Okay. Hello. Thank you. Welcome, everybody. I'm Simeon Gutman, Morgan Stanley's hardline, broadline food retail analyst. My pleasure to welcome Daniel Danker, EVP, AI Acceleration and Product Design, from Walmart, most recently with Instacart as Chief Product Officer in online grocery. Thank you, Walmart, for being here third year in a row. It probably took three years to be anointed as a tech company. One introduction for Daniel before we get into this. I was talking with Doug about two months ago as an outgoing conversation. We talked about some of his hardest decisions. I asked him about one of his best decisions. I didn't know Daniel yet, he mentioned it was hiring Daniel as someone at the enterprise level who can help advance AI. High expectations, sorry about that.
Starting with your role, it's new for Walmart. What drew you to Bentonville?
I mean, it's pretty unusual to find a company that's as grounded, I think, as Walmart is, that's got really strong values. It operates at such tremendous scale that it can serve so many customers. A little bit more unusual to find a company like that that also recognizes the role of technology the way it does. You know, Doug often would talk about Walmart as people-led and tech-powered. Those aren't just words. They matter a lot, and I think especially right now, it gives a lot of purpose to technology in a way that I think really matters and I really wanted to be a part of it. It's easier said than done. To be truly tech-powered, you need to invest in technology like a technology company, and I think Walmart has done that.
Now the creation of this role is interesting because it reflects both a desire for growth and speed using technology that I think we can really make happen. Also a recognition that if we're going to make the most of what AI can enable, that it needs to be a role that sits at the executive council level, alongside our business teams and other global functions as well. It's been great to see the discussion of AI and the discussion of technology permeate all of our exec team meetings in the way that it has, and I think that makes its way throughout the whole organization.
This nuance between growth and cost efficiency.
Yeah
thinking about how AI can be applied, is there one priority versus the other for you?
You know, of course, we're always going to focus on operating the most efficient way we possibly can because we wanna serve customers with the best prices we possibly can. That's always been the case. I think it's actually easier to figure out how to drive efficiency than it is to think through growth because you only have to look at the things you're already doing and think about how you might do them a little bit more efficiently. I think growth is much more interesting as a vector here. In some ways, there are things we've wanted to do for customers for many years that we did not have the technology to do them with, and all of a sudden, we have the technology now.
I think that's a much more exciting way to go, and it's reflected in how we're prioritizing. We view AI as something that will enable us to move faster and drive more growth, and that's a big part of my focus.
Curious, just as an outsider, anything surprise you so far?
Yeah, because for many of us, I think we associate large company with bureaucracy and moving slowly. That was my biggest fear in joining. I knew that I was very strongly spiritually aligned to the values. I knew that I was very strongly spiritually aligned to the core customer value prop that we focus on. I was worried that the company would move slowly because it's big. I think what I've discovered is just how quickly Walmart executes. I think in part it's because of this constant retail-led this retail mindset of you have to win the customer every single day, and that competition is a core part of how the world works, and you win by serving the customer better than anyone else can.
I think that has not only driven a good value to customers, it's also driven velocity inside of the company. It moves really fast and that's been both a relief and something that's very exciting to see.
Jumping in, is agentic commerce good or bad for Walmart and why?
I think it's good for a couple of reasons. I look like I hesitated when I said that. I think it's very good for a couple of reasons. One is there's a lot of things that people buy every single week, and AI and agentic commerce enables us to put those things on autopilot. You don't really wanna spend all your time thinking about how to fill the fridge or replenish the pantry, or know that you're about to run out of laundry detergent, only to realize that the next time you do laundry, and then you're scrambling. We can put all of that on autopilot, and AI enables us to do that. Agentic commerce enables us to do that. Those are the things that I call the chores. Even just saying the word chores evokes the right emotional reaction in all of us.
It's not the bit you wanna be spending your time on. Agentic commerce lets us put that on autopilot. I'd say the other thing that enables us to do is take all that time that we get back and focus it on much more personalized, much better product discovery than has been possible in the past. In a sense, if you think about it, the store does such a good job of creating this immersive, panoramic experience where you're delighted in all the senses are delighted with thousands of items, very carefully and intentionally placed. An online experience has always had this promise of better personalization, more attuned to your household and your individual needs, but the immersiveness was lacking.
All of a sudden, with AI, we get to upgrade the personalization and upgrade the entire shopping experience around it in ways that weren't possible with prior technologies. Agentic commerce enables us to solve both of those, put the chores on repeat, automate them, make them not feel like chores anymore, and then really draw customers into these additional shopping occasions that are far more interesting, far more nuanced, far more personalized. That's a really big part of what I think we're gonna talk about today 'cause I think it's a really important part of what agentic commerce does.
One of our premises is that agentic actually expands the market for commerce.
Yeah.
Do you agree with that?
I do. I do. Because you have to ask yourself, what are people buying today and how are they buying? What is it that agentic commerce enables us to do that we couldn't do before? What kind of shopping occasion does it unlock that we couldn't unlock before? That's, that's a really important starting point for the conversation, because otherwise all we're going to do is create a slightly different interface to the thing that we're already doing. That's not that interesting. No, agentic commerce is an expansive solution. Now, it needs a little bit of definition. I think a lot of people equate agentic commerce with robotic commerce. This idea that you've somehow handed the wheel to a computer that will now shop on your behalf.
If we go back to the scenarios we just talked about, the chores part might actually go that way. You might just benefit from letting an agent do all that shopping for you. By the way, that's new, but not that new. We've always had the ability for customers to put certain items on subscription or repeat order. This is just a more sophisticated version of that. It's the next version of it, if you will. I think that's good. That's let's almost call that robotic commerce, because agentic commerce just means that one AI agent and another AI agent might be working together to complete a task. It doesn't mean that the computer has replaced the human in that. You know, if anything, I think what we've seen is that we're, what?
30 years into e-commerce, and people still love going to the store. because people love shopping. That's what's really happening. People love shopping. I don't think they're in a hurry to try to let something else take over that experience because it's actually quite enjoyable to see options, to see what goes together, to choose. Something we love to do as humans. I don't know that you need to choose the same butter for the eighth week in a row. Maybe not. But if you're shopping for apparel, for beauty, for home, for, I mean, countless categories, of course you wanna participate in that shopping experience. I think when we say agentic commerce, it almost implies that those journeys become automatic. I don't see that future at all. I almost would love for us, let's just do this.
Let's start this right here. Invent this new word of robotic commerce to distinguish it from what agentic commerce can do in how it serves people and how people engage with it. I think they're quite different.
I wanna drill down on that and talk about disintermediation. I neglected to read a disclosure, so I'll read it real quick. Sorry for this.
This is the moment to talk.
Yes, exactly. For important disclosures, please see the Morgan Stanley Research Disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. Thank you. There's a lot of chatter on Wall Street, and I see headlines about incumbent retailers like Walmart being disintermediated, so traffic going somewhere other than your app, for example. How do you react to that?
Well, first of all, let's maybe break down how people shop online with Walmart. The traditional, by far most common place that people go inside of our app, you all do it's the search box. The search box drives the vast majority of items that people are adding to cart. There's this second way of shopping with us that we've launched over the last year, which we call Sparky, and it is our shopping agent. It's not automatic. It's just a different way into the shopping journey. It lives at the bottom of the app. It's a conversational experience around shopping. That's the second way to shop. The third is what happens when you're using Sparky, but inside of ChatGPT or other LLMs that live outside of Walmart.
One of the things that I spend a lot of time on is trying to understand what are people buying in each of these places. When we look at what people are buying on our search box, it's food and consumables. They're doing their weekly grocery shop. Top items, probably not going to surprise anybody here. Bananas. Americans love bananas. Bananas, bread, milk, eggs. It's your weekly grocery shop, and that's what people are doing inside of search. Inside of Sparky, the kind of a bit of a fork in the road that we're seeing. One direction that people go with it is automating what I just described as the weekly shop. It's when Sparky pops up and says, "I notice you tend to buy the same items every week.
Do you want me to just add them to your cart?" "I notice that you're about to run out of laundry detergent because you seem to replace it every 56 days, and it's been 53 days. Do you want me to add laundry detergent?" Okay. That's automating the weekly shop and the replenishment, the keeping the house stocked. I think that's great if people wanna automate that. The second track is discovery-oriented, and it expands from that weekly shop into a whole bunch of new scenarios that I think are really interesting. I think these are scenarios where, through my eyes, I see growth. I'll give you examples of some of the top items that show up there. Phone chargers, like I just need a charger that will work with this phone. Okay, cool.
Easier to do in a conversational experience than in a search box. Tires, we sell a lot of tires because a lot of our super centers have Auto Care Centers. If you just think about what would you even type into a search box if you needed tires? Like, would you remember the size, the exact... I wouldn't even know where to start. In Sparky, you just say, "I need tires." It says, "What kind of car do you have?" You tell it what kind of car you have, and then it says, "Do you want one tire or four tires? Do you need them today, or can you wait till tomorrow? If you need them today, these are the ones that are in the super center right now.
If you can wait till tomorrow, you have a few more choices. Makes tons of sense for a conversational experience. These are all very new shopping occasions and very additive to the grocery scenario that we just talked about. That's search box versus Sparky, but they're both inside of our app. We recently launched Sparky inside of ChatGPT, and we started seeing patterns. We started noticing what are people buying there. You know, one version of this story could have gone the direction that they just buy the same things that they're buying inside of the Walmart app. Here's what we're learning. Top two items that people are buying from Walmart on ChatGPT, vitamin supplements and protein supplements. You're probably all thinking the same thing at this point, of what was the prompt that led to those purchases?
Do you wanna say it? Feel free. It's not a prompt that starts from the place of commerce. It's a prompt that probably began with, "I just started taking GLP-1. What do I need to know?" Something along those lines. Of course, the response probably went down the path of, "Well, here's what you need to do with your sleeping routines. Here's how you need to change your exercise habits. Here's how you need to change your eating habits. Also, you're probably gonna need more protein." Eventually, that conversation led to something of a commerce experience, and that's where we stepped in, and that's where we were able to satisfy a need.
That's also why these are additive experiences to what's happening in the app, because they didn't necessarily start from a place of commerce, but we stepped in when it made sense to have a commerce journey around it. I think they're very additive. They're very complementary to each other. It's why I see growth. I think it's very interesting when you can follow the customer's lead when you design these experiences, and I think that's what we're able to do here.
In that third mode.
Yeah.
than where the user interacts inside of ChatGPT, isn't that disintermediation?
I see where you're going. I would argue it's intermediation. I'll tell you why. Today, you're seeing the most rudimentary version of that experience. It's what's called native checkout. Customer enters a prompt into ChatGPT. It comes back with results that it knows are products that exist on Walmart. You can even buy those products literally without ever visiting Walmart in that journey. This is a very temporary moment in time. By this time next month, you will not see that experience anymore. You will see is that the Sparky experience will travel directly into ChatGPT and Gemini and anybody else that we ever integrate with.
The way that works is that when you discover an item and ChatGPT or any of these LLMs has decided to surface an item in response to a query, and that item is at Walmart, when a customer taps it will open up Sparky, and Sparky will take over the shopping journey at that point. It's gonna happen directly within ChatGPT, and the customer will be able to do a number of things. First of all, they're logged in at that point. By the way, that also means that we know what else they've been adding to their cart throughout the week 'cause people tend to add stuff to their cart throughout the week. We'll bundle those into the same order. We'll know whether they're members, so they can get great fees, great prices.
We'll also know if there's other items that maybe they should add to their cart alongside this item. If you added protein supplements, there may be other things you need as well, and we can suggest those. By the way, they might not even finish the journey there. They might come back a day later and go to Walmart, the Walmart app and complete the purchase then. It's all one and the same. It's very fluid between the two. The important bit here is that the Walmart experience lives inside of and travels into all these places where customers can shop. In a sense, if you compare that to what exists today, it's a deep linking-oriented world.
You go to a website, you do a search, it comes back with results, you click, it takes you out of that site into a completely different site. What's the first thing you see when you land on those sites? Do you accept cookies? Unless you're shopping for cookies, that was not the right answer. That's the kind of disjointed experience that we see today. All we're doing here when we talk about agent to agent is bringing our shopping agent into this other environment so that the journey is more connected. It's more integrated, intermediated. It's more unified. You can pick up in one place and continue and not feel like you're being ripped out of that experience. That's all that is. I think it's a better shopping journey.
You're building this model for us in agentic. I just wanna paraphrase before we go to the next. You're saying Walmart is going to coexist in name your LLM?
Exactly.
Whatever model, that agent will talk to Sparky controls data and the user experience from that point, and then consummates the transaction.
It decides at that point which items to show, what order to show them in. It does that on a very personalized basis because it knows about each customer that's coming in. In all likelihood, by the way, we know about those customers from before. They've probably shopped our stores, and they've shopped us online. We can continue where they left off.
Sparky, LLM.
Why does Sparky get prioritized? Why does Walmart win in that context?
It's a good question. In a sense, the customer is putting in a query, those LLMs are gonna have a choice to make across a whole bunch of different retailers and a whole bunch of different items. This is where it comes back to our customer value proposition. You know, what we stand for and what customers rely on us for are several things. One, we have enormous assortment, half a billion items available across our stores and our Marketplace. We show up a lot because we have a lot of shopping occasions that we can satisfy. Second, price. We're known for a good price, you know you're gonna get a good value when you shop on Walmart. Third, speed. Why is it that those things matter?
They matter to customers, they also matter to Walmart, they also matter to these LLMs because the LLMs have to serve the customer's need. They don't work if they only kind of serve it. If they want customers to keep coming back, then they need to show products that reflect how customers would want to shop. There's another component to that that I just think is important that we don't forget. Even though selection, speed, and price are very tangible ways to rank, trust is an incredibly important part of that journey, and it's something that Walmart has had at its core for decades. If you think about what happens when you're shopping online, you need to know that you're gonna get the product you wanted.
You need to know that if it shows up and it's got a certain price or whatnot, that that's probably the right price. You need to be confident in that. You also need to know that if there's a problem, you can return it. You're not gonna need to haggle at that point. Trust is a really big part of that end-to-end journey. It's on these merits that I think that we can expect that the LLMs their incentives are aligned with the customer's incentives when they show Walmart, and it's why I think that we will show up quite a lot. If you use ChatGPT today, you'll notice that we do show up quite a lot, when you do enter queries. I think we win on our merits.
We have to talk about ads. Before we get there, I wanna just focus back. Traditional LLM talking with Sparky, why does it create an accelerator for growth? Are there examples you can give? Earlier, I heard the Marketplace example, which I think is pretty intriguing.
The exciting thing is that we have a lot of customers that come to us every single week to place grocery orders. Shopping with Walmart is already something that happens on repeat for millions and millions of households. In a sense, they've connected a Walmart-shaped pipe into their house, but what they put through that pipe today online is first and foremost food, consumables, those kinds of items. What I think is exciting about this and why I view it through the lens of growth is that it's introducing us into shopping journeys and shopping occasions for very different items. For electronics, for beauty, for fashion, for home, for categories that go way beyond the categories that people traditionally shop us online every single week.
If we can insert ourselves into those conversations and serve a customer need, that becomes very complementary to what we were already doing with those customers. It enables those customers, by the way, to benefit a lot, most of our orders come from members, and so they're getting more out of their membership in the process. The reason I think that it's important that you evoked the word Marketplace is because, you know, if a typical supercenter has between 150,000 and 180,000 items, in our Marketplace, we have almost half a billion items available for customers.
These shopping occasions can get quite nuanced and quite specific, and in all likelihood, we're able to serve those needs through the Marketplace in a way that customers just don't have as much motor memory around using us for yet, but that's what I think is gonna change here.
Are you impressed by the bones of Walmart's Marketplace?
I am pretty excited about it. It's also growing very, very quickly, both in terms of assortment and price and speed. On all of those dimensions, which matter just as much in the Marketplace as they matter with grocery delivery and online pickup and delivery is what we often call it. The Marketplace is on a very fast pace to growing on all those dimensions, which I think is really, really cool.
Retail media, big dollars, big margin. I think we can have a fair debate around agentic, why Walmart can win, why it's gonna grow the market. The debate that we fall short on is how advertising dollars fall through the funnel, what these models look like. I leave it open-ended how this could all play out and how does it work in Walmart's favor.
You almost have to go back to what is it that we're charging advertisers for. When advertisers pay us, they're paying us because we can help them sell products and reach customers. Even though we all have a little bit of a display ads component to our advertising business, fundamentally, we get paid when we generate sales. We generate sales in many, many different ways, and we generate sales on many, many different surfaces. If you consider that from an ads perspective, the advertiser pays us for conversion, then what matters is that in all the places where we show up, we play a role in choosing the products that customers are going to see and determining the order in which they see them. Of course, when we do that, we have to be extremely respectful of what the customer is looking for.
You know, if we just showed products that where we had ads against those products, but we didn't show products that were particularly relevant to a customer's need in that moment, they wouldn't convert. Wouldn't work for the customer, wouldn't work for the advertiser, wouldn't work for Walmart. It's the incentives are very aligned here. If you consider that we're essentially being paid to drive sales in a very highly relevant way to customers, then that is something that travels neatly, whether it's happening inside of our app or whether it's happening inside of a Sparky experience that lives elsewhere, that shows up elsewhere. That's why we're pretty excited about what happens when we drive new shopping occasions, because it gives us new opportunities, new moments against which to offer that value prop, both to customers and to advertisers.
If I go to an LLM, I have some occasion or some need, and it prioritizes Walmart because of speed, price, assortment, trust. You've won that. Once there's a handshake to Sparky, you still control that advertising process, meaning you should still retain the advertising dollar and then see that growth. Is that a fair way to think about it?
The way to think about it is that inside of that window that shows up when Sparky comes up, we're still playing exactly the same role there in terms of choosing which items show up in what order, what are the complementary items to those? How do we build a full basket? How do we do that in a way that's informed by that customer's past engagement with us, so it's just as personalized of an experience, even inclusive of the things that they've purchased from us in stores and elsewhere. All of that happens in the same way inside of a ChatGPT environment where Sparky shows up as it does when those customers are coming to our app.
Right. As long as we don't go to this robotic commerce world in which, you know, humans are just no longer making decisions, but robots are, where there may be no advertising in place, as long as that doesn't happen, which it doesn't sound like is your premise.
I think even in that world.
Okay
...there is a choice that is made around which items are surfaced to the foreground. It might just be that a human being isn't looking at them, but a computer is choosing. Even in that world, the choice of which products to show and in what order plays a big role. I fundamentally believe that customers will only use shopping agents, truly robotic shopping agents, if they can count on those agents making similar decisions to the ones they would have made themselves. I don't view them as fundamentally different from each other because the customer needs to be happy with the outcome no matter what.
Even in that environment, there's still a choice to be made around which items show up and in what order. It still factors the advertiser's needs in exactly the same way.
Yeah. On that point, more philosophical, how do you know that customers actually want agents to shop for them? Is it a build it and they will come reality?
I don't know that they do. I think that they do in some circumstances. Going back to our analogy of the chore, I think that customers definitely wanna take chores away. Show of hands if you wanna hold on to every chore in your life, right? We got one hand. There's always one. I don't think that's something customers actually want to take the effort out of what they need to do every single week that's just on repeat. For sure. I think shopping agents, robotic shopping agents will play a role there, no question about it. How much of the rest of the shopping occasion do they want completely automated, truly hand the wheel off? I don't know. I think we need to learn.
I don't think it's nearly as big of a portion of those shopping journeys as the headlines seem to imply right now. I think it's just because if you look at shopping behaviors, whether they were physical or digital, whether they were on-demand delivery or shipping-based, et cetera, some things remain true, which is people seem to love shopping and they like to browse and they like the agency. Weirdly, I use the word agency, which is not to say that they want to hand it off to an agent. I think that agents may help in the selection process.
I think they may be able to do things like give you a hand with knowing what will look good in your house or what would, what pieces of clothing go well together, or how to treat a fever in a child that's 3 years old versus one that's 5 years old. That might help in the discovery process. Sparky is very excited to help with all of these things. Whether that results in truly hands off the wheel, I just don't think that it will in most cases. It's not reflective of the human behavior.
It sounds like the Walmart agentic experience is about to change profoundly in the next few weeks. What is the timeframe for when the consumer experience broadly will see what's agentic mean for them? Meaning, we don't know.
Yeah
...you know, in e-commerce, we didn't know how far it would go until we understood what the capabilities were.
Yeah.
It took time to build.
Yeah. In a sense, the way we're approaching this is to add a lot of functionality into Sparky to see what sticks and what doesn't. Now, we don't have to guess quite as much as this might imply because customers are going in and entering queries today into Sparky. So we're pretty obsessively looking at those and realizing where are they interested in going next? Which questions were we able to answer well, and which ones weren't we able to answer well? You know, I think one interesting difference between the ChatGPTs and Geminis of the world and Sparky is that those LLMs seem to like to have a conversation, a prolonged conversation. At the end of every answer, they ask you whether there's more, whether they should make a graph, whether they should turn it into a PDF. We've all seen it, right?
Sparky's a little more goal-oriented because customers that come to Walmart are there to shop. Customers that come to Sparky are trying to solve a problem. We just wanna help you solve that problem as quickly as possible. We are learning in the process which things require more back and forth, which things are we able to do well, which things are we not able to do super well. We're plugging those holes. At the same time, we're also taking a crack at some things that we think would be magical for customers that they haven't even asked for yet because they couldn't have imagined that they would be possible. I'll give you my favorite example right now is today when we all shop for clothes online, we scroll through photos of other people wearing those clothes. It's a little weird.
I don't think the next generation is going to believe us when we tell them that years ago, we used to scroll through other people wearing clothes. It'll seem odd. We didn't have the technology with which to fix this until just recently. We didn't have the technology with which to show you all of those clothes on you, so you didn't have to do the mental gymnastics of imagining what they would look like on you. Now we do. Nobody's asking for that because I don't think it's in our consciousness to ask for it, but we're building those things and as we build them, we roll them out and people suddenly have a different experience, and it's an unexpected moment of delight. Today it's a moment of delight. In 10 years it'll seem like it must have always been that way.
In that vein, maybe we'll close on other barriers. What are other biggest barriers to agentic commerce taking off?
Customers need to be taken on a journey to adopt new technology. Right now there's this spike of new technology reaching customers, and I think it's, at times it's actually ahead of where they are in the adoption journey, and I think a little bit more care needs to be given to whether the technology is working well. One of the reasons that we're not just haphazardly rushing to launch as many new features as possible is because we want everything that we build to work well for customers. There are some just core basic needs that we're still just scratching the surface of.
It's a breadth versus depth thing, I think a little bit, and I wanna make sure that as our team builds these new capabilities, that we do a really good job, that it serves all customers, and that it's reliable. I think there's a little bit of just making sure we build with quality, not just rush AI things out the door. That's a thing we think a lot about.
Okay. Well, on that note, we will close. Thank you very much for participating, Daniel. We learned a lot about agentic advertising, retail media. Appreciate it.
Thank you.
Thank you.