Latest / Retail Media Breakfast Club / Walmart is Betting on AI Agents, While Amazon Goes Proprietary
Transcript
- Walmart's AI bet - interoperable === [00:00:00] [00:00:00] Breaking News: Walmart's AI Shopping Agent --- news broke last week from the Wall Street Journal that Walmart is preparing to welcome its next customer, the AI shopping agent. I'm gonna share an overview of the news, a few perspectives that have been shared by others in the space so far about what this means. And then offer my own. Let's jump in. [00:00:17] Introduction to Retail Media Breakfast Club --- Good morning and welcome to the Retail Media Breakfast Club. I'm your host, Kerry Masters, and every weekday at 6:00 AM Eastern, I bring you candid and real discussions in the world of retail, media and marketplaces in just 10 minutes. So bring your own cup of coffee and we'll get smarter about retail media together. [00:00:38] Walmart's AI Shopping Agent: Key Takeaways --- So part one, what is Walmart doing here? Here's a few key takeaways from the Wall Street Journal piece. One is that Walmart is developing its own shopping agents for basic tasks like reordering groceries and handling complex prompts like Plan Unicorn themed party for my daughter. This goes beyond the basic assistance that we're [00:01:00] seeing every retailer and their mother launch recently, which is a glorified search system. This shopping agents can conduct a multi-step process for a user, and could learn preferences and things like that. So that is part one of what Walmart is doing here. Second is that they are preparing for a future where they're going to interact with external. Agents with external consumer facing shopping agents or perhaps other types of intermediaries. So they are pursuing a, double track here, both with developing their own and opening up the possibility for other agents to interact with their product catalog and assortment and things like that. Walmart's, CTO says that agent to agent protocols are being developed, but they're not yet ready, which is quite interesting. We are gonna get into this a little bit more. These protocols let third party agents communicate directly with retailers agents. [00:02:00] Number three, search ranking influences agent behavior. Open AI's operator considers how prominently pages rank in web searches, including paid ads. Highly ranked pages are more likely to appear in an operator, and this is really one of the big questions that we have in the industry around what is the role of sponsored product ads going forward? Number four, interoperability is a long word That's difficult to say. We're all gonna need to start practicing this one walmart is preparing for customers using both Walmart's own agents and third party agents. And takeaway number five from this article is that agent shopping is fundamentally different. Bots may ignore emotional imagery, and retailers will need split second pricing decisions. So how does this all tie into other AI developments that we're seeing elsewhere in the retail industry? And are any other retailers out there doing [00:03:00] something similar or at least talking about it? [00:03:01] Comparing Walmart and Amazon's AI Strategies --- And the answer is no. Amazon is the only one getting close, but they are not pursuing this interoperability. Path that Walmart is. I'm getting pictures constantly now from retailers who've launched their own generative AI assistant for consumers, or maybe on the employee side, like helping employees locate products and talk about. Features and benefits to customers and things like that. And that's sort of the stage that most retailers are at. They're building sort of glorified search engines within a generative AI assistant on the front end, But it's not really answering this new path to purchase that is emerging, which is I want to use my own AI assistant that I have come know and love and who knows all this stuff about me and can help me make decisions about. Brands and trends and things [00:04:00] like that, I wanna use that same system rather than going over to a retailer and starting over from scratch. So Amazon is the only one that really comes close to what Walmart is talking about here, which is. They have built their own consumer facing AI agents, which includes Alexa Plus and buy. For me, it has a native, shopping assistant Rufuss, and then they have also launched a developer focus tool called Nova Act, which could, in theory be used to create AI shopping agents that could interface with other retailers. So sort of setting a bit of a double standard here, like, oh, we'll give you the tools to develop agents to visit other retailers, but we won't let any agents visit us. I don't see that changing anytime soon. [00:04:45] Technical Insights: Open Standards and Protocols --- So in terms of the tech here, there's a couple of when we talk about open standards and Walmart is talking about these model protocols being developed, there's two that you should understand at least a little about. One is Google's agent to agent [00:05:00] protocol, or a to a, and the other is anthropics. Model context protocol or MCP, which lets AI models connect with external tools and data sources. This is what could ultimately enable Walmart's interoperability bet. So it's sort of like an API where the agent can plug into the backend, get the information it needs, rather than navigating to the website. Typing out a search query, clicking through things just to tell you which, pack of diapers has the best price per unit or something like that. So it's an instantaneous request rather than actually sending sort of web scraping traffic to a website. That's just my. Ultra dumbed down version of how it works. But the long and short of it is that Amazon is building a walled garden, whereas Walmart is preparing for an open ecosystem. [00:05:58] Industry Perspectives on AI Shopping Agents --- All right, let's get into what other people [00:06:00] have said so far about this. . One is that the low hanging fruit here, especially for brands, but also for retailers, is product content. So my old friend Eric Heller made this point that. One thing we should never lose sight of is the importance of complete PDPs. I still see pages without complete or informative images or that are missing key information in the search bullets. It won't matter how good your machine to machine API is if you have incorrect information on the PDP. And I agree it's a great call for many brands and for retailers too, is the content needs to be up to date. Even Amazon, I saw something recently. they have very clearly reconciling information from their catalog with D two C sites, particularly for their agent capabilities too. Another interesting take, which I, I don't. Necessarily agree with is from Brian got, and Brian [00:07:00] says, I feel like a lot of organizations have failed to ask one thing, do people want AI help? 'cause I mostly don't, and it's increasingly annoying to get a forced on me. And my response here is I think we need to look at some cohort data here and. The reality is a lot of people are becoming like, kind of scarily close to chat GPT. Probably seen some memes around people, it's sort of like their best friend now. For what it's worth,, I agree that the AI assistance on retailer sites. Can be very underwhelming, including rufuss. So I do not, I do not disagree there. But what we're talking about here is not just an AI assistant on a website that's kind of dumb we're really talking about your own personal AI agent that understands you, and it's that experience of collaborating with a chat GPT, for example, on product discovery. That is amazing. And that's what this kind of technology is gonna be solving for [00:08:00] not putting another chat bot on a retailer.com. Some other. [00:08:06] Final Thoughts and Conclusion --- Thoughts from my side,. One is, what the heck happens to sponsored product ads? This is really top of mind for a lot of people. Do agents see them? I've seen one post suggesting that operator focuses on the top product results, which includes sponsored placements, but I think there's a little bit more investigation required on how agents actually process paid placements versus organic results. I haven't seen a whole lot of research there. I'd love to hear if you . Number two, retailers are rushing to build these proprietary AI assistance. But my question is, what is the opportunity cost of doing that versus preparing for third party agent integration? Walmart, their move is hedging. They're doing both. They also have a lot of resources so they can afford to do both. And I see all these retailers out there, adding the chat bot kind of [00:09:00] sucks and. Hoping that there's gonna be some excitement around that, and there isn't. People don't like them, I wonder if some of these retailers would spend their time and resources a little more wisely by looking at these new protocols and building around them. Number three is site performance could become competitive. What's gonna happen with all these agents cruising around your website? It's gonna create all these traffic spikes. False impressions, false impressions on your organic content, on the ads. Gonna throw the ad. Units out of whack. A direct agent to agent interface, eliminates these issues. Number four, I think that that's gonna be part of the preference that agents develop is they're probably gonna want to go and transact with retailers that have an agent protocol on the other side because it takes less compute power for them to do that. One of the interesting takeaways after OpenAI operator launched [00:10:00] was people were. Blown away by what it was able to do. But the common takeaway was it took a long time to do it. Watching the agent go and like navigate to this website, type in a query, try and find the dropdown box, have to try and maybe like do a capture form or something like that. So I think that your AI agent is gonna have a preference for interacting with retailers that have a context protocol on the other side. All right. Gone a little long here. But the conclusion is really that brands and retailers that get this right won't be the ones with the flashiest AI features on their websites necessarily. They're just gonna be the ones that make it seamless for both humans and AI to do business with them.