Why Agents Will Buy, But Marketing Must Stay Human

Speaker A: Yeah, it went from being like, I'm marketing to humans and this is the track. And what I hear from many marketing leaders is like that other piece, marketing to robots. Very unclear.

Speaker B: Marketing to robots is not a thing. It doesn't have anything that we use our original marketing suite of tools to interact with the emotional narrative, the storytelling. It's like trying to market to a calculator.

Speaker A: Idea of robots isn't exactly new, so why all the sudden attention? The big driver artificial intelligence. Today, my guest is Katya Forbes, the founder of the C and the author of the field guide on how to do marketing when the customer is no longer human. Katia wrote the playbook on machine customers and how AI is reshaping the customer experience so she can tell us all the details.

Speaker B: I think marketers are going to have to be running a dual track. Humans aren't going to go away and are still going to want to interact with brands. But there's going to be this new track that has to run which is capturing the agentic ecosystem still operates in data. It must be discoverable. Your data needs to be clean. But I feel like this is just the table stakes of, um, being able to operate in an agentic economy. And I think for marketers now, trust is the product that we're marketing here in this moment in time.

Speaker A: Katya, welcome to Marketing Trends.

Speaker B: Uh, thank you. It's an absolute privilege to be here.

Speaker A: Yeah, I'm excited because you are such a unique guest and even on our prep call when I was talking to you, I'm like, oh, man, so many good, juicy ideas and contrarian takes. And so I want to jump right in because I know that you just got back from China and when I was talking to you, you were like, man, the way they think about marketing, the way they do it is so different. And so I actually just want to start there while it's fresh on your mind of like, what did you see and what did you go into it thinking that now you are not thinking about anymore? It's like shifted you completely.

Speaker B: Well, I saw a lot. So the visit that I did was to Shanghai, Hangzhou and Shenzhen. And I was able to visit Alibaba, Tencent Unitree Robotics and byd. So some really amazing, uh, progressive Chinese companies who are, uh, doing amazing things in the AI space. You know, physical AI space with the robots and the cars going and visiting. Those was fantastic. I think the stuff that really stuck out for me, going to see the robotics with my own eyes was super enlightening. And what I think a Lot of people are doing is going and looking at the YouTube videos of robots running marathons, robots doing dances, um, and forming opinions, I guess, on what they see there. I had an opportunity to interrogate their VP of marketing and one of their product people about what they're doing there, how they're doing it, and the, I guess, the approaches that they're taking. I mean, they're very used to showing around foreign visitors. They're very, um, all of those companies have got the ways that they, I guess, market to foreign executives. Um, and for each of them, it would involve a tour of the things that they would want you to see and showcase. Um, and then sitting down and having an executive conversation and executive briefing on the other parts of the narrative, whether they want to do a bit of a deep dive and offering us the opportunity to ask questions. The critical thinking is welcomed. So we're absolutely able to ask, uh, some tough questions. The robotics one is the one that really sticks out because I think as humans we have visceral reactions to a humanoid robot, um, the quadruped robot, which, to be honest, I just found really quite creepy. We observe them showcasing the skills and abilities of these robots by pushing them around and showing us that they didn't fall over. This was some of the demonstrations, but from a, uh, uh, capability perspective, watching them run through their routine of dancing, there was one that could do ballet. I mean, it wasn't a Russian ballet or anything like that in terms of capability, but it was still pretty impressive. Um, and seeing where they're up to, uh, in what they are showcasing as embodied AI. Now, the conversation that I had with the executive, taking it back to the machine, customer perspective is, all right, well, what if the robot needed to transact either on behalf of its human or on behalf of itself? I mean, we see the Boston Dynamics Atlas robot. It can swap its own battery packs. So what business models and transactional opportunities does that create? Um, and so I asked this question and it was a little bit like I'd asked the question of, well, what do robots dream about? They're like, wait, what? Then, in having that further discourse with them, um, came to understand about unitary robotics and that they are a hardware company. They, they sell hardware, they sell the balance, the body, um, but it's up to their customers to put the brain into it. And I didn't know that before I went there. You know, maybe more fool me for, for not doing that deep dive on exactly how unitree robotics does things. But I, I learned about how they create a hardware offering and they market that hardware offering on the capabilities and showcasing of what it can do. Um, and then the customers who purchase them pair them with LLMs to put the brain in there. Um, and I saw a really interesting exploration by a team in India who'd bought a G1 robot, paired it with a LLM to give it a brain, given it a voice from 11 labs, um, creating sort of the deli girl vibe. They dressed it in cargo pants and a crop top, but it didn't have a waist. Like they don't have waists so it looked weird. So they 3D printed it a waist and then had it sort of go around the streets to interact with people who were, uh, you know, informal markets selling tea on the roadside to have a robot come up and say I want to buy some tea. And so seeing how it goes from factory offering as a hardware through to a customer who creates the embodied AI, through to that embodied AI walking around the streets in India trying to order tea and have conversations. This is a huge direction of travel confirmation for me of what machine customers are going to be capable and likely to be doing in the not too distant future. So I think we're going to see robots in home before the end of this, this decade. So what that looks like from a business models to support it transactional capabilities product set is going to be really wild.

Speaker A: Wow, okay. See that's why I'm glad I started with this question because it already is having me think about so many different angles for this one. And also just, I mean the term machine customers, it's not one that I've heard on the show because so many people are talking about AI agents right now and that's you know, mostly what many of the conversations are about. So it is interesting.

Speaker B: One machine customer type. Yeah, like there's many types of machine customers and an AI agent is absolutely one of them. But it's not really big enough to describe all of the different types of machine customers that we, we are seeing. So I've identified about five. One is that like a delegated agent that is that agentic AI that's operating on behalf of its human to go and do stuff in the world. We're seeing a lot of plays in this space. We just saw Google I, um, launching their Spark agent, this, you know, at the time of recording this week. So we've, we've got a lot happening in that area. There's also the autonomous buyer, so the machine customer that buys stuff on its own behalf. Things like um, a ah, factory running predictive Maintenance with AI that can then hook into its erp, uh system and order parts on behalf of itself because it's predicting something's going to fail. In the factory we see already a lot of the co buyer type which is the human working together with the AI, uh to research things, to fact check things in the middle of sales conversations, um, to then go on and purchase things then. The one that I like the most, which I also saw the nascent state of in uh China is the multi agent network. So this is something that could be as down at the scale of a smart home with a bunch of agents that work inside the home to make it all work all the way, scaling it up to a smart city or a smart government. So in China Shenzhen is really moving towards smart city where it has a lot of things that are going to be or already are being controlled by AI. And in the UAE I've also been keeping, so I keep quite a global eye open. Um, in the UAE earlier this month they announced that they're planning to run about 50% of their government services using Agentic AI by 2028. So that's what a year and a half away. And you might think that's a hand wavy future but they back that up the following week launching their work permit agentic AI that can now if it's a clean filing for a work permit, put it almost straight through things that took months, weeks now taking days, hours showcasing in practice agentic AI starting to work on government services. So that's that multi agent network type. And then the, the last one that I've been seeing in market has been the intermediary broker which is the one that sits between the personal business and, and the transaction, the, the ultimate transaction I guess drop shipper is what's happening because of the disintermediation of this one. This one looks like Amazon's Rufus that will you know, help you decide on stuff and read all the reviews for you on the Amazon platform or Walmart. Sparky. Another example of that intermediary broker where it's kind of working on behalf of the human but also on behalf of the platform that owns it. So agentic AI covers some of that but not all of that. Um, and so I think that taking that broad view helps us be ready for all different types of machine customers, AI agents, robots, cars that can buy stuff, um, to come knocking digitally on our door to try and transact.

Speaker A: So before I go outside of the international topic, what did you see at Alibaba? I think you went there as well, and you were like your, your mind was blown by something that you saw there. Tell me more about that.

Speaker B: Yeah. Alibaba is agentic commerce at epic scale. Earlier this year in the Lunar New Year festival, um, their spring festival, which is, that's the moment in time where we see the dancing robots and lots of campaigns happening in China because people are, uh, you know, have time because they've got some, some holiday. And also it's a big moment for the region to be showcasing itself. Um, over spring festival this year, Alibaba ran a campaign to give people 25 yuan coupons in order for them to use the Chen AI model that Alibaba has created to customize and order bubble tea. You had to use the AI agent to find a location next to you or close by, uh, to be able to uh, customize the bubble tea that you wanted. You then use it to place the order using Alipay, which is their payment rail that sits again in the Alibaba ecosystem. Uh, and then what they did was they used the Alibaba logistics network in order to deliver the bubble tea from the bubble tea shop to your home. So the full stack all lives inside the Alibaba ecosystem. And they did 10 million orders in the first nine hours and 120 million orders across the life of that campaign. Which is an epic response to a marketing campaign. And this is basically the most expensive onboarding campaign that has been run in the history of onboarding campaigns to get people to delegate a, uh, buying decision to an AI agent and run the payment through that agent and get an outcome in the real world from that interaction. That's a mind blowing, epic scale agent e commerce story and going and visiting them and talking to them about that, um, coming to understand what they're trying to do there. Um, because the intention for that model is for it to go wide into a lot of different places. And I also visited BYD while I was there and in doing some, some follow up research about the Chinese car manufacturers, nine of the main Chinese car manufacturers, uh, using Chuan as the brain for their intelligent cockpits in their cars. So that model is now in cars. And not only Chinese models, but BMW China is also putting Chuan into the intelligent cockpit for their car. So we're seeing the sort of land and expand, I guess, um, of the Alibaba, um, product suite and ecosystem into all kinds of different places, uh, which is, it's just fascinating to see their intent showing up in so many different places, uh, and creating these new machine customers. So the cars in China, they're intelligent cockpits. I was in the showcase in byd, and, uh, the lady who was running us through it, uh, said to us, uh, almost like she was describing windscreen wipers or something really boring, uh, well, our car, it can tell whether you're unhappy or feeling stressed, and it can change the mood inside. It can play some nice music for you. And I was like, wait, what? What? So there's this new AI driven emotional inference of how you're feeling. Um, and then I had a conversation with the executive after that, and I was like, well, so what happens when the car wants to do a transaction maybe on behalf of what it saw about its human? And again, he was like, oh, but we have Alibaba in the dashboard of the car, and it's also connected, so I don't see why the car would want to buy something. We just use Alipay for everything. I was like, aha. Huh. I realized, uh, what was really nice and confirming for that was that I'm still a little bit ahead of the most futuristic of people, so that's cool. Or I haven't met the right person in BYD yet. But then in the conversation of me explaining why a car would want to do a transaction on behalf of its human, say the human is, you know, it's tired and needs some, you know, some support. The intelligent cockpit goes, oh, you're tired. I'll tell you what I'm going to do. I'm going to buy dinner of the takeaway. I'll navigate you past that on the way home. And you bring, uh, the takeaway home so you don't have to cook. There's a tiny leak that has to happen to wire those two things together. And someone's going to do it. Because when I was talking to the BYD executive, he was like, actually, that's a really good idea, which made me laugh. I'm sure somebody is working on it, um, already, uh, within, um, those car ecosystems. But being in China and seeing how it's all infusing in all of the different places just gave me a really interesting, um, perspective on, I guess, at least the future direction of travel, if not, like, what's in market today, that perhaps in Western markets, because maybe we don't have our eyes on it. We're not aware that this stuff is in market today, and it's actually surging ahead. And, I mean, I can talk a little bit about the reasons why that is in China for sure, but I'll pause there for a second because that was a lot.

Speaker A: I know for today's buyers that next purchase is never more than a tap away. A moment of curiosity at lunch, a nudge at midnight, an impulse on the train ride home. Attentive helps brands show up in all of those moments with the right message on the right channel at exactly the right time. And when every message feels that relevant, it doesn't just feel better, it performs better. That's what marketing made personal feels like. That's attentive. To learn more, visit attentive.com? that's coming to mind now is like, what are uh, how can I take what's happening in China and apply it in the us? When I think about like we have, you know, laws and regulations that don't allow companies to just be a big monopoly that then all of a sudden they're every part of your life. Like in China, like we, you kind of can't do that. Which I think maybe does help with like what success they're having that in many ways like most companies can't have what they have in the US at least. So like how do you take what you see and then apply it to a US based company that you know, has regulations that don't maybe allow for what China has?

Speaker B: Yeah, ah, I mean on that emotional inference, uh, in the UK there's for the eu, um, sorry, in, in the eu, with the EU AI act you can't do that. You like, you can't infer people's emotions and then, you know, create outcomes based on that. However, in the cars that are in the eu, um, and also in the States and Australia and other, you know, markets like that, um, they have the same sort of sensors and they will have the same sort of AI. But it's presented as driver safety. Are uh, your eyes on the road? Hey, your eyes are off the road. Pay attention. Are we creating the conditions for you to be comfortable in the car so that you have your full attention on the really quite important task of driving the car? Um, and so it's the same sort of underpinning technology but presented as a different package because of the nature of what is right, wrong, allowable, not allowable in those markets. Um, so there is, there is a little bit of that sort of workaround um, going on. Um, in China they don't have changes of government so they can set a 35 year strategy and they can stick to it knowing that the administration that's coming in isn't going to undo all of the things that the previous administration put in place so people can run you know, 35 year plans. And because there is that monopoly across product stacks. So Alibaba owns from the AI model through to the payment rail, all the way through to the logistics network without frictional fragmentation. Whereas in the US we see, you know, Visa, Ah, MasterCard, Stripe, WorldPay, all of the payment rails, uh, are sort of fighting for their piece of the pie here, um, we have multiple AI houses who are offering different things and different capabilities and also sort of, you know, pushing against each other for primacy in this space. And so that tension creates, I don't know, better outcomes because competition does, does create great, great outcomes. And also constraint. We can be creative when we have constraint. If there's something we can't do, then we have to find a way around it and be creative to do something differently to get to the same outcome. China's done that with their AI models. They can't buy in Nvidia chips and so they have to create models that are lighter, that are cheaper, um, that don't need that kind of processing power, which they have done with uh, deep Seq and the other models that they have. So with constraint comes creativity. So I think, um, taking that mindset into what we're doing in markets where it isn't quite as streamlined because of the nature of our governments, our regulations and things like that, um, is go, all right, well given this constraint, what's the creative way that I can still get to the outcome that I'm trying to get to? And let's, let's pull marketing down to like the brass tax of, of what it's about. If we look at it at the purest level, and I'm sure this is debatable and somebody on the Internet will go, but, uh, what about blah, blah? Marketing is about the creation of demand for the product and service that the business is trying to put out into the market and generate revenue from. That is something we've done through brand narrative, we've done sometimes through emotional storytelling, we've done through explaining features. Um, all of those kinds of things are ways that we marketed to get to that outcome. And the challenge I think marketing has now is to look at the purest thing of the purpose, the purest purpose that they have and go, all right, well what got us here isn't going to get us to there. So now how do we do this differently for the immediate future? I think marketers are going to have to be running a dual track. They're going to be running a track for humans because humans aren't going to go away, um, and are still going to want to interact with brands and understand things about the products and services that they're buying, whether it be B2C or B2B. But there's going to be this new track that has to run which is capturing the agentic ecosystem, capturing agents, capturing physical AI with the capability to transact as market share and figuring out what's the easiest best way to do that. Um, so I think that's the most immediate term for marketers is still figuring out how to run those dual tracks, uh, successfully. And I think which is like doing

Speaker A: two jobs at the same time. Yeah, it went from being like, I'm marketing to humans and this is the track. And now it's like, now you need to figure out both. And what I hear, at least from many marketing leaders is like that other piece of like, marketing to robots is very unclear. Everyone has lots of thoughts, but I actually haven't heard a definitive like, this is the best way. Now I've had companies who pitch that they, you know, they'll make sure you're found in the LLMs and here's how to do it. And then someone else comes in and they say something opposite and put a. TXT file on there and make one website for robots and one for humans. And then someone else comes on. It's like, no, don't do that. That's not going to work. And so I actually am like, okay, so how, like what's the best path while we're in this awkward middle school phase of trying to figure our life out right now?

Speaker B: Yeah, yeah, I love that descriptor. And it is, I think everybody's like throwing a lot of stuff at the wall and seeing what will stick. Ah. And with all of that be discoverable, I mean that's taking our uh, SEO playbook and going, oh, let's try and make this into something that will matter, um, in this new way of operating. But all of that is tactical foundation level, table stake stuff have become discoverable and become trustworthy in terms of the transactions. Operationally trustworthy. I was chatting about this yesterday, uh, with uh, a, uh, colleague and figuring out how to get the conversation up to a better level I think is where marketers need to set their sights. Marketing to robots, hot take, unpopular take is not a thing. I actually don't think it's a thing because what you're trying to do there, it's like trying to get the sentiment of a, ah, calculator to come up with the outcome that you want. I don't think that marketing to robots is a thing because it's like trying to market to a calculator, a very sophisticated calculator that can do lots and lots of things, but it still operates in data, it still operates in binary ones and zeros. It doesn't have anything that we use our original marketing suite of tools to interact with the emotional narrative, the storytelling. So that's, that's, I think the first thing, the second part for that first

Speaker A: thing, before you go to the second one. When I think about people talking about marketing to robots, what I hear so far is, and maybe this isn't even the term marketing, but it's just setting up files and language without emotion in it in a way that they can read of. Like just putting data there.

Speaker B: Yeah, super basic. And this is a hygiene factor. This is not, I'm not saying that that's not a thing that you don't need to do like you absolutely need to do that. You must be discoverable. Your data needs to be clean, the product feed needs to be there. You have to have the APIs for them to hook into in order for them to know what your inventory is and be able to order from your inventory. So creating those receptors is crucial. Absolutely. Um, but I feel like this is just the table stakes of being able to operate in the agentic economy less than it's marketing to robots. On the first, uh, part of this that we've got to make friends with, we need to figure out because the reasons why people don't want to participate in this have a lot to do with trust. And I think for marketers now, trust is the product that we're marketing here in this moment in time, um, that you can trust that your agent is going to do what you ask it to do. You can trust the merchant on the other side of that transaction to deliver, uh, for the agent in the way that the agent needs to be given a customer experience, that merchants need to be able to trust that your agent is not a rogue and it's going to be able to pay its bill. And, you know, whether that's a B2C, I'm buying myself some new moisturizer, or it's a company that needs, you know, 50 servers, that it's put out a request for proposal to get the best prices for its 50 servers. There has to be the trust from the merchant or vendor side back to the AI or the machine customer that's operating. And so trust is, I think trust is the product that we're trying to market. Right. Now is a great example of this. They went out, they put their developer kit out into market and they made a consumer promise that if you're a registered agent, so an agent that is registered in their ecosystem, that they verify the identity of and who owns it and all of that kind of stuff. If your registered agent makes an erroneous purchase, not an unauthorized purchase with which Visa and MasterCard absolutely already cover, um, but an erroneous purchase, it does something dumb. Amex says it'll cover your costs. That's trust as a product.

Speaker A: Yeah. So registered though, I'm like, I've made agents, I don't know what it means to be registered. So is this just for like enterprise companies who know how to go through that hoopla to make it registered?

Speaker B: There's a lot that's happening in this KYA space right now. There isn't a definitive answer to exactly how that works. I'm sure that Amex has a particular way that they want you to register an agent with them. Perhaps it's something like Google's Spark agent because it is created by a large corporate that has the very strong technical underpinnings. The agent will have a registered ID that it can go out and traverse the ecosystem. Although to be fair, Spark's going to use Google pay to pay for stuff, not Amex. Unless your Amex is in your Google pay wallet. See it's very fragmented. But yeah, this, this uh, know your agent is something that's, that's really happening a lot in financial services and amongst the payment Rails providers. An interesting one that I saw in April is the bank of Bots, which is a fintech that came out of stealth mode, um, doing the world's first loan to an AI agent. So their customer segment is agents, robots and drones. They're basically, they're a bank with financial products aimed at robots, AI agents and

Speaker A: drones, um, which is aimed at the people behind them. Right, that are operating them. So who are they going after? Who are the people behind those folks?

Speaker B: Robots. And that's taking the KYA in financial services we do kyc then you know, we're all talking about kya. So giving agents um, immutably provably consistent identities that we can say this agent belongs to this business or this person and has these governing constraints around it. But there's then the next step behind that which is no, you're human. Who is the human counterparty in the liability chain there who's ultimately going to be accountable to pay the bill like the agent. And I think that's partly what bank of Bots is trying to solve. That the agent has money that it is authorized within whatever constraints it's allowed to operate in, to use for the purposes that it has been instructed to act on. And so they have done a working capital loan to an AI agent that applied for the loan. A human had to do KYC because we don't have KYA ready in market yet to really do the same level of anti money laundering, et cetera, that a KYC would do. Um, but other than that, the agent determined its repayment schedules, how it was going to, how it's going to use the money. Um, and so this is a, a very nascent product, uh, suite for AI agents. And I guess what we've got to look at from a marketing perspective, um, I take it back to some fundamental behavior design. BJ Fogg has this great sort of, um, I guess equation for getting people to do a behavior. So the behavior equals the motivation to do the behavior, um, the ability to do the behavior and the trigger being put in front of you at the moment in time where you want somebody to do the behavior. Now with this, we have to look at what are the things that we can control from a perspective of an AI agent or a machine customer. We cannot control their motivation because they're not motivated. Like it's not a thing. They don't become motivated. The instruction behind them is their motivation. So going, all right, well if that's the motivation, um, what can we infer, understand, interrogate about the instructions that they've been given as marketing people to understand our customer better? Ah, let's again take it and distill it down to marketing essentials. What we do with humans is we deeply understand who the customer is and what they're all about, what motivates them, what's their worldview, all of those kinds of things. We have to take that same set of skills and apply it to this new customer. Not expecting to find the same things as we would find with a human customer. That's not going to work. But if you look for motivation, but in this instance you describe it as these are the instructions and um, the governing constraints around this agent that I can understand. It will operate within disintermediated motivation. Okay, so that's as much as I can know there. And then knowing that can help me serve it better. Um, the ability is where we really have strength and power. Because making something easy for a human is, it's table stakes and it's User Experience 101. You know, creating the conditions in marketing for people to easily take up your trigger offer, uh, your trigger that you're putting at the moment in time, that's where we get some, the traction with humans. It's the same, it's the same with an AI agent. It's the same with an embodied AI. If you can make it easy to do the behavior that you are intending as the outcome from the marketing exercise, then you're already like, you know, 80% of the way there. And with that it looks different when you're dealing with an AI agent or an embodied AI, a car that wants to buy something than a human. So you have to create the connectors that it needs, um, you have to create the clean data that it wants to interrogate. So this is where I think that discoverability does come in as part of that uh, marketing layer. But it's bringing the mindset of the marketer to that which is all right, we're giving it the ability to do something so we can get it to do the behavior. It's not a technical problem to solve in its purest form, it's kind of an everybody problem to solve because you're going to need the people for whom this is a specialist skill to make that all work. Um, so getting them to do that behavior through friction free ability is really, uh, crucial to the whole conversation. And then the third thing is again, it's a marketing conversation is that trigger, where do we put the trigger in the journey that this agent, this machine, um, customer is having with our organization in order for it to, in the motivation part of it. So in that, in that third part of the equation, the trigger part of the equation, this is a marketing conversation, this is a marketing, um, expertise moment. Where do we put that trigger in the journey that this AI agent is having with our business in order for it to bring into play all of those motivations that we've come to understand the parameters that it's operating within, the constraints, the things that it's been told as the outcome it has to get and create that easy ability for it to do the behavior. So like your technology, people are not going to think about it in these terms. They're going to think about in terms of the pipes, the utility, the data. Is it all there? Okay, great. And this is an everybody problem to solve and an everybody opportunity. But it's the marketing mindset that can bring all of that really strategic thinking to this to go. This is the outcome and the behavior that we want to get. This is our understanding of what's motivating the behavior from the human or the business behind the agent and how it's manifesting in the agent, in its parameters, in its behavior and its operating. This is our ability to, this is how we can create the ability for it to do the thing without friction so that we're the chosen party that they want to work with because we make it easy for them. And then this is the way we're going to trigger it to come in and execute on um, the instructions that it's been given with our products and our services. So this is, this is where the marketing conversation needs to be. Not down in the weeds of like, you know, how do we make ourselves discoverable? Because the more discoverable you are, the more interchangeable you are.

Speaker A: Every customer tells you who they are and what they browse, what they buy, when they come back and when they don't. Attentive helps brands turn those signals into personalized messages across sms, email, RCS and push. So every interaction feels relevant, not random, not more marketing, better marketing. That's marketing made personal, that's attentive. To learn more, visit attentive. Com. Now I'm thinking about how you just explained it is making me think that, okay, from a consumer perspective because if I think about AI agents, let's say from like finance, high frequency trading, of course they're just going to go out, they know the rules, they're going to be making the trades and doing all the things that makes total sense to me that like you need space, speed and you know, they can act on your behalf and that's the goal. But when I think about consumers right now, I don't think there's definitely obviously not enough trust now to do that. And how it's being used right now is go out and do all the research and then bring it back to me and I'm going to take some time to make a decision that maybe back in the day I would have just gone on Google, only had two links, went with the more the one with the better story, the brand that I know, the one that has a bigger emotional tie to it. But now it feels like because we have so much information coming, there might be a bigger pause that marketers aren't used to because consumers aren't going to say, hey, go and buy that thing for me. I don't think as of now they might, they probably have need some time to develop trust to say like, I want you to start executing on my behalf but instead bring me a lot of options that are the best ones Based off what I told you. And now I need time to wait and think about it and then how you get chosen in that space. That's the part I'm curious about. Like, because it feels like the brands who used to show up there, let's say Nike, Patagonia, like they built their brand based off emotions, the human element. I mean, they spent so much time there. And so I might think that still matters, um, because people look at it and they're like, well, I still, as of now in this generation, have a lot of trust for, for whatever it might be, Nike, um, because it's so top of mind. So, like, how do you think about maybe a pause? And I will also say I've saw this with enterprise companies too, with them evaluating SaaS and looking and being like, now I have a lot of good options. And there was like the great pause of buying into new enterprise SaaS companies, which is probably for the best at this point since many of them are going to be obsolete. But yeah, so this pause. I'm just curious about how you think about that.

Speaker B: Yeah, I think you're kind of describing the moment of decision paralysis where you have so many options. Amazon has, uh, tackled this with a feature in Rufus called Help me Decide, where you just give it back to the AI and say, uh, this is too much. Please just help me decide. But the thing with Rufus is you think that maybe it's working on your behalf, but Rufus is motivated, uh, to deliver brands to your front door that have paid to be surfaced. Um, and so you're not actually getting that neutral viewpoint on that one. There is a moment there where people are going to go, oh, wow, that's a lot. Okay, I don't really know what to do or how to, I need to process. Because we don't process at the speed of AI. We just don't. Um, our brains, they work differently and they parse data in different ways. And the amount of data that we can parse all together at one time is different. So I did some experimentation, particularly around brand voice coming up in that co buyer, um, scenario. And I did an experiment, uh, going, ah, I'm looking for some running shoes. I don't run very often. I'm a woman. What have you got for me? And it came back with, you know, three different options with three different brands and just described the features of the brands. And I asked it, you know, did you actually represent the brand voice of these brands when you presented this information to me? Uh, and I was working with Claude at the time on this one it was like, well, actually, no, I didn't, um, and I was like, well, can you do it again? But actually represent the brand voice properly. And it went very much from like, uh, you know, EV foam soles to marshmallowy soft soles that will propel you forward into the whole world. And so it is capable of representing brands to differentiate. And I think the theory that I have about this to take it more above that foundational discoverability level and even above that operational trust level, um, is that we actually have to be having a stratospheric conversation about the values that our organizations stand for because that is a differentiator in the marketplace, particularly as, um, you know, younger generations, uh, and even older generations are uh, buying in alignment with their values and statements about their values are becoming part of the descriptor that they're giving to their AI agents to go and find them options. And an AI agent has got all of the patience and time in the world to interrogate your supply chain about its sustainability credentials or its ethical sourcing credentials than a human does. So we're going to need to start encoding our values as organizations in machine readable data. Patagonia is a really good example of this because they have their footprint chronicles which showcase how they participate in the circular economy, the amount of reuse and repair that they do, um, the level of recyclable materials that are content of their products and they showcase all of this in data. And so if I say I want a, you know, a sustainable mountain climbing jacket, um, Patagonia is going to show up there as the option for me. Um, and so these sorts of methods of filtering out the noise in those conversations are going to be more and more apparent for marketers to start taking advantage of. And again, it's not just that. Make your data machine readable, put a. TXT file at the front of your, you know, your website to explain to the agents what you do. Honestly, most agents, if you're a big brand, they already know about you from their training data. Like it's already in their training data who Nike is, who, you know, um, who Patagonia is. Uh, and so they're going to be taking it out of their training data, uh, before they start hitting your website in order to determine whether or not you're a fit for what they do. Again, looking at it from a behavior design perspective. All right, well, people are, they're motivated to find the thing that matches up with what they want and have inferred that motivation to, um, their agent have um, you made it easy for the agent to make you the choice there and are you getting the right trigger, the right information in front of them at the right time? So how well or poorly you have described your brand out into the marketplace with the differentiators that are really clear in machine readable data backed format, like the data backing of this is really crucial. Um, especially third party data. Um, the Internet decides what's good and what's bad and agents take a lot of steer from places across the Internet that surface those sorts of opinions. So those are some of the things that can be brought into play into that, that pause moment. The other thing that is worth noting here is this is a country by country thing in terms of how much people trust AI to buy on their behalf. Worldpay did a really great survey just at the end of 2025, ah, with about 8,000 people across all sorts of different countries. Um, and in China, 65% of people said I'm already shopping with an AI agent or I would be happy to shop with an AI agent. So they are the most progressive market in terms of trusting this and wanting to use this as a way to participate in the economy. Um, whereas if you look at the very lowest end of the scale, uh, France was at 53% of the French people who responded said I would never shop with an AI agent ever. The highest resistant market to this. So there isn't a stereotypical approach that you can say carte blanche across the whole thing. One size just does not fit all in this scenario. And different countries, different markets, different cultures are going to come to this at ah, different speeds with different levels of trust and resistance. And this is again another marketing conversation for your particular territory and understanding what the culture is that the AI agent is operating in, which will determine the types of instructions that it's being given that comes into play. And your ability as a marketer to bring that layer and lens to how this works, um, is also really, really crucial. So I think that marketers, if they're spending all their time trying to figure out how to make their data machine readable, they're operating at the wrong level and they're using the wrong talents, um, to do the wrong thing. I guess that's where I'm going with this one.

Speaker A: That's a clip right there. That's good. Yeah. I mean what I like about this conversation and the one thing that's really hitting for me especially I'll just say in a US market is so many companies have in the past, many have cared about showing their values and showing the things that they care about that maybe are counterintuitive to making a profit. And many of those companies have kind of gone unnoticed and they just did it because they wanted to, which is amazing. But now what's cool is I see consumers and maybe I'm in a bubble in Austin, but even in the Bay Area and certain areas, like consumers actually are starting to care about what are the materials in this and like health consequences of things. And I think it's because we have so much information now. It is like, hey, where is uh, things like diamonds coming from? Like that might not feel good anymore now that I've seen where it comes from. I can't unsee, um, it. And like, so there's a whole new revolution of data and information that I think is making a more conscious customer rising up. And so thinking about that, like, if that's like one takeaway from this interview is like being able to instill your values and really get clear on what are your values, what do you want to stand for as a company and how do you want to make it known? Because now it actually matters in a way that you're going to be found because of it, which is a pretty cool reality to be in now.

Speaker B: And I also, I want to expand that one, that it's not just a B2C conversation, this is a B2B conversation as well. So Walmart has had AI procurement operating since 2022, starting in pilot and now operationalizing where they negotiate with their more than 2,000 vendors using an AI platform called Pactum. They're getting great results from that. Um, for Walmart I think, maybe not so much for the vendors, but definitely Walmart is getting great results from that. But Walmart's procurement runs on a responsible sourcing set of governing guardrails. The things that that organization states as its values in terms of who it will source from and what the parameters are around that in terms of sustainability, in terms of modern slavery, in terms of all of the ESG credentials they're going to be exposed to expecting their vendors to be able to articulate. Now when it's the AI that is interrogating your supply chain in this way, you better be ready for it. You better be ready with everything that you need to signal as something uh, that is machine readable, machine understandable, but not only that verifiable in third party credentialing databases. Like if you say that you're organic, you need to be proving that through the registration with a third party Organic supplier database, uh, in order for you to be able to go, okay, it's not just what we say. I can show you that we're living our values. And here is also the receipts for where our values have showed up. So this is, uh, uh, an important B2B story as well. So I want people to be coming to this with both lenses in play.

Speaker A: So now that you're talking about third party, this is relevant because I was just on an interview, um, with an amazing human, and we were talking about brands having their own media platforms. And the question that came up for me was, yes, Marc Andreessen has said every brand, every company needs to be a media company. He's been saying that for a long time. And so it's always in the back of my mind. But then when I hear this, and this is something I kind of pushed on a little bit, I'm like, but if LLMs are favoring third party content, should you instead be focused on having others talk about you instead of. Even if you have a media platform that's arm's length and people know it's from you, like, eventually the LLMs will come in and be like, well, these two are connected. So, like, I'm gonna go and like, look at what's on Reddit instead. And I'm gonna go and look at this podcast where someone talked about they were actually a good, you know, platform they were using. So how do you think about this of like, using third parties to like, verify that you're good versus, like, having your own owned channels to tell your story?

Speaker B: Yeah, I guess it depends on who you're trying to talk to. Um, and we still have to talk to humans. We definitely still have to talk to humans. And humans are, you know, they like brand narratives and they get emotional about brand narratives. And I was exploring a paper that came out just, I think just this week about how ChatGPT, uh, refers people to E commerce websites. This is, it's a very dense academic paper.

Speaker A: And the tldr, because I definitely didn't read it.

Speaker B: Yeah, yeah, yeah. So the TLDR on this one is, um, that at, uh, the moment, it is a pretty small subset of referrals because it was comparing search to ChatGPT, um, specifically just ChatGPT, saying that, you know, the referrals are still pretty small and search actually still outweighed, um, chatgpt in terms of E commerce referrals. The other aspect of that, um, in some of the other literature that I've been reading is the ability for the LLMs to infer the emotions of people. Because I mean with the narrative of logic replaces emotion. Um, in this I think is true if we have a pure just AI agent doing something. But that AI agent just looking at what I saw in BYD about the emotional inference that the agents are able to do as the intelligent cockpit of a car, agents are going to be able to infer from humans the emotions that we're bringing to an instruction or conversation and use that as part of the buying parameters. I just saw some stuff come out of Thinking Machine Labs, which is a group of people who left OpenAI to go and do some more research based work in AI. Their AI is able to use cameras, um, to sort of infer more about the emotional state of the human or the physical state of the human. Like they've got some demos around helping somebody not slouch at their computer and also inferring where people's emotional states are as the machine gets smarter, um, it's going to be bringing all of that emotional baggage with it as well. So I don't think, and I'm really sorry because this just makes it more complicated. I don't think we can take a pure view in marketing and go all, uh, right, well, logic replaces emotion. So now I'm going to do things like that. This is a really tough space for marketing and if people are feeling like confused and stressed, I'm here for you. I see this and it's right that you should feel this way because this is really, really tough.

Speaker A: So you were going towards like LLMs picking up on emotions. And so then the logic piece around it might not be the way to go. I mean all I'm hearing and what I keep hearing with this is like, I don't know if anyone knows where we're headed because I even look at just the past six months to a year and there's been many CMOs that I talked to who are like, yeah, I threw out my entire annual plan at the end of January. Like I started with one in January 1st and January 25th it was dead. And I look at like what's been happening. I'm like, no one last year was predicting that AI agents would come in, change commerce, kill off a, uh, ton of SaaS companies or continue to kill them off. Like no one predicted that. And even when people saw what was

Speaker B: happening with Claude, the AI agents shop. Yeah, I wrote a book.

Speaker A: Oh yeah, okay. That so, okay. I will say people on the show have said that for a long time. Yeah, I'd uh, say maybe Two years ago. I've heard people saying that, but I was like, okay, that didn't come true quick enough.

Speaker B: It was a gartner trend from 2017.

Speaker A: Yeah, yeah. So I have heard people saying that, but not one person said, hey, Claude's going to come in here and like rip apart SaaS companies and they're going to die and then we're just going to make what we want easily. Like, no one, at least from people I talked to, I didn't hear anyone predicting that. So then I'm like, what? I don't even know if we know where we're going really. And like what to even prepare for fully.

Speaker B: I, uh, now I know, I remember what you're asking about. So on that, on that sort of, you know, brands being their own media companies and storytelling and you know, language models, large language models being biased towards what third parties have got to say about brands. I think the connection that I'm trying to make here is that because of this ability that's of kind coming through with some of the research level work. So, uh, the thinking machine labs work around inferring how a person's feeling and their physical state as well. The stuff we saw in intelligent cockpits, um, with emotional inference and then adjusting your environment to suit the emotional mood. I think it's impossible to see a situation where we have AI agents that are able to infer the emotional state of their human and then perhaps pull on the storytelling that a brand is doing in its own media channels as part of its consideration set to match against the, the, the instructions that it's been given. So the. But I, this is just me like predicting something that I can see as a possible future. I have no experimentation on this. And I just don't think we're in a place with, with AI where we have it in, um, a in a good enough state to, to showcase whether or not this would or wouldn't work to definitively say to a brand, hey, yeah, keep spending your money becoming your own media company. I can only say that I can see a possible state where it could be useful, which I'm sure is not the definitive thing that people want to hear right now. Everybody wants to know what the hell to do.

Speaker A: Yeah, yeah. Which I think anyone who, I don't know when, when I look at a lot of the actions being taken, a lot of people are moving really quick and shipping things really quick, which is amazing. Um, and what comes up too is like, man, what if people are moving too quickly in a direction that's not where things are headed and like, you know, we talked about with just going in and adding certain. Txt files on the website and changing the whole structure of your website and like moving so quick where then you look up and you're like, oh wait, that's actually not the direction that things are heading anymore. And I just, you know, put my budget and my, I campaigned for this to happen and now it's already gone. And so, yeah, it feels like we're in a space of like, also maybe pause for a second and like using your own intuition to also be like, okay, me as a consumer, if I'm on ChatGPT or Claude or Gemini, like, how am I seeing things play out? Because that's usually probably the best place to discover what's happening with the market. It's just like getting in there and like seeing how things are going hands on.

Speaker B: So I would say that our marketers need to start behaving like scientists, um, here and running experiments, um, so we can bring in a lot of things that we learned from doing lean user experience design for this, which is form your hypothesis, be really clear about what you believe to be true. Make the statement, we believe that if we keep running a media company that AI agents will be able to use emotional inference to add that to the instructions that be given by the human or by the business. Uh, and our brand storytelling will still land and we will see measurable results. So be clear on the positions that you're taking and then go, all right, what experiments can we design to prove out whether this is true or not true? Uh, and this is what's really quite powerful, um, in terms of the way that you can think about this. Then run the experiment and measure the results and see whether you're right or wrong. Run lots of experiments, small experiments, but be very intentional about the experiments you're running. Don't just throw spaghetti at the wall and hope that something's going to stick. Have a hypothesis what you believe to be true. Make the statement, write it down, figure out the experiment, run the experiment, look at the results, see whether or not you are right or wrong. Then either keep doing it or scale the experiment if you are right and measure again, or go, oh, yikes, that wasn't, um, what we wanted to do. Let's toss that, start on something else. Now I know that a lot of organizations, they've got all these AI efficiency gains they're apparently getting, which is allowing them to let go of all this headcount. Rather than do that, perhaps keep the people that you Thought you might want to get efficiency gains, um, by having them head out of your organization and redeploy those people into running these experiments, doing innovation. Let's take an abundance mindset to this rather than an efficiency mindset. If we keep on taking and stripping people out to do exactly the same things we've already done, just faster and cheaper, then we will perpetuate the status quo, which is ultimately a slow slide to mediocrity and irrelevance.

Speaker A: If. Hell yeah.

Speaker B: If we take these amazing people we have in our organizations and go, hey, the AI is going to do that thing that was administrative and annoying that you used to do. Let's take your brain power, and we want you to start coming up with hypotheses about things that we could be doing, running experiments showing what's working and not working, innovating. Then we're probably going to get to something that is net new. And when we have things that are net new, we have net new revenue, which is a better outcome for a business because we're selling new things to different people or the same things to new people. And that's how we grow our businesses. Otherwise we're just going to become anorexic businesses that can't support ourselves. So this is absolutely my plea is redeploy these humans into your innovation space to run these experiments and start figuring out for your business what's going to work, what's not going to work, and then please, you know, share what you find so that the rest of us who are scrabbling around in the dark, can. Can. Can help. Can be helped.

Speaker A: Yep, yep. I love that. It also makes me think it'd probably be very interesting to just have Claude quickly look up the companies who have done that. I mean, there was huge headlines for a while there, like so and so lays off 70% of this and that. And you're like, whoa. It'd be interesting to go and look and be like, what happened? Because oftentimes we just see those headlines and then no one ever looks back and is like, whoa, what happened? Like, now that company is like, hiring again and they're, uh, missing a lot of people, you know, so it, it would be interesting just to do a quick look back on that.

Speaker B: Yeah, I think they've already walked back. The AI is taking your jobs narrative. I think even Sam Altman and, and the people at the top of all of this have walked that back and said, actually, it's just a correction for some hiring that we did some overhiring, that we did. But they've got a very convenient scapegoat there, which is point at the AI and say, oh, AI is doing all the jobs now. And I think that if you go inside, scratch underneath the surface of any of those organizations, you will find they probably don't have AI capable of doing even one human's job. I'm sure that there is AI that can do some jobs, um, at the level that it's being presented. No, I don't believe it.

Speaker A: Yeah, there you go. I mean, I think what is beautiful about the time period that we're in right now is that humans get to actually take a step back and be like, what do I like doing? And like, what am I best at and what makes me human? Because, yeah, a lot of jobs, a lot of things can now be done. I mean, I'm watching it for myself where I'm like, whoa, I used to have many people doing this one thing and for me it used to take this much time and I don't have to anymore. So that's amazing. So, like, now what do I get to do that's uniquely me and that's like a fun place to be in. So it's a resurgence of humans and doing human things and figuring out what that actually is. Because for a while we were just. Many people were put in roles that you had to be in that like, you had to choose certain jobs and doing the same thing over and over again. And that was just how things were. And so now you really don't have to, and you get to choose in a very different way, which is pretty cool.

Speaker B: Yeah. And this is a big mindset change that I think has to happen. Um, which is. Okay, well, now that I've offloaded all of that administrative task that took me three hours and it now gets done in less than three minutes, what am I going to do with my three hours? And being super intentional about repurposing our time to try and do something net new, to explore, to experiment, to find something that's meaningful, useful, and interesting to do with those three hours that have been gifted back to us. I mean, what a gift. Time is such a gift. So if we're not intentionally using it, I think we're missing out on, uh, such an incredibly valuable opportunity. The thing that that's interesting around that one as well is that while that sounds like a hand wavy future and it's coming in Chinese cars, and maybe that doesn't necessarily land as much for a U.S. audience. Um, like Mercedes has got Mercedes pay already in the Car. It's had it since 2023. You can like the car can pay for its own parking, the car can pay for its own charging, it can buy third party apps off its platform. So like we already have, in the words of the CEO of Mercedes, um, pay, you know, gentic commerce started on the Internet and it's coming to the car. And that's his main intention, is to bring agentic commerce to the car. So we have got in market today vehicles that are able to buy. Mercedes doesn't have an intelligent cockpit, I don't think yet. To be fair, I haven't research that explicitly because I was just looking at the BMW China and the other China examples. They definitely have the intelligent cockpit coming in. Um, so when you couple that idea of agentic commerce in the car with the emotional inference, you kind of get something that if you wire it together, it has the ability to buy based on people's emotional states depending on the governance and constraints that are around it. I don't know.

Speaker A: And the consumer trusting to allow that.

Speaker B: I tested some of it out like I tested putting these out, um, on LinkedIn and wrote a substack article about it. Um, and one of the comments that I got back was yeah, and it could tell that like my daughter got into the car when I picked her up after school and she had a total meltdown and drove me crazy all the way home. And then by the time I get home it will have bought my favorite chocolate bar and it'll be waiting for me and I can eat it when the kids go to bed. Ah. And I was like, whoa, okay, that escalated quickly. You actually took this and made uh, an experience out of it. Um, so I, I see that some people immediately see the benefit and how this could, could help them.

Speaker A: Katya, thank you for joining Marketing Trends. This was a super enlightening conversation. I feel like it sparked so, so many new ideas and thoughts. So thank you for coming on and sharing everything. Where can our listeners and our viewers find out more about you and your work?

Speaker B: Oh, thank you. It was a really terrific conversation. Made me think of some new things. So thanks for asking those questions. Um, I am very easy to find, so I work out loud on this. I work out loud on LinkedIn, I'm easy to find there as well as on Substack, uh, where I write as the CX Evolutionist. My book obviously is available for people if they feel like they could use a field guide to start navigating this. Um, that one's available on Amazon, um, in every market. Around the world, so I'm happy to have conversations, please, if you're interested in talking about this. I'm always interested in talking about this. So reach out and let's chat about it.

Speaker A: Amazing. Thank you so much.