it for that, and you still get all the enjoyment of finding it, but now you're even more happy 'cause you saved half off.

[00:14:50] jason: Yeah.

[00:14:50] scot: What's not to love about that?

[00:14:51] So, so that's one way I could see the fashion thing work.

[00:14:54]

[00:14:55] kiri: Now, I love this example because Scott's daughter keeps the part that she [00:15:00] enjoys. She enjoys the thrill of the chase. She found the perfect dress, but the agent is the one going and doing the haggling to work within her budget. I wrote about this a while back in a piece called [00:15:15] Not All Shopping is Fun.

[00:15:17] A lot of the time when people push back on agentic commerce, a reason that they give is, "But shopping is fun. People enjoy shopping." And my argument is, yeah, some of it is. Some of [00:15:30] it's fun, but plenty of it is a chore, and which parts are fun versus a chore depends a lot on who is shopping, in what category, and what that mission looks like on the day.

[00:15:44] [00:15:45] Restocking the pantry is a chore for a lot of people. Tracking down a secondhand dress on marketplaces can be the fun part for one person and a slog for the other person. My argument in that piece was that [00:16:00] people are ready to hand over the chores

[00:16:03] And what's been holding them back so far is the technology which hasn't yet been mature or trustworthy enough to hand them over to, but that is starting to change

[00:16:14] So wrapping up [00:16:15] here, every thread that got pulled in this debate leads back to the same place. Whether agents do our discovery or just the grunt work, whether we're browsing for fun or we're restocking on autopilot, somebody [00:16:30] has to pay for the agent.

[00:16:31] And once they do, it changes whose interest the agent has. And so that's the question I'll be chasing down on Adgentic. If you wanna listen to the full episode of the [00:16:45] podcast that I just took some snippets from, that is The Jason and Scott Show, It's linked in the show notes along with Not All Shopping Is Fun.

[00:16:53] And if the monetization question keeps you up at night too, come find me over on Adgentic. [00:17:00] Thanks for tuning in, and I'll catch you tomorrow.

[00:17:02]