[00:00:00] Phillip: Hello, welcome to Future Commerce Podcast at the intersection of culture and commerce. I'm Phillip.

[00:00:04] Brian: I'm Brian.

[00:00:05] Phillip: Brian, you got the glow about you today.

[00:00:07] Brian: I know. I've got the the travel light. That's what I've got. The travel mic, travel light.

[00:00:12] Phillip: I didn't know that you'd, like, travel with a ring light or...

[00:00:17] Brian: Travel with a ring light. This is the light you gave me.

[00:00:20] Phillip: You don't seem like the ring light travel type.

[00:00:23] Brian: No. I'm not. I am not. I am the opposite of that. We were...

[00:00:27] Phillip: Shop Talk, and there was the whatnot streamer who, like, that's their whole job.

[00:00:34] Brian: They just showed up ring light.

[00:00:36] Phillip: But they showed up to Shop Talk without the ring light, and I'm like, you have one, like, literally one job. That's your whole job. Your whole job is that. Well, have the earrings to sell. But...

[00:00:51] Brian: I hear that. I... I forgot my, like, AirPods on this trip, and I just... I Amazon-ed one day a pair of headphones. So I'm just like, I need headphones.

[00:01:01] Phillip: The... what an innovation. We have lots of innovation. We're gonna talk about innovation today. Before we get there, I just wanna remind everybody, we have our newest book. It does talk about how innovations in the world in, you know, media, technology, commerce, retail, and the way that the world is being shaped with social commerce. It's all wrapped up into 80 pages. It's our newest book. It's our zine. It's called Strata. You can go get it right now. Futurecommerce.com/strata. You ever wondered why aesthetics rule the world around us? Why the things that cause us to buy, or the things that's like social peacocking, the social commerce of it all? It's all wrapped up into 80 pages, our newest book. Go get it right now. Futurecommerce.com/strata. I would love to get it into your hands. There's only 500 of them available. And if you grabbed it at POSSIBLE, it means it's possible that we have fewer available these days, so go get it. Strata. And people are going gaga over that. Would love to hear more from you guys when you get it into your hands. I can't wait for this next conversation, Brian. I've been looking forward to this one for a long time. Sounds like it's been years in the making. Our next guest is the founder of M7, and M7 specializes in emerging media, AI, AR, VR, evolving internet, and here to talk with us about all of it today. He is an independent researcher and technology adviser. He's helped Fortune 500 companies gain that first mover advantage, and he's the thirty fifth official member of the MIT Media Lab Consortium. I think that's the first time we can ever say that this ever happened here on on the show.

[00:02:35] Brian: Far from now.

[00:02:36] Phillip: And he has his hands on in actually doing technology rather than us who just, you know, we contemplate. We just are the pundits these days. I just stopped doing technology. And I can't wait to get into talking about our current moment. So Matt Maher, welcome to Future Commerce.

[00:02:53] Matt Maher: Thank you so much, Phillip and Brian. So excited to chop it up with you today. This is going to be a great episode. I can't wait. I can't wait.

[00:02:59] Phillip: Yeah. And we do have a lot to get to today. But, you know, we are... I think you and I go back a few years now, but what have you been up to and what are you working on at M7? And what is this current moment that we are in right now? I think there's a lot to talk about. You're talking a lot about how AI is going from one to a 100 rather than zero to one. Tee us up for our conversation.

[00:03:20] Matt Maher: Yeah, I mean, when I founded M7 in 2018, I did start a lot of my keynotes then with this thesis that like technology and media moving at breakneck speeds. I look back and I'm like, what an idiot. Like how slow were we moving back then? Well, I'd say relatively quick in 2018. And now obviously we're in 2026 and we've entered the AI era. So yeah, I like to say, we don't like to just talk the talk. We like to walk the walk. And I really appreciate that introduction that, yeah, I roll my sleeves up with the M7 team and we're kind of in the trenches with a lot of our clients building, really just trying to not only see into the future and where things are gonna be but kind of build towards that future and frankly start to carve our own paths respectively. Because I think that's something we'll touch on as we get through this, that it is truly the wild west. There is no set path on which a company or retailer needs to go. It really is difficult in how to navigate. And that's why, you know, at M7, every client we work with, that relationship is very, very unique because I started this company with the idea of like, I just didn't want to be another solution desperately looking for problems. Wanted to really understand the business. And with AI, we're seeing so many opportunities to solve problems, but also a lot of noise and also a lot of new startups and tons of solutions desperately looking for problems. So we're just there in the trenches trying to navigate with our clients.

[00:04:57] Phillip: I keep hearing that enterprises are behind, right? They buy solutions, they don't use the solutions. But I have to say, I'm sure that's not the enterprises you're working with. Are they moving as fast as the technology is evolving? Is that happening?

[00:05:11] Matt Maher: Yeah, I mean, everyone, we're all giving it our best shot, Phil. We're giving our best shot. I think the tricky thing is, again, you mentioned it, I'll kind of maybe double click on it for a second. I equate, yes, the internet, right? That was zero to one. We did not have digital, information superhighway. We got it. And that was going from zero to one and it was slower, but businesses started a posture of, okay, I guess I'll need a website, retailers. Okay, guess I can sell online. And now we are in that, what I say like one to 100 moment. And you can look at it from a macro standpoint and then I'll give a more practical use case with the client. But macro use case, take like, you know, Claude Code, Claude Code comes out, trillion dollars gets wiped off the market, software is dead, it's not. But it's this idea of like, my god, are you kidding me? Now we have millions of people that are at parity with the best developers in the world. Then, you know, Claude Code integrates with Canva. Boom, two seventy million users of Canva. Now we're going to have the design skills of the top PR and comms companies in the world. Then Higgsfield AI, that's another great AI startup where when you think of Avengers and Disney, you're talking hundreds of millions of dollars in visual effects. Now you got 22,000,000 users on Higgsfield AI that are at parity, can create essentially multimillion dollar visual effects adjacent visuals with this tool.

[00:06:27] Matt Maher: So that's what I get... all of a sudden like we snap our fingers and we're at parity with all these capabilities. And I can say with our clients, I'll pick one just anecdotally, Cox Automotive. Three out of every four car sales in America runs through a Cox property. And back in the day, five, ten, fifteen years ago, problem solving with innovation and technology was kind of like a one to one. We have a problem, okay let's layer on this technology and that can essentially help us. In the age of AI, it's like, well, we can pick out thirty, forty, 50 issues we've had that are somewhat interlinked with each other and then layer on, okay, let's do Copilot for the entire organization to raise the baseline of intelligence. Let's use Claude APIs in the backend systems to be way smarter in how they can talk to each other. And then all of a sudden you kind of extrapolate that out. You're at a dealership and it's like the car salesman knows what to say at the right time, when to say it, who to say it to and how to push that sale along. So you can kind of solve a lot of problems at the same time with some of these tools. And again, that's only one use case but we are in the one to 100 era right now of problem solving in AI.

[00:07:34] Phillip: It reminds me, Brian. We had a study we did two, almost three years ago, coming up on three years ago about early adoption in enterprises in AI. One of the top line questions, sort of a qualifier in the study, Matt, was this: you know, do you have an internal policy around usage of AI in your enterprise? And for the people that said no, it was... or for the people that said yes, it was, is it, you know, permissive or restrictive? And for the people that said it was restrictive, it was, you know, do they allow it at all? And for people that said no, it was, do you still use it? And for the people that said yes, it was, how many of those still are using it and they're not allowed? 93% of people were still using it on their own dime and they were using it, you know, off the books. And just two and a half... and a lot has changed since then. And to that end, I feel like that that was the beginning of the reskilling. Now you're talking about specialized tools by team. Now you're talking about Claude building specialized skills, right? And building its own tools that are individualized not just by role, but by, I think, by, you know, hour to hour use case in granularity. So there's not even... I don't even know what to call it anymore. It's not really a playbook. Like, we're in a in a totally new era of building out a brand new org structure that is different to anything that we've seen before. How are... how are enterprises, how are businesses dealing with this brand new shape of an organization?

[00:09:06] Matt Maher: Well, first of all, I love that analogy. I think that was just like a really succinct and concise way to bring it to bear. I'd say with enterprise, I love playbook too, and let's stick on that for a second because we do try to build playbooks for our clients, but it also is really difficult. And when you think about it, it's like, okay, this is going to give us strategic guidance within the game that we're playing. That game has set rules. Therefore, the playbook can help us, you know, succeed within the frameworks of that game. But the reality is this whole world is moving so fast that the rules of the game are changing very... So therefore the playbook has to adapt pretty quickly. So the best metaphor I could use is probably like, let's say the NFL, you know, number one sport in US. So, you know, they change rules in the NFL, but they usually wait till the end of a season and then they enact those new rules and then you're able to go the next season. Your playbook could change based on that. AI gives us a world where the rules are changing, not even game by game, like almost play by play where you're about to hike it and you're like, well, that's not 10 yards anymore.

[00:10:09] Matt Maher: It's only two yards to the first now. And by the way, a touchdown's 12 points now and a field goal might be worth 20... in terms of how should we optimize for the best output. So we do try to build these playbooks but with just a lot of fluidity to say, okay, we know this to be true. Let's take retail as an example. Google announces the Universal Commerce Protocol, big deal, right? This is going to be a great way for people to shop. You're gonna have businesses be able to upload their Google Schema. It's gonna be a quick way to match inventory to agents. And you peel back the layer and there's nine competing agentic protocols right now. Everyone from OpenAI to Google to Microsoft to Coinbase. So if you're a company, do you place a bet on one of the nine protocols and just hope that that's the one that wins? Because you can bet your bottom dollar, every one of those big tech companies wants their protocol to be the de facto protocol. They want to be the HDMI cable that won out back in the day. We just don't know which one it's going to be.

[00:11:11] Matt Maher: So our playbook has this adaptability to say, okay, here are these different things. We plan more for what's directionally correct than what is actualized right now. So that is a lot... I can say with our clients what we're doing is saying, we can't bet the farm on one thing. We have to be able to spread our bets and then when we see a signal, when we see something that feels like this is going to be it, then we can double down or triple down on that and then sprint ahead. But a lot of it goes into that preparation, that business transformation, kind of having orgs rethink how they operate. And I think that's been a lot of cardinal sins with a lot of Fortune 1000s of, they have not good data hygiene, data's not structured, they needed a reorg anyway. And all of a sudden they think, well, what about AI? It's kind of like a big universal outlet. Let's just plug our business into and that maybe everything's going to be fixed. And I think that's where the biggest mistakes happen is when you just think the problems you have today will be a catchall fix from AI and that's generally not what's happening.

[00:12:11] Brian: Yeah. It reminds me, back in my very first professional job, there was someone in the org that would always say "automagically." And it just killed me every time. But I feel like that's what people are kind of like, like when they're like, just AI, like just AI it. Just throw AI at it. It's like, it'll automagically work, you know. Right. There's a lot of... there's a lot of stuff to do to get things to a point where you can sort of use it, whether that's organizational change or whether that's data cleanliness or whatever it is. Like, there's a lot of... lot of steps sometimes in the process before it's just gonna work. Now that said, a lot of things are working. And actually, sometimes, automagic is actually real because you step into these situations and you're like, this data is a mess. I don't know how to go... like, that's gonna take me a week to go through and, like, go get all that information that I'm looking for that I need to provide to that client or a deal or whatever it is. And now it actually feels like you can accelerate that to, like, what would have taken a week might take a day.

[00:13:21] Brian: To your point, one to a 100 rather than zero to one. But one of the things that I think a lot of people have been struggling with is when to use AI and like how. And like, you know, I... I think about like, you know, some of the ways people have been using it creatively. Seems like there's a lot of blowback towards that ends. I think it seems like the kind of tool where the more you lean into it for creativity, the more it leads you down a specific direction. Or, you know, in other scenarios, maybe the more you rely on it, the less you actually know about the actual thing you're doing. So you then start to use it in a way where you don't actually have the filter you need to know when it's going down the wrong path or whatever it is. And so you talk about this idea of like the productivity paradox of AI. And so what is that and why does that matter to enterprises?

[00:14:24] Matt Maher: Oh, definitely. So let's start, Brian... we won't even get to the productivity paradox. Let's talk about what you just hit on, which is our cognitive load starting to, I don't want to use the word decompose, but I'm going to use the word decompose because we have studies out of... out of MIT. I'm going to be on a train there right after we wrap here. They had this amazing study that they had three focus groups. One was an LLM, or large language model, group. One was a search engine group. They could use Google to look for information. And one was a brain only group. So they had no tools they can use. And I mean, I probably don't even need to say what actually happened, but you can probably guess that it was in descending order. Those who only used a large language model consistently underperformed in neural, linguistic, and behavioral levels. And the most fascinating part of that study is that they then flipped the group. The LLM group then got to move over to brain only and brain only got to move over to LLM. The brain only who moved over to LLM had some cognitive capital built up from using their actual brains. The LLM group who was then forced to just use their brain, they found performed even worse, right? Got even worse, they just didn't have the recall, linguistic, those tools to essentially pull from. So there's one issue you're very much describing, that if we are going to continue to delegate our cognitive loads and critical thinking to these tools, those are muscles that will absolutely atrophy.

[00:15:48] Brian: Absolutely. Now...

[00:15:48] Matt Maher: The second part, which I think is a fascinating piece of this and this is what I'm trying to call the productivity paradox. And it's that I always try to look at things through an anthropological lens. I also think that's why Future Commerce is so powerful, is that you look through culture and commerce and you kind of predict the future and where it's going to be. And I think, you know, human beings are part of that. And I try to take that same lens when I think about all these tools. It's like, oh, I'm so productive. Like, look what I built in Claude Code, this like really watered down version of Spotify that no one's ever going to use. It only took me two hours, but I built it. And I think people start to like glom onto that to say like, I feel so productive. So there were two studies that really opened my eyes. One was the METR study. This was late twenty twenty five. To Phillip's point, these things move so fast so these tools are better, but in this study they set out for coders to have them use AI tools and they really wanted to quantitatively track how much time was saved. So the hypothesis before was that completion time was going to reduce by 24%. That's what they thought. The developers themselves went through the study and then qualitatively were asked, how much time did you think it saved you? On average, they believed it saved 20%, so a little under what the hypothesis was. In actualization, the AI increased overall completion time by 20%. And they gave that information to all the developers. Seven in ten said, it doesn't matter, I'm still going to use the tool moving forward.

[00:17:17] Matt Maher: And I think that shows exactly how we are as humans. It objectively takes more time, but our dopamine receptors are feeling good when we're that productive. So we'll happily take 20% more time, claim that we didn't take as much time even though we're easily seeing that we're just spending more time prompting, reviewing AI output and literally waiting for the models to be done. So the second and last study, and I'll shut up after this because I think this one is the duality we have to live with, is this Faros study in March 2026. It found that there's 91% longer code review times using AI tools, 9% more bugs per developer that are using the tool, but 21% higher task completion rate and 98% more merged pull requests, which means collectively there absolutely is more output and more deliverables, but individually it's taking a lot more time. And the promise that all of us as humans are going to have all our time back for high value things is not happening just yet, but we still have that feeling, that automagical feeling of like, it feels so good to be productive even if I have to do so much work to be productive. So I think that's where there's going to be this divide of, how much time are you actually saving? Even if you're token maxing, even if you're doing all these things that big tech wants you to do, are you actually saving time?

[00:18:40] Phillip: That's the saving time, right? So the... where you start to become a little bit, uh, enamored with the AI is... there is no saving time. It's the the charter of the job starts to grow. And the time that we're reclaiming supposedly, I think we're actually starting to do other things with. Right? The job never gets completed. And from what I... this is anecdotal at best, you had a lot of data there, Matt. So again, like, here's the punditry. But we here at Future Commerce, we gather with them and break bread, round tables with them often. We spoke to people last week, Brian, at... mhmm, one of our, you know, salons, and spoke with more than one person last week alone who said that their productivity increase is leading to a lot more stress in their role. And it's leading to them feeling like they have more on their plate. They're having to balance more context switching. They're spending more time, you know, with more Claude Code windows, more Claude projects, more, right, more automation open. They do feel more powerful. They feel way more productive. The dopamine is at an all time high, but it does have this other sort of cognitive load now that didn't exist before. And when the productivity then drives, you know, folks... the other effects start to kick in as to, do we need to hire other resources, right? We're having all of these gains without the the OpEx impact. You start to wonder if that is going to be your life forever. "I've become important in my job and it's ruining my life." I feel like there's a lot of people who are really good with AI right now who feel that very intensely. I don't know if you have any, you know, if you have, you could commiserate on that, but that's what we're starting to see in our feedback in our executive network.

[00:20:43] Matt Maher: Yeah, I mean, I will say you're spot on there, Phillip. I think, again, this is somewhat unique, I'd say, to a US culture. We have clients... I think it's a little bit different in Europe. I mean, if you just look at how our world is set up, it's like it's set up in a world to maximize shareholder value, find more efficiencies. And I think what happens is, yeah, as you're on that treadmill and you're running speed eight and you're feeling pretty good. And it's like, well, yeah, you can get me to speed 12 and I'll get to speed 16, but you can't slow down now. And now you're going really, really fast and you got bionic legs and you're feeling good because you're like... but the reality is like, again, macro capitalism in America is then, you have CFOs looking to say, hey, you have X amount of people in this division. I think if we lower that by 10 or 20%, we're still going to get the same output. And all of a sudden you don't realize that output becomes on you. And that's exactly the meme you're kind of saying where it's like, there was the opportunity, there was this window of opportunity to say, okay, as this was starting to shift, what does higher value time look like? What does it mean for bigger thinking and to actually use that time and not instantly reallocate it to just more productivity? Because if we need the output to be two to three X, then of course all that time is just going to be filled into actually producing the thing. I think there's two second order effects that become really tricky. One, you see entry level jobs starting to just go away because it's like, why would you pay that when you could have an agent or someone essentially do it, which leads to... you get the next middle and higher management leadership if you're not training the young folks. So I think that's one thing that's just difficult to solve.

[00:22:20] Brian: To add on to that, like, even if you do hire a junior role, if they're using AI to do their job, they are not gonna develop the same instincts that they would have if they hadn't used it.

[00:22:32] Phillip: So short term anyway.

[00:22:33] Brian: That's what I'm... short term. Yeah. Exactly. Sorry. Keep going. I just wanna interject.

[00:22:36] Matt Maher: No. It's such a good point. And the second other major piece is that if everyone's then starting to use it, then you start to think of like where the net new idea is gonna come from. And that's what I always try to say to clients too. Like as much as I'm a proponent of AI and all the good, it is a reduction to the mean when you think of what large language models and even these creative platforms do. I always struggle to find metaphors. I always like to try to stick to sports metaphors, but I'll stick with basketball because now it's finals now. You know, if you asked like any AI, the smartest one in the world, GPT 5.5, like what is a great way to free up Steph Curry for a three? Two or three years ago, it could have come up with a couple ideas but like it literally could never have come up with the elevator screen, which is like a brand new screen or play that was invented a couple years ago for him specifically where two defenders come together, create an elevator door, he pops through and hits the three. AI could literally never have created that because it did not exist in the world before. So it would have given a bunch of really good ideas based on a hundred years of history, but it couldn't have come up with that.

[00:23:39] Matt Maher: That had to come from the intuition of a human being saying this is an anomaly of a player and we need to create something net new. So I think that's the larger second order effect for many companies too that are trying to find the next big thing or the next revenue stream from. And that's what I always try to say to clients too. Like as much as I'm a proponent of AI and all the good, it is a reduction to the mean when you think of what large language models and even these creative platforms do. Because these tools are only built on what exists in the knowledge base of our world today, where does the fringe live anymore? And then who's going to come up with those great ideas to your point, Brian, if everyone's starting to use them and then we're just kind of inundated with every single one of these tools and we start to actually lose those critical thinking skills, then I think that's actually dangerous for business in the future of like, we're not going to have great net new ideas because the famous phrase you'll hear with everything with AI is, it only took me five minutes and it's just as good or almost as good as what the other output is. What do you think about, well, we're in a world where you need to be better, and it by definition cannot be net new or better than something that existed. I think that's something that no one's really talking about now, but we'll feel that five to ten years down the road.

[00:24:40] Brian: Or does it? I think that sometimes, and this happens even before AI, like, are certain things an organization can do that aren't necessarily better. Like, they just need to get them done. Right? And this is where I think AI is actually having a real impact. It's actually... there's a lot of stuff that just has to get done, and people have actually been pretty under-resourced to get a lot of those things done. Great point. And that... what's what's happening is those things are the things that are getting done with AI, and now that actually is putting people in the position of having to be creative. So this is where I think that cognitive load comes in. It's the last 5% we're talking about, but that last 5% is actually a 100% because... and and you could see this even before AI. Maybe maybe for...

[00:25:34] Phillip: You, Brian. But yeah.

[00:25:36] Brian: Well, you could see this, like, with developers, you know, they'd be like, I'm 90% of the way there. This is before AI. I'm 90% of the way there. All I have left is this final 10%, and that 10% took just as long as the first 90%.

[00:25:50] Brian: Yeah. It's true. Yeah. I do think there's something... so there's this quote. I'm a... hat tip to Alastair Roberts. It's from G.K. Chesterton, who was a turn of the century author, writer, unbelievable thinker in in Britain, influenced Marshall McLuhan and a bunch of others. And I'm gonna read you a quote. It's a little bit... you have to kind of apply it, use your brain a little bit to kind of like contextualize it. But he says...

[00:26:19] Phillip: The 5%, Brian? Is that the... yeah.

[00:26:21] Brian: The 5%. Sorry. 100%. "The political instinct or desire is one of these things which they held in common. Falling in love is more poetical than dropping into poetry. The democratic intention is that government, helping to rule the tribe, is a thing like falling in love and not a thing like dropping into poetry. It is not something analogous to playing the church organ, painting on vellum, discovering the North Pole, that insidious habit, looping the loop, being astronomer royal, and so on. For these things, we do not wish a man to do at all unless he does them well. It is, on the contrary, a thing analogous to writing one's own love letters or blowing one's own nose. These things we want a man to do for himself even if he does them very badly. I am not here arguing the truth of any of these conceptions. I know that some moderns are asking to have their wives chosen by scientists, and they may soon be asking, for all I know, to have their noses blown by nurses. I am merely saying that mankind does recognize these universal human functions, and democracy classes government among them."

[00:27:37] Brian: "In short, the democratic faith is this: that the most terribly important things must be left to ordinary men themselves, the mating of the sexes, the rearing of the young, the laws of the state." This is democracy, and in this, I have always believed. And I think the point he's making, and obviously this is a political statement, but the point he is making is there are certain things that we just kind of need to do even if we do them poorly. It just requires the... like, an individual, individual context that's so unique per person that we bring something to the table that nobody else can bring. And that's just part of being human. And I think that that is what we can say to a lot of tool application. There's a part of the job that just comes with being who you are. This is the the people factor, having the right team, and so on. And, like, there are things you're just gonna want people to do that are part of who they are.

[00:28:37] Phillip: And... "left to ordinary men, the mating of the sexes, the rearing of the young, the laws of the state, and the purchasing of things with one click shopping." That's... that's the argument you're making?

[00:28:50] Brian: That's... no. No. I'm saying purchasing of things with one click shopping is something that should be done with extraordinary... what do you say?

[00:28:58] Phillip: Even if you're bad at it. That's what G.K. Chesterton is saying. Matt, get in here. I need you to...

[00:29:04] Matt Maher: No, no, no. So first of all, I love a long form quote read because it's like, you can kind of like get into it. I transported for a second. I went to Audible and you were just like a great narrator. But I think it's actually really powerful. Because I'm a fan of your show, I'm just gonna then reference an episode of yours I just listened to. It was episode four fifty eight with Gillian Katz from Hannah Grey. And you talked about AI symbiosis of, you actually have to live with these machines. And I think that is a critical piece because what you're describing, Brian, is what a lot of big tech is saying is like, it will replace, replace, replace.

[00:29:36] Brian: Right.

[00:29:37] Matt Maher: And what that entire quote is, is like, no, there is a clear delineation of like what a human is gonna do and then essentially what an AI is gonna do. And I'm gonna double down on...

[00:29:46] Brian: Your quote.

[00:29:47] Matt Maher: And I'd say, you know, I reference a lot to clients that the E.O. Wilson quote of, we have paleolithic emotions, medieval institutions and godlike technology. And it's the same type of issue as like our brains, the MIT study shows it... like are not ready to handle, forget smartphone addiction, forget all the other things it can't handle. But now this new level of delegating cognitive loads to say, we're going to just atrophy this muscle in our brain because we're gonna get a dopamine hit if we just give it to the agent. And then on the receiving end, I do that and deliver it to you. And then you, instead of reading what I said, you know, put it through your system to get it. It's like that dilution of information, what it actually does to the human brain is not good. So I think there is something very powerful and potent in what you're saying there is like, we are gonna have to exist in the world where we do things as humans that actually are unique and better than the machines, believe it or not, no matter how much they advance. I mean, Sam Altman said that. He said like, in... he's saying in the future, you know, five or ten years from now, the best model will be an okay poet, like barely okay or barely passable.

[00:30:54] Matt Maher: Like, because these are probabilistic statistical models. Like they're not great at math and they're not great at poetry or a really good written word. So it's like that is a reality and there's no world in which it's getting better unless there's more human data to train on. So I think that's the biggest thing for the business world and enterprise and retailers is to live in that world to say there are things that happen in a boutique, in a retail store that a human is going to always be better at than the humanoid robot or any level of front of house AI that you have there. And you need to lean into that and you can't throw the baby out with the bathwater and get rid of all these folks in hopes of efficiency because then, you know, your business is going to end up crashing when nobody wants to go there to ever shop again. So I think that's the larger thing that we have to continue to push forward with, is that humans and AI have to live together in a way that makes sense. And hopefully we get there. I'm really hoping we get there, but we'll see.

[00:31:49] Brian: Yeah. And I think that, to Phillip's point and actually Sam Altman just sort of regurgitated this, we need to have...

[00:31:57] Phillip: Unwittingly, I think. Unwittingly.

[00:31:59] Brian: Unwittingly. Yes. I... I don't think he read Future Commerce and then said this, but like...

[00:32:03] Phillip: We should just say that...

[00:32:04] Brian: He did. A human world and an agent world, and that split them. We need to have interfaces that both can engage with. And I... that everything you just said, Matt, reminds me of one of my favorite, you know, authors, Norbert Wiener, who's... I've quoted many times on this show and the quote unquote father of cybernetics, but he... or whatever whatever his title is. Anyway, he had this whole thesis back in the fifties that there are some things that should be human to human, some things that should be machine to machine. And we need to figure out what those things are and, like... like, create those buckets. And the faster we can actually create those buckets, actually the faster we can go after transforming the world and figuring out how things work together. Because... and I think, you know, his book was titled "The Human Use of Human Beings." We're not very good at figuring out what what we're supposed to be doing, I think, and where we fit in.

[00:33:03] Phillip: We like to incur a lot of friction where things could be automated, and then we like to automate a lot of things that should have friction. That's where things get a little bit wacky in the world. And I think that's also where some of the agentic promise versus the reality starts to rear a little bit of its... the hype head, if you will. There's things that... look, they sound really good. They sound really good. I don't know that they deliver in the way that maybe we hope or we can envision that they will. For instance, conversational commerce. We're all talking to our computers, but it's not the Alexa in in the room that we all thought that they would be. It takes on a different form. Matt, I know that you've done a lot of thinking and a lot of work specifically with, you know, a lot of brands. You had a real moment recently on, you know, how you're engaging with, you know, say PayPal to Target, or, you know, this... you were touching on Universal Commerce Protocol, for instance, earlier. You know, we're thinking about agentic as a consumer experience of shopping on our behalf. I think we just talked a little bit about how that automation may not always play out the way that maybe we were thinking it should be, or what should or shouldn't be the role of a computer with, you know, deployed on the behalf of a of a retailer. I'm curious. How does that play out in the best possible way, and how does it play out in reality when...

[00:34:38] Matt Maher: Totally. Great question. And I think something that's worth kind of spending some time on because I've started to map it out. I try to... I call it the human handoff of like, yes, we're using AI tools for discoverability. Google's not going anywhere, but we're still using them. We're hitting critical mass. I mean, you have 900,000,000 weekly users in GPT, 500,000,000 on Gemini, tens of millions on Perplexity. Like it's a critical mass of people actually using AI tools for discoverability. But where I try to map it, especially for retailers, are across kind of three ways of low cost, low emotion all the way up to like high cost, high emotion. And what we're trying to figure out is where does that human handoff happen where the human says, thank you for all that help. I'm going to take over from now. And here's our argument. For low cost, low emotion, so think like toilet paper, toothbrushes, not a ton of brand loyalty there. If an agent, you know, if Amazon's Rufus can rip through 10,000 reviewed product and get it to my doorstep in twenty four hours, then sure. I think people are going to be comfortable with an agent getting you your toothpaste, getting you your toilet paper, all those kind of low cost, low emotion items.

[00:35:43] Matt Maher: So the handoff might never even happen. An agent might do it end to end and certain retailers need to plan for that. Middle cost, middle emotion. Now think like a bicycle, a crib, a refrigerator. I don't see too many people walking around with like a Whirlpool hat or like an "I love Samsung" or LG shirt on. But at the same time, if you're going to spend 5 or $10,000 on a refrigerator, I don't think people are going to be too... they're just not going to say, sure agent, like with all the mistakes or hallucinations, go get me that $10,000 fridge. Let me know when it's scheduled and coming, right? The handoff probably happens later. Awareness, discovery goes through all these different tools. GPT is finding me the best, comparing using Reddit and YouTube, and then it's handing off and giving it to me. But I'm going to make that purchase because it's a high ticket purchase. The last one, high cost, high emotion. Think cars, think houses, think of the handbag. If I'm going to buy a beautiful brand new Chanel handbag, the last thing I'm going to do is have AI end to end buy that for me.

[00:36:42] Matt Maher: Even further so, again from an anthropological lens, I'm walking down the fashion district in New York City with a bag and someone says, that looks fantastic. And I say, Gemini picked it out. Like there's nothing worse that would happen than to be like, oh, you have no taste. You don't care. AI's picking your fashion now. So I think that handoff happens way, way earlier where a human says like, no, I want the craftsmanship. I want to make that decision. This is my identity. My identity gets tied to some of these higher products, these high cost, high emotion products. So I don't want an agent at all messing up. Forget the purchase, but even curating. So I think that's the spectrum we kind of map on. And then you as a retailer say, where should we be? Where should we step in? And again, this goes into the bigger question of what does our website mean in 2026? Are we making it for humans? Are we making it for agents? Are we making it for both? And how real is that future going to be?

[00:37:39] Brian: Yeah. It seems like... it's interesting across those tiers, like, there's probably an arc for AI usage. So you brought up, like, toilet paper as, like, the first tier. And I mean, that tier is not gonna require a lot of, like, engagement with AI to, like, make that purchase. The second tier seems like the tier to me where you're actually gonna use AI the most in the process. Now you won't check out on AI because it's, like you said, the trust in making sure you get the, like, the right thing and then the payments going to the right people and blah blah blah. Like, all that's that... that you don't wanna do it in context. But you're gonna research. Which one's the most dependable? Which one's the most, you know, like has these features? And so on. And then the final category feels like you may use it somewhat just in case, like, you do a little price shopping, you might do a little bit of, like, feature comparison, like, if you're thinking about certain things potentially. But mostly, you're buying it because you really, really wanna show something off or you have personal tastes or something along those lines. The middle category though is the one that I think is most exciting in terms of like actual usage of of AI for the consumer. Sorry. I know that's a little bit of a side note, but it is interesting.

[00:39:00] Phillip: I do. Matt, there's something that we're seeing too, though, that, again, you know, speaking... either seeing on the conference circuit or seeing folks in our, you know, in our circles, subscribers to our premium content. You know, they're speaking back to us saying, consumers, especially the savvy, savvy ones, they're using GPT to their advantage too. Right? They're using it to write CX tickets. They're... some of the crafty ones are using it to falsify returns. They're creating images of broken product that arrived perfectly fine. There's just a new level that we're all having to contend with. And so to some degree, you have consumers at one end using AI to manage, you know, real and maybe not so real interactions with brands, writing reviews, etcetera. Then you have brands, you know, operators who are having to use GPT and LLMs to manage their relationship with the consumer. And we're meeting in the middle, m-e-a-t. We're the meat in the middle. Right? And it's... at some point, like, you have to wonder when we get disintermediated, then disintermediated, like, where are the friction... anyway, it'll probably be easier if we just had those two things talk to each other and figure it all out. And that's why I think the AI agent making a purchase decision instead of the human is to some degree, that's where you start wondering about the... there are things like you said, that are probably high high trust in your agent, low brand loyalty. Mhmm. It doesn't matter. There is a gradient there. But for other things, we want the friction. It's the whole point. So I don't know. How long do you think? I mean, let's... where would you put the line? Like, where could you hazard a guess on how long we're going to see before that plays out? Or is it happening right now and we probably don't realize that it's happening among early adopters?

[00:40:56] Matt Maher: It's happening among early adopters, but actually it's a great point, Phillip. With clients, I try to delineate two things because, again, think of the whole marketing funnel, right? You start with awareness. Let's just pull it, make it simple. Awareness, finding out, just transaction. And there's two different worlds happening with AI on both of those. I think for retailers specifically, they don't know where to put their essential energy, right? So you start with discoverability, and I think Brian hit on this before of, yes, that is happening. That is happening in the billions of queries per day that are happening with AI first search engines that people are learning and understanding and getting more information. They're more savvy, they know more about your brand. They are equipped with information and you need to plan for that future. That means you need to make sure that when all these LLMs are crawling your site, they're getting surfaced the right information. You don't have long load times with JavaScript because they will instantly go away from your site and go to Reddit and YouTube. There's a whole bunch of hygiene that needs to happen. So that is happening right now because we've hit critical mass when it comes to people using AI first search tools. So retailers, fix that immediately. But let's go to the other side, which probably gets more oxygen now, is agentic shopping, agentic commerce. Yeah. That the agent is going to your site. It's ChatGPT's Atlas, it's Perplexity's Comet browser. They're purchasing all these things and my site is not set up for agents.

[00:42:15] Matt Maher: "I'm gonna lose all these sales." And law of numbers, that is so early. It's so early days. I joke with the meme from Arrested Development that there are dozens of us using agentic shopping tools. Meaning the people that are actually letting their agents just go end to end and buy all this stuff is really small. As someone who has done this a ton as part of what we do is test every one of these, the fail rate is very high. Like I've had things delivered late, I've had wrong orders, inventory hasn't been checked. That's because the obviously relative barometer of the maturation of digital commerce is really high. Amazon, Walmart, Target, they're all fantastic. They get you stuff immediately. They rarely mess up. But the reality is people aren't doing that right now but yet everyone is racing towards that, get my website agent ready so it can take the agent's hand, white glove, walk it through the red carpet. I think this is an issue, or it's perceptual to say you need to plan for the first thing, that you need your site to be very clean and make sure it's read by bots because if you're using Reddit, if you're a brand and your site is not optimized for large language models and it says, no, this is too loaded, let's go over to Reddit or YouTube, you're not controlling the narrative anymore. And if there's negative sentiment on Reddit, that's not going to be great.

[00:43:27] Matt Maher: So again, it goes to the point of there is an actualized future right now where retailers absolutely need to clean up their site and make sure everything is optimized for LLMs, but they don't have to race right now to try to set up the right protocols for agentic commerce. Yes, it's good to set it up now, but don't put all your energy and all your capital for perhaps 10 or 20 people that might purchase something agentically. So I think it's uneven remainders in terms of where time is spent, that we have an actualized future, directionally correct future that, yes, to Brian's point too, low cost, low emotion. I do think people are going to let agents do this and the toilet papers and the toothbrushes of the world need to prepare for that and they should put their effort into agentic commerce and autonomous shopping. But I think that's where the mismatch is, is that, as we said this before, it's like people say AI and just want all of it. You really have to break it down and say, what are we trying to solve for? Great, we're losing market share in terms of our discoverability on AI platforms. Partner with Profound, partner with Evertune, get those reports, understand why you're not showing up and start making the small fixes to show up. That is what you should do now. It's not hiring 10 AI startups to, you know, make sure your agent to agent shopping is perfect by the start of 2027.

[00:44:40] Phillip: Yeah. Evertune's an AI marketing platform that, in case, you know, people weren't aware. These are... I mean, these are... if you're not on top of these things, I feel like these are... you you need to... you need to be on top of this. We should probably do an entire sort of tactical episode on on the new stack. That's something maybe we'll have you back to chat about, something like that.

[00:45:07] Brian: One thing I just wanted to add, one of the reasons why the agentic thing is really tough right now too is that actually the things that you would typically buy in that range are on Walmart and Amazon, maybe Costco. Right? So a lot of people are getting a lot of those things in their, you know, their weekly Costco trip or their weekly grocery trip. And it's almost... it's very like low mental load because they're already there. They're gonna be there. People are gonna continue to wanna pick groceries. I know that grocery ordering is up, but there's still a lot of like grocery that happens at Costco, Walmart, etcetera. And then this is the other kicker. Walmart and other major sites, when they see a bot on their site, they're asking, are you a real person? And this is a big problem for agentic shopping right now because a lot of it... a lot of agentic shopping should be... it's sort of targeted at budget. Right? They're looking for a good deal. It's passive. It's looking for things that will, you know, you're gonna be able to get what you need that are cheaper. And a lot of the sites that have those kinds of deals are all blocking. You're not gonna use this to buy jewelry. You're not gonna use this to buy a lot of the things, like you said, that are high taste items. The ones that maybe have like that... that aren't blocking, that don't have that sophistication. The things that sell basics where you would wanna apply agentic are all blocking you. And so...

[00:46:41] Matt Maher: I mean, it's f... I mean, we have Amazon won what I call a landmark lawsuit against Perplexity against... no agentic shopping on their site. eBay just straight banned it. Target has updated their Ts and Cs to say we have no agentic shopping on our site. So I mean, we're seeing, you're right, Brian, we're seeing it across the board. And for, I'd say ulterior motives, Amazon has a $60,000,000,000 ad business to protect. Think about the make goods when an agent is on their site. And we've done with all our clients... I've agentically bought tons of things. Again, it's taking over my browser. So their backend data is showing essentially a human browsing. So that messes up their first party data, messes up their ad, messes up everything. So I understand the backlash of banning all this, but you're right. It's... what it would be most used for has been banned now. So again, the consumer loses in that one.

[00:47:28] Brian: Yep.

[00:47:29] Phillip: I call this agentic ghettos. I'm like, we shouldn't be... we shouldn't be trying to create these little pockets of the world either through legislation or through, you know, blocking and robots dot txt, like, allow listing. We shouldn't be creating these priority lanes or, you know, blocked lanes for certain agents of, you know, in deal making of what is allowed to shop where and when. If truly agents are an extension or a proxy for human behavior, then a human is supposed to be what's driving the interest. Right? And that is... we're in an early stage right now where everything feels very competitive, but the future should be open. And if we don't see that, if it's still being driven by the platforms, then I think we see things, you know, come to sort of... you know, there's an egress where we're running more from our local machines and it's gonna be harder to block and stop. And I see that's gonna be an interesting future to have to work the kinks out in. Yeah. Matt, it's been such a pleasure to have you. We would love for you to come back. It's not enough time for sure. We always love to pick these, you know, conversations apart. I'm going to ask you, think out twelve months. What do you think brands are going to get wrong in the next twelve months? What do you think they're going to get right?

[00:48:54] Matt Maher: I think the biggest mistake or problem brands will have in the next twelve months, again, we hit on it before but it's worth reiterating, is naturally identifying the problems they need to solve and instead just continuing to feel like they need the treadmill to speed up and race towards a future but not actually move forward because they're on a treadmill. I think the idea here is that they make the mistake because, again, they're applying solutions to things that aren't problems. I think the brands that will succeed will understand exactly how they're getting more productive, where they can layer on AI in the right sense and measure its ROI. And I think that's the big piece moving forward. It felt like 2023 and 2024 was the year of aspiration, GPT's out, AI can do amazing things. 2025 was the year of action, start doing pilots, start getting copilots, start getting Zoom AI assisted and Claude and all this. And 2026 and 2027, I think is going to be the year or years of assessment of saying, is this productive? Are the people getting better or more burnt out? So I think an assessment of really where you are with using these tools is going to be the critical piece for the next twelve months. And those that come on the other side of the assessment to say this was good and this was definitely bad and here's how our new playbook is going to be for 2028 and 2030 are going to be the brands that succeed.

[00:50:15] Matt Maher: The ones that will fail are going to essentially keep going all-in and keep doubling down on every single tool without having all the infrastructural hygiene that they need. And then all of a sudden that water level is going to rise higher and higher until they're just drowning. And then if they're, you know, from the outside world too, they become more and more invisible. And I think that's the last thing I'll say. What I say to clients is all this AI first search, all that's happening is the aperture of the internet continues to shrink and everything becomes more personalized for each and every... what people see, you don't exist anymore. You literally don't exist. So I think the brands that are smart will stay in that aperture, assess, admit mistakes and move forward. And the ones that will fail are the ones that don't care about that aperture, think it's all business as usual. And all of a sudden that water kind of rises too high and they're no longer relevant or making revenue.

[00:51:07] Phillip: What a way to end. Thank you for the mic drop. Matt Maher, he's the founder over at M7 Innovations. You can get more information, m7innovations.co. Follow him on social, Matt Maher. Thank you so much. We'll drop all those links in the show notes. Thank you for watching this episode of Future Commerce. If it sparks something for you, like, follow, subscribe, futurecommerce.com. Get our new book, futurecommerce.com/strata. Remember, commerce shapes the future because commerce is culture. We'll see you next time.