Barry McGeough on AI’s Retail Disruption | Retail Rebels 2025

So, um, hope you guys had a really good conference. And Michael, thanks for asking me to come and speak at another PI event. It's really good. Uh, my name is Barry McGough. I run, uh, corporate applied innovation for a, a, a company called AmeriCo Group. We're here in New York, uh, and we're— I'm fortunate enough to be able to do a lot of material science and a lot of, a lot of digital transformation work over the last few years, and we're tracking all the, the things that we are fascinated with and addicted to trying to understand and turn into revenue and, and positive outcomes for our business. So been tracking everything that's happening with AI. It's really important. So what we're looking at right now is what's happening with visual AI, what's happening with commerce, and what, and what are some of the implications that we need to think about, some of the opportunities and some of the watchouts. So we have been having a lot of fun with AI. If you guys remember this in 2023, this is the Pope in a puffy jacket. We thought it was awesome. You can't believe you could do this. And we thought this is unbelievably groundbreaking, meme-worthy, fun stuff. Just a couple of years later with Midjourney, we can do this. So we can have the Pope playing Jimi Hendrix. It's awesome. So it's even more awesome. So it's getting more awesome, it's getting more fun, it's getting more experiential. We know that we're using AI a lot in, in business. So if you look at this McKinsey graph, um, in almost every geography, in every country, people are, are implementing AI at about an 80% rate. This is the number of companies that say they're using AI technology and computer vision, machine learning, uh, and, and putting it into at least one of their business functions. But where are they using it? And where are we using it in society at large? So Harvard Business Review does a great, a great job in, in tracking what the killer AI use cases are. Last year, the number one use case for AI was content creation and editing. What's very interesting is still number two, but in one year from another, for another thread, another conversation, the number one use case is using AI for companionship and therapy and personal advice. That's very, very interesting and we'll kind of touch on that at the end of why that matters, right? Everyone's looking for the killer use case. So where are we using it in business? This is another good McKinsey study from March, seems like a million years ago, but they were really talking about in our vertical, in consumer and retail, we're using AI at about 46% clip in marketing, 21% in product development, and in IT about 20%. So we can see where, where we're using it in our industry, which is really important. We're also trying to figure out where are we gonna make money. So this is BCG from, from last year, but it's still really salient because what they're listing now, what they're talking about is that about 96% of companies are playing with AI and only 26% have moved past pilots. So companies that are trying AI are putting money into it and it's not necessarily a satisfying meal yet. That's really, really interesting. So when's it gonna become relevant to our businesses? By 2027 though, these same companies said that they think that they will have, they will be, then that's in 2 years, they're gonna realize 45% greater cost reduction and 60% more revenue growth. So they're gonna start to, we're gonna, as businesses, we're gonna start to use AI to lower costs and find new products. Lines. So where are we doing it? According to the same McKinsey study, uh, we're— while people are using a lot of text generation and text-based ChatGPT searches, we're still being used a lot for images and video. So I had a really interesting experience in way back in February where I went up, uptown to Fordham University, and I work with a company called Machine Cinema, and it was kind of a sit-in. and they got, um, they got 60 people to, to, to make a 5-minute video and they had 4 hours to do it. And what I didn't really realize is the complexity of being able to do this. It's not just throwing it into a ChatGPT engine. You have to figure music, you have to do voiceovers. There are act— they used actually 15 different tools in this seminar. And it was humbling to understand that it's not any good innovation. They say is akin to magic, but it's not magic. It's a tool, and you still have to learn completely new languages. This conversation about upskilling is really important. You're going to have to learn 15 new tools if you want to be an editor of AI movies. It was amazing to watch. And this is what they came up with. I won't play the whole thing. So the prompt was, do something in New York and make it sort of edgy and period. And you can see You see that her face is a little bit fake. You get a little bit of hallucinating, weird license plates. But it's not bad. That was February. This is Sora last month. Welcome to a nonexistent car show. Let's see some opinions. I mean, man, the acceleration is crazy. You look far, step on the pedal, And you are there.

I feel safe with him in an SUV. And it seems to be the right type of car for him.

I think the range is only going to get better. Sorry. We don't want to drive gas cars anymore.

Yeah, no more gas cars.

You can see I'm kind of a misfit here. But don't tell anyone, I've just bought an electric car.

I think it's really great for families and for little babies. With all the safety features that these SUVs have.

But what you're really seeing is that technology is going to be very, very important in terms of how we go forward.

It was great to come to the conference because my husband loves cars.

So you get the idea that what we've done— and we're trying to see this in real time— is we've crossed the uncanny valley. The uncanny valley is this idea that something is fake enough that you don't believe it until it becomes so believable that you do believe it. So that one was VO through Google VO3. This is, this is Sora. And you can see that, that these look like Instagram influencer pictures of people expressing real emotion. And I, I believe it. I, I'm there. I think this is actually feels quite authentic. So that's how fast this is moving. It's moving so fast right now that the way people are using it commercially— and this just happened in August— is Vogue magazine actually posted two AI-generated models, Vivienne and Anastasia. And a lot of people didn't know for a while until they got kind of outed by an Instagram influencer. And she is very cool. I really like her a lot, Karinya. And she went crazy on her Instagram post. And she said, they're taking away jobs from gaffers, makeup people, real models. These are real jobs. She was pissed. She was very, very angry. And so were a lot of other people that decided that they were gonna cancel their subscriptions to Vogue because Vogue is about real people and real model and real luxury. So if you think about what's happening in the microcosm of our industry, we also, it's very important for us to think about legislation and what's happening in society at large. So this is a very interesting development in California where AB 213, it was on Gavin Newsom's desk. It actually got signed about 2 weeks ago that would put regulations, and this is the nation's first regulations around how data is compiled and how it's treated and how it's processed in these AI engines. What was passed a year ago is even more interesting. and that is AB 2602, and that basically says that it protects people's name, image, and likeness so they can't be used and can't be replicated without permission. That went into effect October 1st. So what's going to be really interesting is how will it be enforced? We don't know yet. This all happened like 2 weeks ago. What's super interesting now is we're watching corporations try to make sense of this counter— of this new technology. So a year ago, Gartner, in their recommendations about being resilient around AI, they suggested as a way for enterprises to protect themselves that one of the things executives should do was revise employer contracts to get permission to use their name, image, and likeness so that they can use that as data and assets on the go-forward and they can't get sued for it. This is corporate legal advice. So, going on to some of the technologies that we're seeing, I'm probably telling you a lot of things that you know, but Google Nanobanana just came out a couple weeks ago, and if you haven't tried it, it's super cool, it's very easy to use, it's drop and drag Photoshop, and it's really, really uncanny. This is what people are doing now. With Sora too. This was actually just released about 3 weeks ago, and I'm sure everyone has seen Michael Jackson in Star Wars, has seen, you know, Hitler on The View. I mean, this is getting to be like really, really weird stuff. And the family of Martin Luther King actually said, "You need to cease and desist. This is not okay that you're having Martin Luther King do these goofy things or say these things that he didn't say," right? How are we going to control that? It's a real question. This is something actually that's used in a more practical way. This is something that was activated at Magic. This is a Champion billboard, and both of these are, it's the same video. These all started with some technology that we're actually using in our industry from Raspberry AI. And it, these started as 2D Illustrator images of socks, t-shirts, and underwear. They were turned into 3D assets. The 3D assets were then put on 3D synthetic models of people that never existed. These people were written in a product brief and turned into a 7-second video that was at the entrance of Magic for, for a week, for 5, for 8 days. It was a 7-second animated video. It got 1.8 million impressions, and everything about this was synthetic. So when we're talking about using these tools in a practical way, not in an AI slop way, but in a practical way, we're starting to see the practical application of being able to use AI in our industry. Now, a couple years ago I talked about the, a road to AI ubiquity. How, when would we see AI everywhere and it would just be the ingredient layer in everything? What we talked about is that it would start to work its way into the AI workspace. It did with Gemini, Microsoft Office, GitHub if you're a coder. It started to also work its way into search. So if you guys haven't used Gemini Search, well, you have because it's now in Google Search and it's integrated that you can't not use it. So Perplexity, these smaller engines like Coggy and You, I really like Perplexity Claude a lot, starting to work its way into search as well. And so what we thought is that once we saw that in the Office suite and once we saw it in search, then it would be the ingredient layer in everything. And it actually is. This just happened 4 days ago. So ChatGPT just launched Atlas. I picked this image because I think it is kind of funny. People are still trying to figure out whether it's better than Google or not. So this is a Reddit clip that someone said they're going to just move it into the trash. Don't know whether it works or not and how it's going to be useful, but it puts a ChatGPT query bar on, on, on your taskbar. And it allows you to use ChatGPT as your search engine. A lot more is happening here as search evolves because the next thing that happened, and this was last week, is that it was announced with— by Walmart that, that ChatGPT would, would allow— you would be able to search and shop inside ChatGPT without ever leaving ChatGPT and buy Walmart products. That's really, really interesting. That has implications because you can, you never have to leave the app. It's kind of like what's happening in China. You never leave the app. So then we get into this question of whether we're in this era of incumbents being challenged by emerging insurgents. Right now, over 40% of shoppers use Google to get new, to get information And Google lead services generated $850 billion in economic activity last year. By contrast, ChatGPT drove about $1.9 billion in commerce. So the economies are, are really, really off. I put this in here because I think it's really important. If you don't know Amy Webb, she's from the Future Today Strategy Group, and this is her post in LinkedIn. I, I wrote the whole thing. Right. And I know it's very wordy. It's not what you're supposed to do on stage, but it's really important because what she said is the conversation is you won't Google which blender to buy. You'll ask ChatGPT to find you one and sell it to you. That means no search, no marketing, no affiliates, no click-through. And that revenue stream for Google is gone. This is a totally rewiring of the way we do business, the way we transact, and even the economy itself. 'Cause all that Google money is gonna have to go somewhere. And will it go to ChatGPT fast enough? This is consequential. So let's talk about agents. MIT Technology Review did a really, really good job in studying. There was a, a whole issue on agents in, uh, August. and what they talked about was its use case in coding. And what they found was that if you, if you apply ChatGPT and, and you applied agency rather to coding, that you would double coding speed every 7 months. So in about 4 years, a, an agent could do a month's worth of coding in a couple of days. So the acceleration of the, and the promise of agents is really, really attractive. Salesforce is all in for agents. So they have Agentforce. So you have an agent builder, so you can work in natural language. And in natural language, you tell the agent what you want it to do, and it builds you the guardrails around what you want it to do. Customer service, sales development, marketing. You just work in natural language, and these agents will go out and do multitask functions and then bring it back to you. I'm gonna talk a little bit about Walmart as, as we kind of plow through this. And so what's interesting is Walmart is actually very much on the cutting edge and very impressive with what they're doing with agents. So on August 25th, just a couple months ago, they decided they were gonna take all their AI efforts and they were gonna put it into 4 platforms, Sparky, Associate, Marty, and Developer. So manufact— so supply chain, employee support, AI testing and development, and then the customer-facing applications. So they were doing, they've been working on data and how and what to do with data for a decade. They're probably the best at it. And so with, they decided they had so many disparate efforts that they were gonna focus it in. In the same month, they announced Wibi, and Wibi is a super agent. So they're taking all their agency efforts and putting into, into one super agent program that unifies 200 AI agents and, and, and, and brings all that into really purposeful delivery. So this is the delivery system of, of agency. A few weeks later, it's all happening in real time. Apple— Google launched their AP2. AP2 is their, uh, advanced, uh, purchasing protocol. And they, they partnered with about 40 different major, major corporations. And what they said was they want to be able to, to have consumers have their agent buy something around a certain set of parameters. I want to buy this green jacket. I don't want to pay more than 20% more. And then you have to give it permission to put it to shop, put it in the cart, accept it, and buy it. So there's guardrails around it theoretically. But this is being launched right now where you as consumers can actually have an agent shop for you. At the same time, Amazon did something for their Amazon Marketplace and Amazon sellers. How do they have an agent optimize how they sell so that they can get the right holiday promotion, they can use the right imageries, they can have the right hooks, they can— and they can also use it as an analytical dashboard. So agents are starting to get involved in all aspects of selling, marketplace, selling ourselves, our own individuals. And what's interesting is when you, when we're starting to imagine or see this protocol where agents are talking to agents, then what's really interesting is if we're getting advice from agents, as humans, we believe agents more than we believe advertisements and people because we think that they know us. So they have authority, they have weight. Now, what is this weight multiplied by? It's multiplied by a billion searches a week. Half of web traffic is, is conducted by bots, not by humans. And the traffic from AI to retail websites jumped 1,200% last month. And 40% of us use AI to help us shop. So you can imagine the adoptive curve, adoption curve is that we're also gonna start to use agents pretty soon. We're gonna be comfortable with it. And we're gonna trust them. And that means that we have a crisis of disintermediation. So what does that mean? That means that when we have our agents do something, and a retailer has their agent do something, then it's agent to agent discourse. And we don't have any conversation, 'cause we've given parameters, we wanna pay 20% more and no less. So we may get served a brand we've never heard of, and then we say, go ahead and buy it. So what does a brand relationship look like? These are beloved brands. John Deere, Lululemon, Brooks. These are brands that know you. You love them because they know you. They know how you know how they run, how you can take a yoga in their studios, and they give you fresh coconut juice, right? You love them. And so what's gonna happen to brands? This is a really, really big open question. And when we talk about synthetics, like we discussed in the Champion billboard, we're also seeing the cost of this, this intermediation really, really drop down. So this is a company called Inception Point AI, and they do podcasts. They do 5,000 podcasts and 3,000 episodes a week. They just run 'em through a machine, right? And they have these synthetic influencers that talk about everything from cooking to travel. It's very charming, right? And it costs them about $1 per episode. They've had 10 million downloads since last year, and all they need is 20 listeners to break even. Imagine. So they're monetizing synthetics and they're monetizing influence. This just happened last week too. This is Tencent, and they're talking— what they have a new capability of being able to do is allow advertisers to go into existing movies, films, television, and other media, and basically lasso out whatever old advertisement was there and put in new branding, and also put in branding where it doesn't exist. So here's some table cards, right? And then they, so this can also be geolocated by promotion, by time. So you can, this can all be happening. This is marketing in real time. So if you think about all this, what is the number one use case that we're actually using all this AI power for? Well, we're using it for slop, right? So I showed you all this stuff before. This is actually just a couple days ago. This was the number one TikTok video of cats shooting bazookas and guns on front doors. So this is where our heads are at. With all this awesome technology, this is what we're tempted to do with it. And the reason why I bring it up is because there's also, 'cause we're talking, I'm talking about right now brands and brand trust and brand loyalty and who do you know and who will you listen to and will you listen to brands or will you listen to agents. And there's a concept that actually was a word of the year 2 years ago and it's called enshittification. It was, there's a book about it from Cory Doctorow and he's from the Electronic Foundation, Frontier Foundation. Trying to protect a saner and less rage-oriented internet, just like Tristan Harris. And so the idea of enshittification is engagement and rage, just like that cat videos shooting bazookas, right? So if you can change the internet or social media from being useful to be engaging, then you can monetize it. So actually, I wanted to get some examples for today, so I pulled some examples. So dating apps, they actually actively work to keep you single. They don't want you to find the person of your dreams because they want you on the app. Duolingo is working on gamification to keep you on the app longer so that you don't learn a language, you're on the app, right? And Google will serve you things now that are totally irrelevant, second and third order search ideas. So they want to sort of double and triple down. So if you search pants before, you might get these. Now you might get someone who's yanking someone's pants off at a Super Bowl game, right? You might get something humorous or enraging. So this idea of enshittification is actually real, and it has real-life consequences in terms of what we believe and what we don't believe. So if you guys have not heard of the dead internet theory, it's the idea that there is things like link rot, all the bot traffic that you're seeing, things that you don't believe, algorithmic death, right? And so if we create all this junk, all this slop, all this enshittification, then what we're gonna get is an internet that we no longer trust. That's a problem too. Because who will we trust? Where will we purchase? Will we trust brands? And what we will, what we will do with this. So when we're thinking about agency, there's the other aspect of it too, which is what's the downside of agency? That's so— I described some of them, but the other ones are really, really also consequential and they make you kind of scratch your head and think. So what if you ask an agent, hey agent, make me a million dollars. You didn't say how. So what if they go in and they con old people out of their retirement savings because they have a really good Ponzi scheme and they do $100,000 or $10,000 times 100 people and they bring it back a million bucks. The agent can't go to jail. What are you gonna do, arrest the agent? That's a problem, right? What if there you had rules of engagement around border conflicts like the one that just happened with Cambodia and Thailand, which affects our industry, right? There were border skirmishes around a temple? And what if you said, hey, agent, I don't want anything to happen while I'm sleeping in the middle of the night. So if you see more than 20 people come over the border, shoot to kill. Is that a good idea? So it could happen. So ThunderForge is part of the US Department of Defense initiative, and they're working with Scale AI and, and, and Anderol, which is a big defense company, and they're actually working on agents for the military. This is actually happening, right? So their ThunderForge marks a decisive shift toward AI-powered data-driven warfare, ensuring that US forces can anticipate and respond to threats with speed and precision. That's interesting. So I'm gonna end here with Noah Yuval Harari. And what he, what he's talking about, it's a really, really great interview. And he is really ringing the alarm bells around AI, as are a lot of ethicists. And a lot of people are saying, because we talk about ethical AI, but that's just two words in the English language. What does it actually mean? So what he's basically saying is we have to think about agents. What are the dangers of unchecked agency? Because agents will do what agents are supposed to do. They don't have morals, they don't have ethics, and we're not at AGI yet. So what he's basically saying is the very definition of AI is learning and changing and doing things. So what we would say is, okay, agent, we're gonna give you rules. Don't break these rules. Cart engagement, sales protocol, crossing borders. Don't break these rules. And if you could ring fence AI, then it wouldn't be AI anymore. It would be like a toaster. A toaster does one thing, it makes toast. So an agent would only do one thing, but that's not the whole point of agency. So it's antithetical to say that agents won't be agents, 'cause agents will be agents. Bill Gates is very, very bullish on agency. And what he says is agents are smarter, just bit smarter than us. They're proactive, they can make decisions, they can do stuff, they can work across applications, they're endlessly intelligent. and they can predict your patterns and your behavior. So one thing he does say though is that based on the information, they'll give you what you need, but you, the human, always need to make the final decision. So the one thing I wanna leave everybody with, 'cause there was some kind of Heisenberg rollercoaster ride in our conversation, is we have to make sure we're people that are making decisions, we're people that are deploying the frontline of this, we're, we're touching the technology. We really have to think about keeping the human in the loop. So that, I'll leave you with that, and thanks for listening. Yeah. Thank you. Hi.

Hi, Barry. Great presentation as always. So, two questions not related. One is, you know, if you've got agents shopping for you, what is the— where does this leave consumer experience, I guess, from a discovery perspective, right? Because a lot of times, shopping is an entertainment activity.

Yeah.

And if you've got a bot who can presumably get you exactly what you need in a fraction of seconds, like, where does that leave the whole discovery process? That's my first question. Second question is, what about model collapse? Because if you have a whole bunch of synthetic data that is feeding these models and you constantly are generating slop, like, eventually you'll get to model collapse. So I'm just curious about your thoughts on that.

Yeah, those are great questions actually. So the, the terrifying scenario for us is that there is no, there is no engagement. Now you can imagine that maybe people will use to buy commodities first. So simple things, soap, tea, you know, you know, toilet paper or whatever. We're gonna have to really think about what engagement looks like. Engagement might be, and there's a big move back to physical retail, but we're gonna have to get very, very creative, and the reason I bring all these things up is they are things that we have to think about. So I don't bring up enshittification lightly, because we have to understand that our customers are being flooded with cat videos and other baloney. How do we get through? One of the things, I saw a really great article on why Japanese ads are so crazy. They're so crazy because there's a lot of competition for eyeballs there, and they have to really stick out. We're going to see some really, really— so we could see the sort of Japanification of imagery for us that could be more outlandish and more extreme. But we're going to— or there may be things that are more authentic, the kind of experiential things that you might get from Patagonia when they have long-form videos of people in Pangea. So it's an open question about what consumer experience is gonna be, but we do need to be aware of the fact that our consumers are not being only inundated with what we want them to see. They're inundated with everything from every vertical. Model collapse, fascinating. That is a bigger problem because, because like you said, data keeps training on things. So the more slop we feed, into the machine, the more it gives out slop responses. A lot of people think that the answer to that is AGI that can start to make the decision between what's, what's good and what's bad, sort of qual decisions, not just quant decisions. But I've, I've been reading a lot about, about the idea of model collapse and how, how risky it is that we are doing all these cat, cat, cat, cat with shotgun videos. So So that, I don't know the answer to it, but I know it's a risk factor. And it's an awesome question. Thank you. Yeah, thanks.

Like I said, really good presentation. I think I took pictures of every one of your slides.

Please, thank you.

One question is with the regulation piece, which you mentioned. Obviously, this is a wild, wild west, like the internet was years and years ago. What are you hearing as far as where things are moving from a regulation standpoint, especially since right now our government is so tied into these big companies, and there's a lot more partnership than you may have had in the past, and just what your thoughts are on about that.

I know it's the end of the day, so I won't go long. I could, because yesterday I was at the NVIDIA GTC conference in DC, and I didn't really actually understand why Silicon Valley is so all in to this administration, when the previous administration was not bad to Silicon Valley. They had the CHIPS Act, they were putting in fiber optic cable, they were working on intermittent energy, right? And energy right now is more important than it was 3 years ago. Granted, we understand it now, we didn't understand it before. But in the absence of real AI policy, what I'm finding is these big tech entrepreneurs and well-funded entrepreneurs are going in and they're creating a framework for us, 'cause there is nothing federally. So the AI Acceleration Act of July 2025 basically talked about taking away guardrails, taking away deregulation, making, speeding up the authorization to build, making sure that the primary objective was to rid Silicon Valley of a woke agenda. But it was— but, and, but so all the guardrails are actually being taken off. The legislation as it reads now says that they have a 10-year reprieve from any liability to any negative impact of AI. And if something happens, we'll deal with it later. That's the way it reads now nationally. So what's happening in EU is defanging itself. And that, that's very, very interesting to watch too. So that's what makes what's happening in California point a lot more interesting because some, there's going to have to be some adults in the room somewhere. But the, but the really interesting thing to me is I do actually understand now why Silicon Valley is all in because there is no, there is no policy. There's no overarching policy around what AI should be. So it can be whatever NVIDIA wants it to be. And so they're coming with a huge laundry list. They, they did this in DC, not in Palo Alto. They had an audience of one, and they said, we need deregulation, we need open source everything so all our data sources are free, we need to have free of liability, we need big government contracts so that we can sell chips. They actually got 7 server farm, sorry, supercomputers that will be installed in the next few years And that contract with the DOE was signed last week. So, and they also want the American government, the US government, to pay for all the energy that's required for data servers and server farms. So they have their cup out, their empty tin cup, and they're saying, if you want Industry 5.0, If you want the fifth industrial revolution, this is what we need, and they're gonna get it. That's the interesting part, is they're gonna get it. That's why they're all in. I, I, yesterday was one of the most fascinating days I've had in a long time because I really did understand why, why this is happening at such an aggressive clip. So we're gonna get a lot of the things that I'm cautioning about now. It's an open question. It's why I do like to talk about this, because you guys are leaders, right? I want us to think about and be at least clear-eyed. There may be things that we can do or choices that we can make, but it is— yeah, it's the Wild West right now. It's going to be the Wild West for at least the next 3 years. Yeah. Thank you. Good question. Hi.

[Speaker:KATHRYN] We talk a lot about agents. Where do you think agents are going to show up, or how do you think humans will interact with agents? Do you think agents will be personal shoppers and sort of individual things that we— that are not related to a website or a retailer or whatever? Do you think they'll show up as part part of our credit cards? Do you think they'll show up as part of ChatGPT or one of the LLMs that we might be using for search? Do you think they'll show up as part of a retailer and that's where the loyalty will come from? And do you think that individuals will, will sort of latch on to one type of agent, or do you think that an individual you will use lots of them?

It was forecast and predicted that, that we would embrace our own personal agents, uh, in a big way. I, I myself have a hard time because the number one use case that people put forward for agents is booking holiday trips, and the number one use case for robots is folding clothes. I don't dislike doing either, so I don't understand why that's such— so compelling. Um, I do think that agency— the reason why I talked about Salesforce is I think you're going to start to see agentic AI in the background. It will perform mundane blocking and tackling and data gathering and collection and some decision making on the backend. And so I think in supply chain, in marketing, in, in CRM, there's a lot of work. And so I can imagine deploying an agent for CRM and marketing big time. It's, I do think it may take us a little while to trust it, but then again, that's why I brought up AP2 because AP2 is a big, big development at the consumer level to use agents for shopping. They're enabling it. And, and most of the partners were people on, in, in the, in the, uh, the pay part of the supply chain. So they worked out the pay part first. So it'll take a while for us to trust it. Like we didn't trust giving the internet our credit cards, but we will. I, I, I predict that the adoption curve will be probably on the enterprise and the backend first. And consumers a little bit later as we give it a shot. Hi. Yes.

Hi. Thank you for that. You've given us a lot to think about. And certainly, the idea that things are moving so quickly— when you say March seems like so far ago and what's been developing. I know you're really plugged into all this. But to those of us who are on the brand side, are there some top-level ways to really keep up to date on what these major shifts are and how to— keep an eye on all of the—

Yeah, there's innovation. I think it's just called AI Daily. It's a, it's a YouTube. Do you know it?

No.

It's, I think it's just called AI Daily. It's on, it's a YouTube link and it, there's something on AI every single day. It's really, really good. I, I didn't, I didn't subscribe to any of these magazines before, but I actually am finding the MIT Technology Review amazing. It's really well written for civilians and not scientists. It comes every 2 months, so it's not too, too much to read. I really, really like it a lot. So I think MIT Technology Review, they had a whole issue on agency, and I learned so much, and they'll talk about power and everything else. So those are a couple, a couple good things to kind of tap into.

Yeah, thank you so much. That was incredible.

Thank you.

Definitely going to reach out to you on LinkedIn.

Yes, please. Yeah, reach out to me on LinkedIn. I love to connect.

And share.

Yes.