AI Intelligence and Competitive Advantage | Matt Krepsik & Vas Bakopoulos | Signal & Noise LIVE at Advertising Week 2026
What happens when everyone has access to the same AI?
In this special live episode of Signal & Noise, recorded in front of an audience at MCM Creative Studios during Advertising Week New York, Brett House and Rio Longacre sit down with Matt Krepsik, CEO of MediaRadar, and Vas Bakopoulos, SVP and Head of Industry Research at MMA Global, for a fast-moving conversation about AI intelligence, marketing, creativity, and competitive advantage.
The discussion starts with new research from MediaRadar examining how advertising and media professionals are actually using AI. Among 500 industry professionals surveyed in July 2026, 87% said they actively use AI in advertising and media decisions, while 92% reported changes to products, services, or pricing as AI reshapes the industry. But widespread adoption raises a much bigger question: If everyone has the same tools, where does real differentiation come from?
Matt and Vas dig into the tension between AI adoption and AI trust: data provenance, outdated information, hallucinations, proprietary data, governance, and why brands need to think carefully about the intelligence and data sitting underneath their models. The conversation explores why the competitive advantage may increasingly live not in the model itself, but in the data, memory, skills, guardrails, workflows, and proprietary knowledge surrounding it.
From there, the panel goes deep on what AI means for the future of advertising: agency operating models, in-housing, the ad-tech supply chain, creative intelligence, and the opportunity to finally reconnect planning, creative, activation, and measurement. And then comes one of the biggest ideas of the conversation: when everyone has the same box of AI-powered crayons, the advantage shifts back to the painter.
Matt and Vas make the case that the future of marketing will place an even greater premium on taste, craft, judgment, perspective, and distinctive ideas. AI can accelerate production, analyze enormous numbers of permutations, and expose new opportunities—but it still takes people to decide what is worth creating in the first place.
It’s thoughtful, provocative, occasionally irreverent, and a lot of fun—exactly what a live Signal & Noise conversation should be.
A huge thank you to MCM Creative Studios for giving us an incredible home for the evening, and especially to MediaRadar, the main sponsor of our Advertising Week activation, for helping make this special live event possible. This was also a milestone for Signal & Noise: our first episode recorded in front of a live studio audience.
If you work in advertising, media, marketing, AI, data, or creative, this is one you’ll want to hear.
Topics include:
• AI adoption in advertising
• AI trust and data provenance
• Proprietary data
• Brand data sovereignty
• AI governance • agency transformation
• In-housing
• Creative intelligence
• Taste and craft
• Human creativity
• Advertising measurement
• Agentic AI
• The future of marketing
#ArtificialIntelligence #AI #AdvertisingWeek #Advertising #Marketing #AdTech #MarTech #MediaRadar #MMAGlobal #CreativeAI #AgenticAI #MarketingAI #SignalAndNoise
Transcript:
Miles (00:03.816)
signal and noise. Bringing clarity to data, tech and AI.
Miles (00:13.698)
Hey everybody, Brett House, Signal and Noise live at Advertising Week 2026, joined by my co-host, Rio Longacre, and we've got two phenomenal guests, right? But before we introduce the guests, we're gonna say thank you to all of our sponsors. Media Radar being one of the big ones. The main sponsor, the main sponsor, right? A company called Gradial, a company called PubEx, Telium, and then a little stealth company called Nexa, that Rio and I are doing some work for on the side.
thank you all for joining us. This is actually we've done about 200 pieces of content with this podcast, and this is the first time that we've actually had a live audience. so we're gonna keep the doors open. We're gonna keep the doors open for vibe. Yeah. So so so yeah, it's been a labor of love. We're both doing this as sort of our side gigs, but so for the people to know who's on on stage. And before we before we jump into that, so we've done about 200 pieces of content, but
For all intents and purposes, Signal and Noise really started a year ago. This is our year anniversary. We were in the basement, the bowels of the the old Manhattan mall for ad week. We were filming in the corner of the press room, terrorizing the poor journos, right? We were vlogging them. With an iPhone and
You're trying, right? But you know, not good influences. Like they were they were not happy, but so w from the bowels of the press room and at Ad Week to this incredible spot. Yeah MCM. Mike Cenzanero grew up together. Incredible spots. Yeah, so thank you to MCM Studios, right? We got the John Hamm is upstairs, by the way, on the fourth floor. John Bon Jovi was there yesterday, we saw the back of his head. Brian Williams podcast is happening at the same time, so maybe maybe signal noise is like
Turned a leaf, right? We're we're we're over the hump. We may lose the live studio audience if you keep dropping names. Yeah, yeah, yeah. So so Matt Krebsick, the CEO. They're gone now. Yeah, we're gonna we've lost the audience completely. They're chasing John Hammond up this way. Yeah, exactly. Like, my God. Right? So Matt Krebsick, CEO of Media Radar, him and I go way back, he was at Nielsen for 16 years, right? Right? Decent bit, yeah. And then quotients, you're almost a lifer at Nielsen, right?
Miles (02:31.054)
For becoming the CEO of Quotion and now the CEO of Ra Media Radar. great guy. We we interviewed at we interviewed you at Cannes at the Hearst House in Cannes, which was a great conversation. And then Vasily Bacopolis. Vass. Vaz. First time? Yeah. Yeah, yeah. I wanted to say it correctly. And he's from MMA Global. He usually screws up every guy's everyone's name. Yeah, like this. And he's worked with Greg Stewart at the MMA Global for those.
community and marketing organization associated. And that that says a lot. Another great mind in the industry and we had a ton of fun. He's certainly provocative when he wants to be. Yeah. And you guys do a ton of great research for the industry. Yeah. Yep. Yep. And so yeah and so today we're we're here to talk about really well real why don't you introduce the theme and the topic right from a well the th Sigil Noise all of our episodes and all of our discussions are thematic.
Right, so we had to continue the tradition. So the theme of this is AI intelligence and competitive advantage. I know that and to to further elaborate, the synopsis, how can marketers turn widely available AI, when everyone has access to the same models, right? Is it really a competitive advantage at a certain point? How can they turn these models into a competitive advantage? And so we're gonna take a look at a study that Media Radar did recently. Into the subject we're gonna talk about it, we're gonna unpack some of its findings.
And just go from there. So normal signal noise discussion, this will be conversational, right? Yeah. Yeah. And we'll see where it goes. So question number one, Matt, this is for you. Among 500 industry professionals surveyed in July 2026, 87% said to actively use AI and a in and ad and media decisions, and 92 percent reported changes to product, services, or pricing. Where is adoption, in your opinion, translating into actual better decisions?
Like that marketers are making or that platforms are making. Can you elaborate on that? Yeah, I mean, I what I would say is maybe from our perspective or from my perspective, we're in that experimentation phase. I think they are there are certain decisions that marketers and brands are making to adapt.
Miles (04:45.218)
very, very quickly, or platforms. If you think about the success of PMAX to kind of improve like the the targeting algorithms. Yep. I think a lot of the productive d decision making so far has been gains around efficiencies. Right? We've just we've been chipping away at the tech tax, we've been maximizing more of our CPMs, getting more like bang for the buck is probably where that's been. And that's been a progressive trend in our industry of just like
How do we tune to kind of get smarter and sharper with our ad spend? I think what's been interesting, if you look at that same study, while like everybody's utilizing AI, everybody's super afraid of AI, right? In the sense of like it's afraid they're gonna lose their jobs or afraid that they're just
Don't even know it's like they like the idea of it. I love the idea. I can talk to my agent, it'll give me some research and tell me some answers. But then when I read it, I'm like, is that true? Yeah, yeah. Or am I or am I outsourcing my own agency? Am I outsourcing my own agency? And so it's like this it's this interesting paradigm of like I kinda like the idea of it. It's sh demonstrated some like really nice productivity gains, but how do I build confidence?
Yeah. In the underlying recommendations and decisions and like the actions and like do I trust the data, do I trust the information? And and how do I ensure it's not just querying Reddit and it's somebody's opinion? Reddit Reddit and Wikipedia from a source of How much of the distrust is like based on perceptions that were formed a year ago when the models are much less accurate, right? Because they've gotten better. And people have created better harnesses.
And we're able to like get much better outputs and tweak them. And you still have to be careful. I I think you still have to be careful. I think the experience has gotten cr incredibly better. Like you feel like you can have a conversation with your agent or with a chat bot and you feel like this was like a normal conversation with Brett on a Tuesday. But then the difference is
Miles (06:53.422)
It's like where do you be less obnoxious? Maybe slightly less verbose or maybe equally verbose. Definitely equally verbose. Maybe just a problem with Claude. Slightly less Slightly more compliments. Like that's a great idea. Yeah. A little sycophantic, right? A little sycophantic. But but the the difference is is like can you trust the the actual like substance of the data? Where did the source come from? And I think something that we're all learning today.
And like weirdly, I was on an investor panel panel earlier today, and I'm sitting there talking to talking on a panel about AI and how we're applying it in different industries. There was a gentleman who's the CEO of a la a legal AI company. Yep. And what was fascinating is like everybody is like really leaning into this quite aggressively of like this isn't a great enabler for us to go faster, bigger, bolder, do more, be more efficient. But in the legal world,
Where do these sources come from? And so they were com they did a comparison that was showing, like, hey, if I get legal advice from a curated view of lawyers of knowledge versus just an open LLM Claude, is the example. Sure. Claude went out and there was sourcing information and gave contradictory legal advice.
To what the actual laws were in that country. Raise your hand if you if you had that experience, like at some point, right? Yeah, yeah. For sure, for sure. I th I think I might be taking too much legal advice. Yeah, right. So one, like don't take legal advice from your agent, but like it highlights something that I think is really, really important for all of us in our industry, which is like, how do you I love the experience and the relationship you can have there and the efficiencies and the productivity you can drive.
But how do you ensure it's built upon strong, robust knowledge and data that you can trust? Yeah. And it's like how do you pair those two kind of components together of saying the experience and access to information is absolutely amazing and the speed at which you can access it at its fingertips br absolutely brilliant, but how do I plug it into and can curate the data sources and the information sources are the most trustworthy. Wait, l let me add to that because you took too much.
Miles (09:11.47)
I haven't said a word. Point point is. This is gonna be a long podcast. Completely agree with with what Matt was saying. And I think we have to differentiate between AI as a technology that keeps surprising us, right? We were surprised three years ago that this thing can talk back to you, right? Yeah. And that you're surprised that this thing can go
surf the internet and then now you're surprised that it can do things, right? Well pass the touring test, right? Pass the touring test. So we're surprised by that. And as the frontier is moving, we keep kind of putting the thing that, yeah, but it hallucinates. Yeah, but a lot of the information is crappy. And we kind of keep kind of putting that under the rug. And there was a piece of information in the study that you did that was a huge red flag. And it has nothing to do with the research. The research's kind of
Communicates what happens, but it showed that 92% of people, of marketers, were saying that, well, you know, I can trust AI or I know enough to be able to trust. You don't know nothing, right? You can't really trust, and I think very, very few people are at the point that they can actually have the guardrails and can do that. So the point here is the differentiation between AI as a technology and what we now call AI.
Intelligence, which is a funny term. We're just talking about that before. It's not even a term, right? AI intelligence is like artificial intelligence intelligence. It's like a double on Do they cancel each other out? But if one of those things is zero, the data, then the sum of this or the product of that is gonna be zero, right? So back to point A, data matters, like where you get the data matters. Like AI is
Pleasing people, obviously, sick of funding, all that. And all the post training is designed actually to have you kind of provide safe answers that are plausible. Not necessarily true, but plausible. Yeah, yeah. So I mean data sovereignty is a big topic that's talked about, especially on the brand advertising side, where they can can they have a level of governance and control over their first party data assets as well as their partner third, second party data assets.
Miles (11:26.708)
In order to train their AI, which hopefully is going to be with the topic of sort of a gentic sovereignty, something that they manage within their own walls, right? To create some sort of brand IP. I mean, how critical is it for, you know, my question is sort of how critical is it for brands to really build out sort of a like a sovereign strategy around all of these? The study said like forty four point nine percent said verified proprietary data, right? So like like would would make them have like
Right. Increased confidence. It makes sense to me. Yeah. It makes sense that way. Proprietary AI as well. Well, I so I here's what I think. I think we all like I think we st we we need a new definition or we need a new language for this world because I think if you think about AI, it's what do one of colleagues she she cho coined this term like fat words in the sense of like it's a word that encompasses like fifty five different meanings for things. Yep. I do think AI as we all throw it around is a really is becoming a really, really fat word.
that it it encompasses a lot. And I think if you break that down, there's like different components to AI. There's the models in the world that we're all learning now, there's like frontier models, and then there's like open source models, which are really just probability scores on characters and tokens, right? Right. That's all a model is. And then what you're also we're also learning is that there's things that I think brands need to own, which is what I call more at the for lack of a better term, like the harness level, which is like
Where do I embed skills and memory? Yep. And I tend to think about that harness level as like the spine. The the spine or the control board for the the gentleman who's running the the podcast right now and all the video content and sound. The context layer? That that it's where you can put the controls and guardrails in, right? So you think about I may have a model that will run a query against. And so I'm gonna have data that I have access to. I have
models that are gonna produce probabilities. Yep. I'm gonna have skills that I've built for my team. And those skills actually say, please don't query Reddit for information. Yeah, that's your governance layer how it's gonna work. Those skills often become like the governance layer for a brand or an agency or even a publisher to say, when we wanna source this type of information anywhere in our ecosystem, it has to come from
Miles (13:52.61)
verified source A, B, C, or Z. Yep. Do not surf the open web and see what somebody throws out in an opinion board. Yeah. But I think that harness level is going to be income incredibly important for brands and companies. For brand IP, for brand differentiation. And for brand differentiation. That's what you can own because your harness becomes yours. You can tap into any model underneath that harness, whatever's cheaper for you or how you want to protect your data.
From briefing to creative production to let's say bid management, like let's the whole like the whole process end to end. So at Amazon Unboxed last week, one of the one of the keynotes, they they had the guy from from Dentsu up there. Yeah. He's talking about basically how
It was kind of the Zuck's vision from a couple years ago. you just like tell us your budget, tell us what you want to sell, and we're gonna do everything, right? And then pressure button. That was kind of what they were saying that they're you know, that what was formerly the artist formerly known as their DSP is gonna do, right? And he was kind of saying like agency people are gonna be out of a job is kind of what he was implying. I actually thought it was kind of an awkward presentation, to be honest. Like in, you know, like reading the hostage, like reading this hostage video almost seemed like. But but you know, that that aside, what in order to do that, let's say that is the end goal.
Like in order to do that, like what confidence will we need? 'Cause these are black box models, right? To your point. I mean like like black box thirty. And they change, by the way. What Matt was saying before, it's ev you put all your guardrails, you put everything. Now the model changes and the way that it's gonna interact with that is gonna change as well. Which is the very scary piece that you need to be able to just check that and check that again and check that again. In fact, when it like
Like the hallucination because I also also actually think that's a bad term. It's telling you what it w thinks you want to hear and it'll make it up if it thinks that it doesn't have the right thing to tell you. Like I I remember I I asked it for a bunch of for cr to create a slide, I asked it for three data points. And the third one, I was like, This is perfect. This is the best. It totally validates my entire thesis. Where did you get it from?
Miles (16:02.782)
It basically told me I found three different things and I combined them and averaged them and it's like you made it up. I thought you would like it. Yeah, yeah, I thought you would like it. Yeah, it's It was a data point that doesn't exist that would support the point you wanted to make. But it supports you the point that I think you're wanting to make. So no, but you mentioned with this Dinsu example, I mean the danger of outsourcing and automating all of that to a third party that's not a glass box, it's a black box, where you don't know why it's decisioning, and you're also outsourcing your IP as a
brand. Like shouldn't that pattern to to the player who's selling the inventory. Like the when you think about incentives, like like the incentive compatibility of that is like, wait a second. It's the platform it owns it's owned and operated inventory. It has its own mo its own model it's using for everything, right? And then you're gonna trust and it has its own reporting as well, right? So which I mean
Yeah, and th and those that model, that predictive analytics that it's developing is training all of your competitors as well. Right? And so when you're when you're working with vendors, that contractual language is pretty important. You're giving them the alpha. Your alpha, you're giving it away, you know, if you're doing that, right? So back to your point about, you know, what models you're using and are you using your own and open source and all that, you have to be able to kind of, you know, really build your own IP and protect it, otherwise you're giving it away.
I every I mean the models are are I mean I think we're in a moment where I do believe the agency model changes. However, I believe the agency model changes to becoming a platform model versus an arbitrage model, right? Yeah or like a services company, right? What like and you think Well define that. Define a platform model. So so the idea is like, hey, I think Publicis like
Doing a really good job right now executing Not because like they're a better agency. They're actually building an operating system, right? They have a good pitch team too. Yeah, they've got a good pitch team as and is the operating system
Miles (18:04.238)
Yeah, that's that that's I think that was Or is it an AI veneer over a lot of outsourced resources that are in the world of media we're in the world of media advertising. Often it's a very thinly held veneer. This is not really your your bedroom or your studio living room. Yeah, let's be real, Al. It's an OBS pod podcast, right? Let's be real. There's a lot of thin veneers. Yeah, it's like the AI team is actually like nine hundred people in like a facility in you know Calcutta or something like that.
To get back like brass axe, I do think the agencies are going to have to build like an operating system and a true platform that allows their customers and their clients to really scale and differentiate and own that own unique system curated for them. Yeah. I don't think so that has to live headlessly within their own environment. Amazon, Google.
Right. and and they they have to have total sovereignty and control over that. But would agencies be willing to give that up, right? To say we're not gonna train models for you that we can't then go and resell to your competition because that's I think I think they will have to and I think also we we just did a study with interviewed sixty major CMOs and we asked them about the future of the operating model and how the work is gonna be done. Yeah. Yeah, marketing operating model, the agency or
The marketing operating model that includes the agency partners includes everyone, right? and there was a ton of stuff coming out of that, one of them being that if you think about the division of labor, all the CMOs are predicting that there's gonna be a lot of more in housing, which is not a surprise, right? There's th there's a tendency to think that now we can do a lot of more of that in-house. It might be we'll see.
Right. I'm not saying I'm just just reporting the findings here. But I have to see guy here, right? Yeah. Yeah. So I've I've seen that, I've heard that, you know. but the point here is that you know there is a tendency to just say, well, the talent mix is gonna be changing, we're gonna be hiring more senior people, we're gonna build it in-house. Some of those workflows now we're stitching them and we're kind of kind of connecting them end to end. So eventually agencies will
Miles (20:27.956)
still be around, but they have to think differently and to your point. They have to kind of think about like, yes, I mean y that's your proprietary model and I'm changing the way that I work with you. I'm changing my incentives. All of those things are gonna change, right, in a way, right? And and we just have to kind of think because there's competition from everywhere. Yeah. I mean to your point, big publishers are saying I can do this end to end. No no no sweat.
Yeah, it's like Yeah, and publishers saying we want direct connection with the brands, right? We want to remove the middlemen, all of the middlemen ad tech and ad tech tax so we can get more working media. A lot of people are salivating over that. And we can share more of our publisher data with the closed loop, exactly. I do think as part of this, you can see some of that tech tax go away. Like you don't need all of those hops. Well, go from like fifty percent to like
Yeah, I think it's gonna go. I mean, at one point you were saying it felt like it was almost like sixty to seventy cents on every dollar was being spent on like intermediaries before he actually reached as a consumer. Yeah, but like I I think it goes back, I think that goes back down to like 15, 20 cents. I do think there's like supply chain efficiencies there that kind of come to fruition. Supply chain optimization, yeah. And well and in part of I think a lot of people don't fully understand it. We had this conversation with Andrew Mole, who's here tonight. I don't know if he's in the audience at PubEx, right?
That a lot of that public rich publisher data, cohort based data, all that rich audience behavioral data and contextual data is not actually being pushed into the bid stream. It's not being pushed through to the DSPs because it would raise their cost exponentially because compute of just data passing through the pipes. And so they they they squeeze all that out, right? And if you can make more of a direct connection, arguably you're you're you've got a richer data set to work with as a brand. Yeah, there are lots of people trying to do that, trying to put put more or less contextual signals and just
First party data from publishers into that? But then the point is if you think about the end-to-end kind of workflow, starting from planning into production, into like say you do pretesting, into you know, launching, optimizing, and measurement, you know, the later part is more likely to be kind of engineered for. I mean, it's kind of been algorithmic for a while, it's closer to the closer loop. So
Miles (22:46.124)
So that that is the the starting point of machine learning and all that. And now it's going backwards into like, okay, what evidence and and intelligence can we kind of add into the creative production? Can we find kind of connect it all the way to the closer? Yeah. So whoever kind of kind of can do that integration, either the the the publisher from from the back all the way to the end, or the brand from the end all the way to the back.
I mean that is the point at the end of the day. You just need to kind of connect all those different things. The promise of the closed loop. We've heard The promise of the closed loop, yeah. And do you do you think this is something that I mean, d is in housing the solution to this, or do we think agencies are gonna help fill like brands can't hire armies of people to use? In housing trend is like the to Rio's point, it's like the classic like Marty supreming of this problem. Yeah.
It's like we just ping pong back and forth like the brand's gonna solve it. They tried to do that with media buying. we can do that in house. There's only a few core platforms we're buying from anyways. And then they realize the complexity. That is just that in and of itself is something that makes in housing of actual media buying. Maybe you can argue like the Martech, the data, a lot of will migrate in house. But when it comes to paying Google and Meta, net thirty you'll be shut down. How how quickly your brand customers, what do they pay? Net ninety?
Net ninety, if you're lucky, maybe like net one twenty. Yeah. We do in housing exercises in the clients like there's no way we're paying Google into three. Well, you're gonna get shut down. You're gonna learn to. And then the next, we're gonna just let the agencies handle that. Yeah. So that would not surprise me. So yeah, so the transformation of this agency model, you're you're seeing it's moving more towards it's a new delivery model, right? It's more of a productized solution, but it's not a SaaS platform, right? There's still a managed services component.
Mm. I think there's definitely still a services component to it. I think that you're gonna see a reframing of what I think what Vass was saying kinda quite well, I'm not gonna try to pronounce Vass's full name 'cause they'll be here the rest of the time. please try. No. appreciate it. If I could do it that anybody could do it. You were been practicing in the corner for an hour. You got an Italian accent going with it. Yeah, that was more Italian than Greek.
Miles (24:59.084)
But but the the point is I think you're gonna see a framing of like the broader end-to-end workflow. I think for many times, it's kind of funny even that this conversation, like we all hone in onto the ad tech martech side. Yeah. And we forgot about the most important part of advertising. The consumer. No. The creative. the creative. Sorry. The actual art, the messaging. Like
Yeah. The thing that we built. Yeah, pi pivot to creative intelligence. We've neglected for 15 years. Yeah, the thing that we've neglected for 15 years. The non-working. I've never met a criteria in my life that doesn't complain of for a complete lack of like a data-driven feedback loop that actually helps them make decisions. Whether they're they're the promise of real time. I mean, initially we broke off creative from media, right? Like which, yeah, and then and then I think programmatic was like the final nail.
That we're okay, we're going to like completely automate this. And I mean, not that creative has no it has no value, but it's value is diminished. Well, I I don't think it's value is diminished. I think it's just been we've focused our efforts and energy on elsewhere. And I think what I actually love about AI coming into the world of media and advertising, all of a sudden it becomes the great equalizer. And it becomes like this concept I've been thinking a lot about lately, which is like, is it the painter the painter? Right. And so when everybody has access
To the same 64 box of Crayola crayons. Yep. if you got a bunch of little kids. Yeah, you're not differentiating yourself with your crayons. You know. You're not going to differentiate with your same crayons. Or if everybody has access to the same tools, the same paintbrushes, the same paint, all of a sudden, value creation goes back to the artist, to the painter. In fact, you can argue consumers have access to better crayons than companies. Because if my $20 a month.
Chat GPT subscription is better than the one than my corporate one that I have at Omnicom off. Yeah, which might be a couple generations old. And you probably are accessing a whole bunch of But it's like the one you have as a consumer is the consumers have access to the best AI and yet AI is only getting better. So I th I think the winners in median advertising over the next, I'd love to say 18 months to two years, which will probably be like
Miles (27:15.55)
six to nine months, because I feel like pace just keeps getting faster and faster and faster, will be the ones that figure out how to differentiate as the painter. Not trying to out outwin the race for like I have a better And that's where human creativity, it's where human sort of like spark and soul comes into play, right? They like the r the original ideas, the just do it that come into play. I think that like the two most important words in in our vernacular for media advertising should be taste and craft.
Yeah. Yeah. And like I've been on this like on this on this pitch and belief now for I feel like a year. But like taste is like having an opinion on something. What I what I think some of the best brand marketers have, they have taste. That's that's how they're successful at trade. And that doesn't come from raw intelligence, which is what AI can do for us. It comes from wisdom and experience. Where is it distinctly? And perspective perspective. But perspective comes with intelligence, is what you do with it.
Yeah. Right. So we can I mean to your earlier point, you know, when we all have sixty-four crayons, i it raises the floor, you have to raise the ceiling, right? Yep. And then you have to kind of see what do you do with this. But I do think that there is a yes and point here. I think I completely agree on the part that you need more important than ever to have taste, you can scale taste.
No. You cannot scale taste. So it's important you have to be able and this is what's gonna be the differentiating factor when we all have So what Vass is saying is we all have jobs because we're focused on taste. As long as we have taste. As long as we have a job is like filling out a spreadsheet. Exactly. As long as we so they might be in trouble. But I think there's and there is as a Greek, I can say that there's a you know Greek philosophy kind of thing that says questions is all we have.
Yeah. Especially when answers are free, quences questions are all we have. So taste and what questions you ask and where you go and how you think it matters. But I do think though, if you think about creative, there's two buckets of creative in my mind. Tell me if you agree. One is the big ideas that you know, I was a planner, I was just writing briefs, and so you it's like okay, it's you're thinking about like catching lightning in a bottle, right? Yeah, yeah. And this is where more than ever, I mean you will need
Miles (29:39.064)
kind of taste and you know that's gonna be a differentiator. Then you have a bunch of other kind of bread and butter creative that you still kind of need to have there. It's coming with other, and I think your production there and your creative intelligence is going to be important here in having the right data and having the right ability to score that creative. We've done some work, interesting work that has to do with like really kind of understanding for your brand what are the drivers of creative success.
You can do that now better with AI than anything else. And you can build your own creative drivers and guidelines. And you can see the adherence of that. And if you're like, you know, like you you can put gold. So I think there's two parts, right? That you know, definitely the differentiation comes from taste, but you still like to But the permutations of of that is is what it the there's so many permutations based on cohort, based on creative type and messages. Exactly, exactly. But the simplified is gonna be better off in this, right? Because you're gonna get better, better
insights on how their creators performing. They're gonna be able to churn out different versions of it quicker. But I think and I think that what this is exposed to is a lot of the what we consider creative, at least on the UX side, was really just pixel pushing. Just pixel pushing. And it's like that production like is being automated. And so people are just doing that. I think that they're struggling now. But I for real like most like real like great designers know they're they're thriving. They love this. So that that that's why I go back and just simplify this whole part of it. I call that crap.
Mm-hmm. Right. Craft is the ability to say I can elicit like a human emotion or a response. I know how to choose which messing element goes where. Because when I meet a customer or a consumer who's doom scrolling on Instagram, that is a very different experience than a customer that I'm trying to convert the context of doom scrolling about data centers on Instagram while they're like Yeah, the context of the consumer experience. Right, but but that's all that's all craft.
And so I think in many ways, like our future and like where our industry has to evolve is around those two very simple concepts. Is this about taste and solving taste, or is this about solving craft? Yeah. And I think the winners in our industry are gonna be the ones that understand how to kind of build both. I have taste, I have an amazing taste. AI does does neither. And I have craft to deploy it. But no, AI is a AI is an enabler. It's a tool. It's an enabler, it's a tool, but it doesn't. Yeah. The other thing as well though, building on what
Miles (32:02.028)
Saying is that I think we're moving to a new model where you know we were designing creative as winner takes it all. AI have done a ton of pre-testing that says, like, okay, what is the winning creative, right? And obviously now, if you have you know AI activation and optimization, building cohorts on real time, you will see that instead of one creative, you have a frontier of creative assets that kind of ma
Work and play with different audiences and different cohorts, right? As a human, I can't play three dimensional test and think this through like in that, right? I need the machine to kind of give me to the map and all that. So that's where the machine comes in, right? Yeah, for sure. It's not gonna come with necessarily the big idea, but it's gonna give me like the signals of shit, there's ten more opportunities that can I can I can go after that I didn't know before. Yep.
Yep, yep. And it's and it's and it's sort of mapping to to all the permutations we talked about. It can be cohort permutations, it could be context permutations, where you've got so many variations. And it's not just an iterative process, but it's like how do you multiply the the original idea across all of these other fronts? And AI helps facilitate and speed up that process without taking away the human kind of judgment from the the original idea. So where to go from pixel pushing to becoming painters and and artists. Yeah, I I look at how it's impacted my job, right? I I
Enjoy it more. The stuff that annoyed the crap out of me I do less of, right? It's quicker. Like I don't have to take notes and calls anymore. I used to like and I'm a violent typer, right? I used to just I would break keyboards. I'm not kidding. I've been on one call, folks, with you. I do. it's bad, right? So I I worry for your your computer. Like I just follow you for abuse of like just but I don't type anymore. Yeah. Like it's it's it's or barely anymore, right? So
I mean v or much less than I used to, right? So i so the things that annoy me I do less of. But like but I guess like
Miles (34:00.622)
Closing this out, like based on the study, so what are some tips for let's say for a CMO or marketing leader that came out of that that you'd like to kind of leave as final thoughts here? man. Well, I want to start with it'd be interesting to hear Vass's perspective on this too. But like my first initial thought is like have a ton of empathy for your neighborhood CMO today. You know, honestly, yeah, it's like
You're like searching for a needle in a giant like haystack right now. And like that's the challenge of a CMO where they're just trying to figure it out. Because it's like on one hand, in that haystack is like this golden goose or this egg or this lightning in a bottle that like if you can find it and capture it, like winter, winter chicken dinner, like that's just
Absolutely amazing. And so I think for for CMOs, they're 100% right to say, how do you lean into the space? But lean into it with confidence and trust in your underlying kind of data. Yeah. And look at this as a great opportunity to get back to the art of taste and craft of advertising. And so at the end of that rainb at the end of that search, there's like an amazing like
Analogies I can come up with here, but like there's an amazing like needle in that haystack that's like it's worth it. Yeah. But like you gotta really make it's gonna be messy along the way, and you gotta be so careful with like the data you're capturing, how you're leaning, how what you're allowing your models to touch, and how do you kind of differentiate in that space? And maybe what what decisions you're outsourcing to your AI agent of any sort, right? Think of it as like you're like.
a a tool, it's great for brute force problems. Like it's like but it's not like a fine-tuned machine. Yeah. Yeah. You know? If you need a lot of compute, it's there for you. If you need to take notes, it's there for you. Interrogate my idea against these these particular qualities or these particular but if you're just looking for like the shortcut to give me the answer. Like I'm Yeah, you're you're gonna be you're gonna hit a median because brute force it's a median. Is is a really good point. So there was that announcement recently that
Miles (36:25.334)
One of the models had like cracks some like incredible like big math problem that about Navier Stokes. But they did it by having ten thousand. I feel like I'm the biggest writer on the stage right now. Yeah, yeah. Well, and that was a problem that was how gonna take how long for a a mathematician or a team of the It was gonna be basically to solve. It was gonna be about ten thousand years of computational brute force. From humans to humans. Yeah. It was a thousand years. How do you even wrap your head around?
Yeah, but it's it's what someone said is the Doctor Strange theory of marketing in this case or whatever. Like, you know, if you've if you've seen the the movie Doctor Strange love? No, Doctor Strange, the one that says like, you know, Avengers or whatever. It's like okay, I calculated there's a billion scenarios, we only win in one. Yeah, yeah. So yes, I mean you can do that, right? That's you know, you can actually get to that eventually. We're gonna find that one. We're gonna find that one. Through brute force, right? Through brute force. But I think
The the other scenario is that you really need to have good data, but you also have to your point craft in taste. You know, one of the work, and if we have a minute, that we did with the team from Meteorator is that we analyzed data from over the last five years and for how different sectors go to market. And it's interesting that every sector has almost like a very specific DNA and playbook in terms of the media allocation, in terms of
You know, creative choices that they make, messaging complexity, messaging complexity complexity, messaging type, volume of creative, right? It's like, okay, I can map them very, very clearly. So if you are someone in QSR and you see that, what do you do? You just say, okay, this is where I want to go, or do you say, okay, this is what I I need to know that, but I need to think about do I want to go there or not. Now we did another study with this data and with the Advertising Great Foundation.
and we found brands that have defied gravity of the last 20 years. And then we put them into this map and we just say, is there anything that we can see? What we found is that those brands specifically kind of defied gravity by defying convention and by simplifying their message.
Miles (38:36.716)
Yeah, divine defining convention is the is the key. But that's having taste. It's like I have perspective and perspective like I'm gonna do this differently and gonna chart my own course. Yeah, yeah. That's where a lot of great creative comes from the counterculture. It comes from stuff that is is pushing against what is the norm because again that's just gonna be the sameness that everybody sees. And word is the big idea. In the world that is coming with agents that are gonna be consumer agents, and then you're gonna have as a market to talk to
Market to people and market to agents. Like we used to say marketing to hearts and minds. Now it's gonna be hearts, so brand and be distinctive. And agents, right? So the agents gonna be algorithmic, you have to do something else. Yeah. But the my the the my the heart is still gonna be important. So you have to be distinctive. You need to have a clear voice, you need to have a perspective, you need to have taste. Otherwise you know data, obviously, not just yeah, be artistic. But yeah. I think it also every compute bills gonna go up, so start investing in
Yeah. Token prices are falling, right? So it's a that's good. And the complexity of the topic. Yeah, well, yeah, I've I've noticed that recently. Any model, it's like you burn through your your daily allocation pretty quickly. Yeah. So this has been terrific. We want to put a little questions from the audience. yeah, let's ask Any scared by any questions? Yeah, any questions? About the study, about what we talked about, about
How many Doomers versus kind of optimists? Who thinks AI is gonna just kill us all? Yeah, yeah. Right? Right? Who would consider themselves like I call it an apocalyptimist. For the listening and viewing audience, we have about 15 to 20 percent doomers. Yeah. Wait, but but we don't know if the rest are optimists. How many optimists? Okay, all right, okay. By the way, we have a podcast on data centers that drops Thursday.
Where he talks to s some dude in Wyoming that's actually responsible for lobbying for managing the data center rollout in a state that's just going full bore. So it's really and so he sees he deals with both sides of the fence, right? It's like and a lot of a lot of the anti-data center arguments are really coming from fear of superintelligence, fear of AI, loss of control. Yeah. Not about the industrial build-out. And it's interesting, Wyoming they have a populist right.
Miles (41:02.298)
Against data centers, whereas like in Colorado it's popular stuff. So it's interesting to hear that, right? It it reminds me of the old Ducks Unlimited case where like you saw like like all of a sudden it was this case around prairie pro potholes, which are like these ponds that like the co like the environmental lobby was like, We want to protect these ponds and then the hunting lobby is like, We want to protect these ponds because we wanna hunt ducks. And it was like this like beautiful combination of like, let's protect the environment. And it's like
Yeah, t in order in order to kill some animals. You know, so a few ducks were harmed, but like that was a really interesting combination where like you see this combination of like groups kinda come to the question in the back, is that right? Yeah.
Miles (41:49.226)
We're we're trying so yeah, so I think the comment just for the for the audience that's listening virtually when this gets post-produced, is that that there's a crayon full of boxes or a box full of crayons, not a crayon full of boxes. By the way, we're sponsored by Crayola Crayon. And there's one crayon that we're all leveraging was just the general. No, so the p his point was like, okay, we're all working with this sixty-four box of crayons. Yeah. However, what today we're doing is we're all searching for that one.
additional crayon, the sixty fifth crayon, shall we say? It's like the silver bullet of crayon. But it's like, it's my sparkly blue crayon. You know, I feel like you've got like a a a magenta pink crayon. I don't know. But like we all've got this like unique crayon that we're like we're like Yeah. Okay, they have the same sixty four, but this crayon is ours. Yeah. And I think it's an interesting paradigm. What what I would say
I think it's an interesting thought experiment is at what point because I I would argue five years ago we had like we were competing with like you had twenty-four different crayons and I had twenty-four different crayons. And so the differentiation on tools was pretty wide. I think with AI and technology and the ability to build software and access to data, though those differentiation, the unique crayons that we had as an agency are getting less and less and less.
Yeah. Yeah. And I think that's where if I were to put a premium on it, it's like, yes, I may still have that one unique crayon or those two unique crayons in that box. But what's more important in my mind is like how do I build a differentiated, how do I think about that as a platform and drive differentiation on the c on the craft of how do I apply those crayons? I I think I think this might be a paradigm shift beyond a box of crayons.
Right. I think we might be talking about something completely different in terms of of the j you theoretically speaking. Well, I mean if you take the analogy from the crayons, someone kind of invents a a pen and then someone comes with a big brush and just deletes everything. Right. So it's like the point here is like, you know, you will have different types of crayons or things to paint. The question is back to your thing, like what are you gonna paint with it? Yeah, yeah. Well we try like I recently tried to register to business in the state of Delaware and they have a backlog of
Miles (44:08.254)
Over a month. They said they've never seen this many new businesses get created. Yeah. It they shut their website down for certain days. It is three million new businesses last like this year and counting. Six million. I six six million up to year to date in the US to these companies. This is an unprecedented like like many of them will f fail. Most of them will fail, but that's okay. It's a credible structure capitalist. But I think to the to the audience member's point, like there might be a new shiny crayon out there that could break through. Yeah. And that's an interesting thing to keep
keep a pulse on is like as much as we I believe like we're shifting back to a a paradigm a a madman era if you would say. Yeah. Of like John Hamm is upstairs, right? John Ham's upstairs of of of taste and craft. I think he's filming right now actually. Yeah. With Brian Williams. Yeah. We're not lying. This is not a as we as we shift back to that madman era of taste and craft, I think there still will be like innovation on the on the paint side. Yeah. That's gonna keep us all like
interested to see like what happens next. Like where all what how quickly do we get access to it. I think the rate of adoption our industry has on new technology is wild. I don't know any other industry in our industry. In media and advertising, I don't know any other industry that adopts new technology as fast as and the common theme is that advertisers themselves are well behind where we are in terms of how they're thinking and they need there needs to be education and training to bring them along the journey. Interesting point. But like
You look at professions like roles in companies, marketing over the last twenty-five, thirty years has has adopted technology probably more so than anyone else except the except IT, right? Like other other other everyone uses tech now in in in the enterprise, but marketing is a completely like a a a technology driven occupation now. Yeah, but it does child. It does come with additional, obviously, complexity. In you know, like a year ago to years ago, when I was thinking about AI.
And and the the the classic example that people say is that, you know, when we move from calculator to Excel or Lotus Nodes, right, there was a whole other kind of spectrum of of you know analysis that you can do and then new jobs and all that. Yeah, or accountants gonna lose their jobs. Right. So so I was thinking two years ago and you know, I made a wrong prediction that that's not gonna happen here because this is the answer machine. And of course I wasn't seeing it. Now I see the complexity that to your point, you know, like, you know, there is all these things that
Miles (46:32.93)
You know, like new pieces of content and deliverables and dashboards and things and that that are being produced. You couldn't have imagined, right? You couldn't have imagined. I actually think I think the analogy in my mind is like imagine eighteen eighty America versus nineteen five America. Yeah. If you look at that comparison of like how quickly we all went from like horse and buggies to no building was over five stories.
Everything that's been invented has already been invented. Yes. And they really believe that. And this is before like electricity. Yeah, before the automobile the automobile, discovery of oil, disc kerosene, like you can just go like computer, you can go on and on and on, right? I mean like they really thought that, right? We're in another one of those moments. I d it does gonna happen. We can't imitate you can't imagine what's gonna happen. It's I think it's like an interesting moment of like, I think we're bracing for a lot of change, but what was cool about then and which I think is cool about now.
still comes back to like human creativity. Yeah. Like I deal I still go back to it's like taste and craft. Yeah. The tools will change, the field will change, what we have access to will change. But that ability to have an opinion to elicit a human emotion. Like one of my favorite like movie quotes lately has been like from Nottinghill, which is like I'm just a girl standing in front of a boy asking him to love her. Yeah. Like
That's that's really the heart and soul of advertising. Yeah, it's kind of the heart and soul of humanity. It's the heart and soul of humanity. You just wanna be loved. That's it. Right. That's a great message. Yeah. It's a great way. Yeah. We're well, I think most of us are optimists in this crowd. We're not we're we don't think we're gonna be, you know, that AI is gonna kill us all. So I think I think we can we can end on that. We might end up being batteries or something. Right. We're like ants, right? We're gonna they're just gonna AI super intelligence is gonna pave over us.
Yeah, so so thanks everybody for for joining us live in the studio. thank you thank thank you Matt and Media Radar for sponsoring this awesome global. And thank you, thank you, Vasily, right, for for joining the with MMA Global, the event. And we're gonna have a band on in a couple of minutes, Afro Andy and Funk, world class, some of the best musicians. And part of this is we we gotta we got a bit of a cu we call it communitas at signal and noise, right?
Miles (48:53.602)
This is a big community. A lot of us have known each other for a long time. You and I, him and I, the band I've known for 30 years, right? So everybody enjoy yourself. Apparently there's a Mescal sponsor outside. Yes, there is. So there's gonna be Mezcal drinks throughout the rest of the evening, and this is an ad tech crowd. So I'm sure it's gonna be kind of fun, right? So stay tuned, we're gonna clear out the room, and then the band's gonna be Band's gonna start and we'll have a good time and everyone can get some drinks. And shout out to Mike and MCM. Thank you, Mike. And thanks everybody. Thank you.
Comments