AI in Advertising Codex Agency Adoption Super Bowl Ads Signal
Garett Sloane · February 8, 2026
advertisingai in marketing
**Introduction to AI Insiders and Guest Garett Sloane**
(0:05) Hello everyone. I'm David Burkowitz and welcome to another edition of AI Insiders by AI Marketers Guild by Architecture Media. I am here with an old friend, Garrett Sloane, chief tech reporter at Adage. Garrett has been I've read a lot of your work recently, but once I saw all the excitement around how you were describing agency operating systems I
**Discussing Agency Operating Systems and Super Bowl Week Changes**
(0:32) wanted to learn more about that. I know we got a lot of people in the community want to learn more about that and all these other exciting things you're seeing Super Bowl week and constantly changing crazy times Claudebot malt book week it's a thousand weeks in one welcome Garrett great to have you here thank you David I'm proud to be here for
**Garrett Sloane's New Role at Ad Age**
(0:57) the first time. I have an update breaking news I'm now the senior editor of technology and AI at Ad Age. Chief was a good title for a reporter, but now editor is in the title. I had to swap out that cushy chief looking thing, taking on even more tech coverage and putting AI in the title. On my resume now.
**Congratulations on the Promotion**
(1:24) Okay, now it's official. Congrats on that. Excited to have you break that news. As long as we get more of your coverage, whatever your title. My grandmother used to always brag because my aunt was a top person at a record label and my dad was a senior person in this hospital. My grandmother loved talking about having a
**The Relevance of Agency Operating Systems**
(1:52) daughter who no, my dad was a chief and my aunt was a president and she talked about that to no end. Senior editor is pretty good too. I'll take it. It's been crazy like you said week. We're here to talk about the OS, and that story I think was two or even three weeks ago, but it's still going to
**Ad Age Article on AI in Advertising**
(2:16) be relevant going forward. I'm going to bring this up, and I'll put the link in the chat for those who aren't subscribed to Adage. You can even get some of this stuff with the free registration. It was one of the best summaries I've seen of what's going on in the space. I love
**Origin of AI Platform Names**
(2:42) did you come up with the names yourself? Did AI come up with these names? This is amazing. I could take you behind the scenes. As I've been reporting on AI, I am using AI a lot more in my reporting, not to write these things, but to brainstorm as people say. Let me analyze some of these platforms, do some of the research, deep
**AI's Role in Brainstorming for Reporting**
(3:04) research on ChatGPT, Gemini, cross-reference them, fact-check, do all that. I was getting a vibe of each one as I was communicating with the LLM. I had to do a lot of tweaking to come up with these names, but these are my names. AI assist. I love it. So what do you
**Agency Differentiation with AI**
(3:32) what's going on here? There's so much more in here, but are these holding companies and agencies successfully differentiating with AI? That's why I did this story. Out of CES, I think we saw so many of these announcements. We've seen these platforms that the media holding
**Evolution of AI Platforms in Agencies**
(3:59) companies and agencies are launching probably for the last year or two. Publix with core AI for years. Omnicom has had Omni AI, I think it made it Omni. Dentsu's changed from Mercury as its key branding to Dentsu.connect last year. These have been coming up for a little bit, but I think we saw a
**CES and "The Machine" as a Breakout Moment**
(4:24) breakout moment around CES. Stagwell, "The Machine" was what set it off. When you get an email about "we're launching the machine," you think, "what is this? What are these platforms? What do they do?" I tried in this chart, where you see seven different categories, to compare them where they do
**Differentiating AI Services and Tools**
(4:46) differentiate, and I think they do. I think there are different reasons for each of them, different ways brands will interact with them, different marketing messages, business strategy. So I think there is some difference, but a lot are coming up with the same services and tools too. AI audiences, say, every agency needs AI audiences where you're creating personas. That's
**The Rise of Vibe Targeting**
(5:08) the hot thing to do in advertising: create target audiences. Vibe targeting, that's a term I tried to coin, tried to help coin last year after Brian O'Kelly first came up with it, but I latched onto it. So vibe targeting is a thing. All these platforms help with these new-age advertising formats.
**Unique AI Use Cases**
(5:31) Was there any one of these things where you thought, "wow, that's a use case I hadn't seen before"? I have to think back because I've done a few. I've looked at Horizon Blue in the past and done a few stories on where these are coming up with interesting "oh, wow" when I look at the
**AI Platforms Making Sense of Data**
(5:55) platform it's making sense. Different platforms make sense of data in ways that when you're not in advertising every day, you're looking at the data. These platforms are relying on sales data, sales data coming from everywhere across the country, products being sold at cash registers across the country. You're like, "Oh, wow. Surveillance
**Brands' Hesitancy to Discuss Data Usage**
(6:17) advertising is real and it's all coming into these platforms." Some of the details when I get into these systems, they do come up with interesting things. I would like more of that in my reporting too because brands are hesitant to discuss this stuff. They don't want to discuss how they're using this data. They don't want to discuss when the LLM comes up
**Credit for LLM-Generated Campaign Ideas**
(6:37) with an idea for a marketing campaign, they don't want to credit the idea to the LLM they may have brainstormed with. It's hard to get, but those are the great stories. If you hear the chocolate maker found the best idea for its ad through talking with LLM or creating audience personas, it found a target demo it never thought of before. Those details are the things I would
**Agency's Fear vs. Enthusiasm for AI**
(6:59) like to get more of. When you talk to folks on the agency side, especially at some of the major players, what's your sense of the fear-to-enthusiasm spectrum for all this? The fear of AI is definitely real. Another behind the scenes after the
**Creative's Skepticism on AI Platforms**
(7:28) story went up, it got a lot of attention on LinkedIn. I think that's where you saw it, David, and came to me. I also got a DM from someone in the agency world on the creative side who wanted to downplay how relevant these platforms are. Anonymously, I know who it is, but anonymously said, "No creative in New York even knows what
**WPP's Open Pro and Self-Serve Marketing**
(7:51) these platforms are. No one's using this stuff. It's not even real." I have some sympathy for that message. But then I talked to WPP about them developing this platform Open. They have Open Pro, a self-serve tool which could bring brands to use the agency on their own. Who do they need if they're using the platform to create
**Agency-Wide AI Adoption Push**
(8:16) their marketing? I'm talking to WPP, and they say everyone's using this stuff. Everyone in the agency is using it. No one's not enthusiastic about using it. I would be surprised. I would like to dig in a little more about the climate in agency land about this stuff because there's a message from up top to adopt AI, use it all
**Innovation vs. Efficiency in AI Messaging**
(8:43) around. I think unenthusiastic workers about that might get a stigma. As far as the messages they're trying to send to the market from these holding companies, how much of it is about innovation versus efficiency? Efficiency is pretty scary when you're working on
**Agency Restructuring and Downsizing**
(9:10) the agency side. I was a publicist for a few years, and I can't tell you the number of reorgs I was part of in just a few years at one holding company. Always pairing down. We've seen it play out. I don't cover the agency world as much as our agency reporters and what's happening Omnicom
**Mergers and Disappearance of Agency Brands**
(9:32) buying IPG. I read the headlines, and it's this classic hundred-year-old agency losing its name and now it's agency plus. With this it's RIP JWT came from the chat right now. Oh man. Has anyone in this room who's
**Fear of AI's Impact on Agency Brands**
(10:05) been on the agency side not had one of their brands on their resume disappear. I can see a few folks here where it's like, "yeah, 360i MRy, all my old brands, all my old t-shirts, they're collector's items now." There's some legitimate fear that AI isn't going to help that. But then is it
**Race to Commoditization in Creative Work**
(10:34) just a race to commoditization when we all have access to all of this? Yep. How much does the creative become the same? You see it happening in some of the commodifying tools that are, "oh, let me create a banner ad for a retail site" or "let me create a quick CTV ad with a little animation." These brands
**Blurring Brand Identities with AI Tools**
(10:58) blend together. You can hard to tell what's luxury, what's common brand. It's equalizing the look of some of these brands. So I think brands are going to have to watch out for that as they adopt these tools. With some of the innovation happening now, are you seeing some of this democratization of innovation
**Democratization of Space-Age AI Tech**
(11:29) and access now thanks to how easy it is to get what I consider space-age tech that a few years ago was unfathomable. It was unfathomable. That's a fact. It's there. No one. It was something that you knew was being worked on in the basement of Google but not available, and then it came out and is publicly available. So it is
**Brand Distinctions and Super Bowl Ads**
(11:54) a great leap, and it does democratize. I think you can see it; there are still distinctions between the brands. Pepsi has a Super Bowl commercial, say, and there's a polar bear in that commercial. I don't know if you've all seen they're taking on Coca-Cola, showing the polar bear moping around because it likes Pepsi more.
**AI's Role in Polished vs. Sloppy Ads**
(12:16) I don't know if that Pepsi polar bear is AI. I should mention Pepsi is not my brand I cover, but I could see it being AI if it's not now. At some point, AI has a hand there. I'm sure it's CGI AI. So maybe AI comes there. That ad is a mix maybe, or it's polished. You see a major brand use AI in a way that's still cinematic and polished, and then you see
**Generative AI for Brands like Kelshi**
(12:45) other brands come out, and they'll quickly create a generative AI commercial. Kelshi came out with one last year. That was a crazy one. That was leaning into the fact that AI is still a little sloppy. It's clearly not polished, but that's okay for a brand like Kelshi. So, different brands are going to have to
**AI and the Uncanny Valley for Luxury Brands**
(13:03) figure it out. If you're Prada, you're not going to mess around with your logos and icons in any way unless you have it perfectly captured. It's a good point. I was talking with someone about the movie Polar Express which brought the phrase "uncanny valley" into public consciousness and
**Future Perception of CGI as AI**
(13:28) I even had to stop myself from saying it was an AI-generated movie. Obviously that technology was not available then. With what you're saying about the polar bears, from the viewer's perspective, a few years from now, we're probably going to call all past CGI AI, right? It all
**Generational View of Non-Real Imagery**
(13:51) looks like my kid's going to call Jurassic Park AI. If it's something that isn't real, she's not even going to know if it's animatronic or CGI or AI. It's going to be, "oh, someone did something that maybe was cool." It's all AI now. I think it must be some of the platforms that
**Generative AI Production Studios**
(14:16) are coming up that aren't agencies. I hear so much about Brandtech Group and Pencil, Secret Level, a production studio. These are all companies we've covered coming up with generative AI in production, and this stuff is getting better. There's a subreddit on Reddit, "Is this AI?" Nobody can tell on a lot of
**The "Super Bowl LLM" and AI Presence**
(14:38) these. Now getting into the Super Bowl this year. Is this the AI bowl now? What are you seeing? I'll coin this with you now. Let's see how this flies. Super Bowl LLM. Can we start spreading that? I want to be the first stamp on that because Anthropic
**Anthropic's Super Bowl Marketing Strategy**
(15:03) just said it's in the game today. We had heard rumblings that they could be, and then we got some word that they're an F1 sponsor. So they're clearly ramping up marketing. Anthropic is an interesting case because they're an enterprise B2B business. What are they doing in the Super Bowl? They want to be mass adopted to consumers, obviously. That's what we've seen of
**Anthropic's Anti-Ad Stance in Super Bowl Commercial**
(15:26) their Super Bowl commercial. It's to appeal to consumers. Fittingly, they're taking on ChatGPT with that commercial, saying, "We don't have ads in our platforms." Again, it's all about ads. It's all about AI. I'm in a sweet spot here. At the same time, it's becoming trendy to make fun of
**Alexa's AI Super Bowl Spot and Anti-AI Trend**
(15:51) AI, like Dollar Shave Club did. We have a Super Bowl Alexa. He's in Alexa Plus. He's in the game too with an AI spot. Amazon's leaning on AI too. Sorry I interrupted. No, I am wondering if anyone here beyond glancing at it
**Amazon's Goodwill and Alexa AI**
(16:14) the day the story broke actually used the Alexa website for anything when they're AI-powered. I think Amazon at least earned a lot of goodwill from fans by putting Michael B. Jordan in an Alexa ad a while back. So, I guess they can stay in the game. They've done some pretty good ones so
**The Anti-AI Backlash from Brands**
(16:44) far. That backlash to AI, which I'm seeing on two fronts. There's the Dollar Shave Club that came out with their new ad that's entirely making fun of AI. Then there are all these brands, even Wikipedia and others, trying to come out and say, "we're anti-AI." It's
**"Real People" vs. AI Messaging Traction**
(17:10) going back to, "we have real models, we have real people, we're real, we don't want to use this stuff." How much traction do you think that message even gets? Especially as soon as the message gets out there and everyone's saying it. That's my Super Bowl story coming tomorrow: Super Bowl LLM.
**Brands' Strategies: Against or Criticizing AI**
(17:35) Look for that. I'm trying to look at what the strategies are. One of the topics is who are the brands that will lean against AI or criticize AI. As AI is becoming mainstream, there's an equal reaction to going another route because that becomes a strategy. "Oh, you're going this route, I'm going to go that." Pepsi, Coca-Cola uses AI instead.
**Rivalry in Anti-AI Advertising**
(18:00) Pepsi criticizes Coca-Cola for AI, or like you said with Manscaped or Dollar Shave Club. Are they competitors with Manscaped? They're both shaving brands. Let's say they're in the same category. Same category, men's grooming, but they're not direct competitors. Manscaped, if anyone
**Manscaped's Ill-Received Generative AI Character**
(18:24) recalls, a few months ago, had a generative AI launch of a new character called Mr. Balszac. I don't know if anybody saw this character, but he was I don't think he was well-received. No, he was not. I wouldn't look him up. He's not going to be making a Super Bowl appearance, I'm sure, because it was not great, but he was
**Opportunistic Anti-AI Messaging by Rivals**
(18:46) AI generated. If you're seeing Dollar Shave Club now with a very anti-AI ad, there could be a subtle dig there. These rivals are setting each other up. When one rival goes into AI hard, the other sees an equally opportunistic message to go anti-AI. So maybe it is which one wins out, but hard to see AI losing out, right?
**Advertising's Push for "Artistic, Human" Content**
(19:10) Earl, is there any more to add? I would say that it's going to be the push for "artistal," "artistic," "human," "not AI generated." That's going to be the push in terms of advertising. "Our stuff is made of real people versus assisted by AI bots." That's what you're going to differentiate yourself
**Hypocrisy in Anti-AI Stances**
(19:32) which I also think runs the risk of being hypocritical. They're like, "okay, we have real people in our ads," and then more stories come out of all the other ways their organizations are using AI. Who isn't? Who can be? It's like being an electricity-free
**The Impracticality of Being "AI-Free"**
(19:57) company. "We use smoke signals." I don't want to hire the accountant who uses pencil and paper. I think that's what will come off. I don't think TurboTax will say, "hey, we have abacuses and we're better for that for some reason." I think if we had to guess on this one, it will be about sentiment.
**Public Sentiment on AI**
(20:21) What is the temperature on what the country thinks about AI? I think it's mostly negative, which is coming probably more from TikTok than anywhere else. But TikTok is not a reflection of the general audiences. G, where are you seeing where the winds are blowing these days?
**AI's Inevitable Future**
(20:41) I'm always torn until it's settled. Everyone said, "you can't." This technology is going to be here. No matter what shakes out, in the future, we're going to be talking to an LLM to search the web or engaging an assistant to book us a hotel room in some capacity. Whichever company wins out, it will be the question.
**Garrett's Personal Use of AI in Reporting**
(21:05) Let's go down to a micro level for a bit. I think it's fascinating for someone like yourself in your role: where do you use AI? What do you like using yourself, and where do you specifically not use AI? I think media is a good example because we're always going to be the
**Caution in AI Use for News Gathering**
(21:34) most careful. The best reporters and editors will be very careful, and newsrooms will be very careful about how it's applied in images. The standards of everything in actual news gathering have to be impeccable, unimpeachable. We, the serious ones of us, live by that. But
**Avoiding Wikipedia for News Background**
(21:59) that doesn't mean I was never going to be like, "Oh, I'm not going to Wikipedia." I don't use Wikipedia ever. There's an example of a tech that I wouldn't go there and try to get information for background or context for a story. There are still some ways you are keeping it pure, but you have to use this. It's like using
**Using ChatGPT for Earnings Analysis**
(22:21) Google. I'll go to, for instance, say, "who's going to give their earnings tomorrow, Google?" They're giving earnings this week. You could take their earnings report, put it through ChatGPT, take last quarter's, put it through ChatGPT, and really do an analysis. That's where it's coming up; that's where you might be using it to run analysis and
**Gemini vs. ChatGPT for Google Earnings**
(22:45) then fact-check it. It would be very funny if you're comparing Gemini versus ChatGPT on Google's own earnings and Gemini tries to massage that. We could see how it positions it, although with Google, there's never anything bad to say; make another hundred billion dollars tomorrow.
**Personal AI Play for Super Bowl Research**
(23:09) but they were supposed to make a hundred trillion, and now it's only 99 trillion. Disappointing. Are there any things that you especially like playing with yourself? I liked for the Super Bowl, looking at, going back, doing some
**Gemini for Super Bowl Brand Mix Analysis**
(23:32) research on Gemini. "Hey, what was the mix of brands over the last five years in Super Bowl: auto, travel, retail, tech?" And "give me a chart over the last five years." It can spin up a chart that's relatively accurate. It didn't look that great. Gemini now Pro, I'm only on the free tier of Gemini, so I don't know what the enterprise is getting, but they're
**Gemini's Multimodal Capabilities**
(23:57) promising multimodal, right? Graphics, charts, images, all this stuff. So, it could spin up these graphics, which is a good way to condense information. "Show me this in a chart," and I can read it. Hence the chart we opened the podcast with, which was the chart for, "show me these agency platforms and what they are, lay out
**Notebook LM Discussion**
(24:20) the landscape for me so I can visualize what it looks like. Notebook LM someone mentions here is the one I hear all the time, and I have to get into it. I'm the proponent. I'm literally the unofficial salesman for Notebook LM. Well, you're going to have to rival Daniel for the sales job there. We're going to fight on this
**Exciting AI Trends for the Year**
(24:44) one. Educating myself on AI topics. It's gotten really good, G. What are the other trends you're excited about this year? What are things you think you're going to be covering way more now that you weren't a year ago? What will pop up this year? I think
**The Frenzy Around Ads on ChatGPT**
(25:09) obviously it's the ads on ChatGPT. We knew it was coming, but we were waiting for that announcement. You saw the frenzy that set off, unleashing all the same debates we discussed: AI versus non-AI, who needs AI, do we need ads in AI? How bad will that make AI if ads influence the responses? That sets off that debate. Also, ChatGPT, OpenAI
**OpenAI's Need for Revenue and Ad Strategy**
(25:36) needs money. It's like looking at Facebook or Google when their startups are not generating revenue. How are you going to do it? They turn to ads. In Google and Meta's cases, you're talking to hundred-billion-dollar ad businesses. Amazon turned on its ad business now, making 60-75 billion a year in ads. Can OpenAI do that?
**Standard Startup Monetization Models**
(25:59) It's a smart play though. Every startup started with a freemium model, then they started with an ad-based sponsored lower tier, then a paid subscription model for regular users, and then enterprise users. It's pretty standard at this point. This is not surprising. Absolutely. It always interests me how you
**Google Search Ads and Netflix Ad Revenue**
(26:20) stumble on the right formula. Google search ads must be super valuable. But it's also, I was reading what Netflix, they had a decent year with their ad business last year, but it was 3% of revenues. So if the Netflix ad business went away, some
**Netflix's Subscription-First Business Model**
(26:46) advertisers might miss it, but it's not like "there goes Netflix," right? That was never their business model. Their business model was the subscription model, which is what they succeeded on. They only added the ad-supporting tiers to get people who couldn't afford the regular subscription model. It's pretty straightforward if
**Netflix's Ad-Supported Tier and Crackdown on Sharing**
(27:07) you think about it. I have a sense that probably people who are submitting to Netflix or getting a new subscription or renewing their old, there's a difference that five a month versus eight a month. I think they cracked down on sharing at the same time they released their ad-supported model. Coincidence? I think not.
**Poor Ad Experience on Netflix vs. Prime TV**
(27:29) and the ad experience is awful. That is a fact. Yes. Which has to be effectively. Sure. Yes. In terms of Prime, I don't mind it on my Fire TV. I did my Cyber Monday shopping, and now I'm getting bombarded with all
**Omnipresent Advertising Across Platforms**
(27:50) the Prime ads. You wanted the Fire, and in the other room, I've got the Telly. I've got these CTV ads going on 24/7 here. Everywhere. Out of home, on your mobile, on your TikTok, on your CTV. Everywhere you're being messaged. No matter where you are, no matter where
**Poor Targeting and Irrelevant Ads**
(28:15) you look, where your eyeballs are, advertisers and brands are going to be there. My biggest problem with any of these ads isn't so much the ad experience; it's that Telly thinks I'm next in line to join ICE one of these days. I seriously The targeting is low. That's
**Opening the Floor for Questions**
(28:35) the targeting issue here. That's just targeting. I think it's or lack thereof. Fair enough. Who else has questions? We got a great bunch of people in the room here. Who's got questions for Garrett? What he's seeing, stuff he's covered? Who wants
**Tamika's Question: ChatGPT Privacy Concerns**
(28:55) to question the questioner? Tamika. Hey, Garrett. How are you? I have a question along the lines of privacy with ChatGPT and the sentiment piece, partly because with chat, people are being way more intimate with their bots than they would. I have a friend who talks to it like it is a therapist. So, are
**Intimate Chatbot Conversations and Regulatory Implications**
(29:23) there any implications from a privacy and a targeting standpoint that people share things with their chatbots that they probably wouldn't share with anyone else? Have you heard anyone thinking or talking about the privacy implications or what type of regulation might be thrust upon OpenAI from a targeting perspective?
**OpenAI's Personalization Opt-Out Strategy**
(29:48) Good. Obviously they talk about that when they launch the ads. Stratd wants to get ahead of that. "Here's how you turn off personalization," but they're being a little cute with it because you could turn off personalization. First of all, you're still collecting all that data. And they're saying the other part is, "we
**Decoupling Data Sales and Audience Targeting**
(30:08) don't sell that data to advertisers," which I think is the Meta stance. "We're not selling you the data, we're selling you the audience that data is based on." I don't think there's any way to decouple that you're targeting using this data. Personally, I wish privacy was more of a concern. It never seems to be. And
**The Sisyphean Challenge of Online Privacy**
(30:34) you can't use the internet without giving your full credentials in ways that are able to be tracked across the web at all times. It's almost a Sisyphean thing. You can't beat it. That's why privacy is tricky. Certainly there will be users concerned about it, a lot. In Europe, this is a much bigger deal. That's where it will
**Europe Setting the Tone for Privacy Regulations**
(31:01) mostly play out. I'm sure in Europe they set the tone, and then hopefully some of those regulations get adopted in the US. I share your sentiment too because someone I know very close, same thing, talking to ChatGPT about their girlfriend, about the problems, about "what do you do, what's my personality issue with this?
**Chatbot Advice and Double Standards for Privacy**
(31:25) What are you getting advice from ChatGPT about this stuff? People are... It is also, one of the things writing my Quadbot piece for tomorrow, it's great to see a lot of these privacy and security concerns come up. It's also funny that the standards we set for the companies that most of us
**Trusting Big Tech vs. New AI Companies**
(31:48) trust? Apple, Google especially? The amount of info that Google has on us across email, browsing history, search history, files, documents that Google has. People aren't too worried about that showing up on the web somewhere? You see some of these new companies
**Lack of Trust for New AI Companies**
(32:17) trying to get access to all this from scratch, and they have not earned that trust. Maybe they're not worried because they haven't seen the consequence. What is the consequence? I challenge you, Karen. What is the consequence? I don't know. If you're looking for a job and
**Potential Consequences of Data Usage**
(32:39) then they take that data and use it against you, or you're running for office, whatever, there could be all kinds of consequences. People are more trusting. I don't think a lot of people give a lot of thought to that. Let's keep the questions coming for G. Derek, you've been waiting. Let me ask. Let me go to Derek
**Derek's Question: Agency AI Use Cases**
(33:03) first because he's been waiting. Questions anyone can answer. It's cool. Derek, why don't you go for it, and Karen, we'll come back to you. Thanks, David. I wanted to ask Garrett or the others on the call, where are the larger agencies? Where
**Focus on AI Personas and Audience Targeting**
(33:20) what are the use cases for the LLM operating systems that they're building? Where are they putting the most effort and time? Is it creative media planning and buying? Where are they putting their efforts? I'm seeing a lot of AI personas, the targeting to build audiences based on data you collect. You go out in the field, you get human research, panel
**Customizing LLMs for Brand-Specific Strategies**
(33:45) data, and then you feed the LLM. I think they're trying to customize the LLMs with bespoke data for each brand. They want each brand to have their own unique place to go and brainstorm these ideas, media strategies, creative ideas. You have to train the LLMs to know the brand's colors. If you're doing generative AI, it needs to know the colors, the fonts,
**Broad AI Application Across Agency Functions**
(34:10) the style. I think across all these areas is where that is. I don't see one standing out over the other, except for the ones that are easier to grasp first. Persona AI personas are a lower lift than deeper training of AI for other purposes. Media buying is coming into play, agentic AI is coming into play. All these things they're doing that I
**Understanding Agentic AI Pitches**
(34:38) want to see the results of before I say it's done. That's what they're doing. When someone gives you an agentic AI pitch, are you seeing substance behind this? Are you able to understand what they're saying? I try to translate it in those stories. When Pubmatic comes to me and says, "Oh, we
**Agentic Trading in Programmatic Advertising**
(35:01) did the first agentic trading in programmatic." You're like, "Okay, what does that mean?" At least they put a brand to it. Using Claude, the brand uses Claude, the agency uses Claude, defines a media strategy, sends it to Pubmatic. The idea is that they don't have to go into Pubmatic's interface to execute this campaign.
**Agentic AI as an Incremental Efficiency Leap**
(35:26) That's where it is. I still think it's a basic leap of efficiency. It's not like bots are handling trillions of dollars of trading every day. It's still an incremental step. But there is a lot of tension about what is truly agentic, what is just using an LLM to give you some help. Hey, on the same vein as the agency
**Independent Agencies' Window of Opportunity**
(35:52) topic, you mentioned Horizon. A lot of your analysis or views on the hold. I'm curious if you have a thought on if independent agencies have a window of opportunity here. I say that knowing that the holding companies have acquired a lot of data assets and things like that that give them somewhat of a
**Leveling the Playing Field for Independents**
(36:24) proprietary advantage. Generally speaking, when it comes to things like creative or planning, or even some of the more nascent stuff like agentic buying, do you think there is a window of opportunity for independent agencies to level the playing field, or does that stay the same? I say that knowing that a lot of the
**Challenger Moment for Independent Agencies**
(36:51) holding company power in the past has been derived by scale and power of the network. Does that change or stay the same in your mind? No, I think you definitely have a challenger moment. I hear Brandtech raised a lot as the model that some of these are going after, where you're creating a
**AI-First Holding Companies and Vayner Media**
(37:15) holding company of different AI-first services and agencies. Brand has emerged a lot. Stagwell is doing what it does. We mentioned Horizon, Vayner Media. I don't know if they've put AI into a product as they're not trying to AI. I have people inside. They're building it.
**The Ubiquity of AI and New Opportunities**
(37:39) and they're building it. I like to say we have AI everywhere. AI is two letters. Why not? They'll have something. This is the general move that people are going, and it will open up opportunities and things we haven't even thought
**Earl's Question: Client Awareness of Agency AI Use**
(38:00) of in advertising services will emerge. Thank you. Thanks, Justin. Earl. I forgot what I was talking about. I want to confirm that G is right in terms of agencies using them either knowingly or unknowingly without knowing what's happening, and I think that clients don't know. So, if
**Ingesting Sensitive Client Data into AI Systems**
(38:34) you're going to build a comprehensive AI system, it's going to have to ingest all the different sensitive data from all the different clients they work with. I don't think they understand AI point of view on that. It's a very good point. Something we have to watch out because brands want to know, and this is
**Karen's Question: API vs. Consumer AI Privacy**
(38:56) being written into the contracts: "my data to train your AI." Awesome. Thanks, Earl. Karen. I asked Gemini and got the answer. My question was if there is a difference between API privacy and general consumer privacy, and it seems like there is. So if I have my own app and I use the API
**Data Privacy: API vs. Consumer Platforms**
(39:34) to connect to the cloud, that data seems to be more private than what you do on the consumer platform. Have you looked into that? Do you have any comments on this? You're stepping into technical territory. I'm going to stop technical
**Custom GPTs and API Data Privacy Choices**
(39:57) and legal, technical and legal. I would be interested to know more what the differentiation is. What do you mean by who's using the API where they would? For example, I have a product called Curator, and you can have your own API and create your own custom GPTs. The data that is generated there, you can then choose in your API whether you want
**Agencies Using Proprietary APIs for Data Privacy**
(40:24) that public or private. I'm assuming a lot of these agencies are also using their own API so that their data will be proprietary and private to them. For sure APIs do mediate that, I know that much. It is technical for me. The other example you were sharing is where which other
**Security of Paid vs. Free AI Conversions**
(40:48) way would it be giving data fully up? If you paid $20 for OpenAI or Gemini, that conversation would probably not be as secure. No, it just knows you're conversing with it and keeps track. But even if you have your own API, it still has a
**API Cost and Custom Access to AI**
(41:11) context memory. I think you can use that or feed it back and use it for your activities. It depends on how much you spend at the API level. The more you spend, the more custom access you get. ChatGPT for $20 a month or now for $8 a month if you use ChatGPT Go. You can't opt out of
**Opting Out of Training Data Collection**
(41:37) training data collection with the $20 Pro account. With the Plus account, you can opt out of it. Anthropic doesn't do it at all. Gemini makes it almost impossible to opt out. Anything you put into Gemini is going into the Google training machine. All of the companies have been asked this question so many times that they're very secure.
**Government Use of AI and Security Requirements**
(42:01) If the CIA and the FBI and the three little government agencies are using Anthropic and Gemini and ChatGPT for government, most likely an advertising agency will have the security requirements, and all of them have reporting. There are a couple companies that do SOC compliance and all of that, so you can get that trust report for any of the
**Walt's Question: Agency Integration**
(42:24) companies. Let's hope. Thanks, Daniel. We got more questions coming in. Let's go to Walt. Hey there. Thanks for joining us, Gareth. This is a great discussion. I'm old AF, so I've been around long enough to have started at full-service agencies. The thing that
**Lack of Integration in Agencies**
(42:46) I've seen that's been left behind is a tight integration between creative and media. First we had creative and media. We used to call data "research" before it became data, and then technology came in. Now you have four things you're juggling, and they're all splintered into so many sub-disciplines. I understand that the holding companies have to
**AI's Potential for Integrated Agency Functions**
(43:07) over-operationalize, so they don't mess it up because they have 50 people touching every campaign start to finish. Are you seeing anything in our new world of AI where we might return to really combining media, creative, data, and technology together in integrated discussions across teams versus the
**AI's Role in Converging Creative and Media**
(43:26) silo effect we've seen happen over the past 25 plus years? I feel it converges a bit with creative AI being able to go across all these and inform, using media to inform your creative and creative informing media, and this is all coming together. I would imagine that's the way it's going.
**Reality of Cross-Team Collaboration**
(43:49) Not really. If you ask the people on the individual teams if they work across the aisle, the truthful answer is not really, not to the extent they could. That's interesting because I talk to the people who are pushing this technology mostly, of course. We have ears everywhere, so not saying we
**Technology's Promise of Walls Coming Down**
(44:12) don't talk to the people inside using it too. But the people pushing the technology would say, "Oh, I can be a media buyer and get the research that a data researcher used to have to provide just by going into the AI platform." So, they're talking about the walls coming down. In reality, that could be something totally different, but the idea is, "I'm a media
**AI's Impact on Roles and Skillsets**
(44:36) buyer. I can make an ad now using these tools." Whether that happens and can you really make a good ad? David's talking about developing his own apps using Claude. That's an example where he wasn't a developer, now he is. So, I haven't read any deep articles on closer integration as a result of all this
**Identifying Leading Integrated Agencies**
(45:02) yet. I'm wondering who the first holding company or large independent is going to be to break out and kick everybody else's ass. My reporter sense, what would you look for? Who would be the best source for something like that? Because I'll go get this story. That's why I do these interview
**Importance of Great Ads vs. Data-Driven Targeting**
(45:20) these talks. We'll go write that story. Who would you like to hear from on that? There's also the bit that could be a whole other discussion here, maybe too long for what we have now, in terms of how much great ads matter versus if you've got the data, if you're vibe targeting, if you're vibe creating and
**Debating Standards for Ad Effectiveness**
(45:45) all this stuff. If it's matching the right thing to the right person, then who cares where it's running and if it's targeting who you say it is? Even if your brand messaging is perfect, if it gets someone to click and gets enough people to buy something, are some of our standards too high? This is not something I
**New Territory: Data Insights and Ad Standards**
(46:07) necessarily want to espouse, but I think these are going to be conversations and debates we're going to have to have as we get more data and have the potential to get more insights from that data. So we're talking about a lot of new territory here that we'll need to get into. Adam, what do you have for us? Very quickly, somewhat
**Adam's Question: Agency Team Morphing**
(46:29) dovetailing on the previous thing about how all these teams are going to morph. That article is great about highlighting what each holding company is doing. But I feel like I'm getting some PTSD from back in the early days of mobile and use of data and custom graphs. I was part of the early team at Jumptap, one of the mobile app
**Past "Centers of Excellence" and Dispersed Budgets**
(46:58) platforms, and WPP Digital was an investor. We worked with the early Zaxis team. Then we went over to Anallect, and I thought they were the coolest things on paper because it was a center of excellence within the holding company. Then you talk to the people at the center of excellence, and it turns out all the IO's, all the dollars, are held by all
**Future of AI Tools in Holding Companies**
(47:21) the different agencies out in the field. Garrett, putting you on the spot, do you have a crystal ball as to what happens to all these things agency holding companies are talking about? Ultimately, in theory, everything that is AI
**Cynical Take on Agency AI Development**
(47:44) meshes up with an employee, with a user, digital experience and journey. Do all these AI tools for the holding companies become a very thin layer over whatever frontier model the agency's deploying in its stack? I want the cynical take on what you're doing right now, which is not
**Cynical Outlook: Agencies vs. Big Tech and AI Ubiquity**
(48:14) a knock against your article, but what's going to happen in 18 to 60 months with these things? My cynical take on it would be that they can't take on Google or Meta in terms of what they're trying to do in some ways. These tools will become so widely used that you may not even need an agency to do some of this
**Need for Deeper Investigation into AI Platform Effectiveness**
(48:44) stuff. That would be my most cynical take. As far as their platforms, there's more digging to do. I appreciate the question because I'm writing about other platforms. I want to dive into more specifics about which sides of these platforms actually give results, or are brands actually using, or are the creatives within the
**The Role of Prompt Engineering in Agency AI Use**
(49:07) agencies actually using? So there's a lot more. And any agency or account that is actually using it, how much are they having to prompt engineer like crazy to the point where why do they need to give up? That could be something too.
**Peter's Question: Client Demand for Agency AI**
(49:22) when they could probably vibe-code up a middle layer or a series of OpenAI or Claude code skills. We'll see. That's why this story was hopefully the first, and I was happy. Great start. That's awesome. Peter, one quick question. Sure. I was wondering about the
**Client Readiness and Agency Investment in AI**
(49:45) other side of the coin: client demand. The agencies are investing heavily, at least the big holding companies. A report last week on AI and marketing showed that on the client side, there's low readiness; they're not doing much. Is this the expectation the agency is going to
**Clients' Willingness to Pay for Agency AI**
(50:10) handle this, and does the brand actually want to pay for this to make up for what they're not doing themselves? I think yes and no. They don't want to pay for it, of course, but they will, or maybe it cuts costs, and that's why they're pushing the adoption. "Hey, save us money in some capacity." Some of this stuff does cost money. Person--I hate to
**Brands Expecting Free AI Services**
(50:36) lean back on AI personas, that's on my mind, but we're hearing from agencies, "Oh, that should be table stakes and free." Brands want AI personas for free. "You should give that to us. Your market research using AI should be free." That's one area. The other area is brands are pushing adoption or forcing agencies to adopt because
**Brands Participating in the AI Hype Train**
(51:00) they want to see, "oh, we're so AI forward. We're doing it." It's a hype train, but they're participating in the hype train. Garrett, appreciate you walking us through what's hype, what's real, and all your ongoing coverage. It would be fun having you back sometime and see how things are changing.
**Closing Remarks and Future Events**
(51:22) You're welcome here anytime. This discussion, I think we had a new record with the number of comments during the chat. Gary, if you missed anything, I can send you all that so you can see the whole likely sidebar that's been brewing. This has been a lot of fun. Appreciate everyone participating. We got a lot of terrific events coming up. Enjoy the Super Bowl
**Thanks and Farewell**
(51:45) everyone. Look forward to your comments in the Slack on the AI ads. We will see you all very soon. Garrett, thanks again. This was terrific. Thank you, David. Take care, everyone.
(0:05) Hello everyone. I'm David Burkowitz and welcome to another edition of AI Insiders by AI Marketers Guild by Architecture Media. I am here with an old friend, Garrett Sloane, chief tech reporter at Adage. Garrett has been I've read a lot of your work recently, but once I saw all the excitement around how you were describing agency operating systems I
**Discussing Agency Operating Systems and Super Bowl Week Changes**
(0:32) wanted to learn more about that. I know we got a lot of people in the community want to learn more about that and all these other exciting things you're seeing Super Bowl week and constantly changing crazy times Claudebot malt book week it's a thousand weeks in one welcome Garrett great to have you here thank you David I'm proud to be here for
**Garrett Sloane's New Role at Ad Age**
(0:57) the first time. I have an update breaking news I'm now the senior editor of technology and AI at Ad Age. Chief was a good title for a reporter, but now editor is in the title. I had to swap out that cushy chief looking thing, taking on even more tech coverage and putting AI in the title. On my resume now.
**Congratulations on the Promotion**
(1:24) Okay, now it's official. Congrats on that. Excited to have you break that news. As long as we get more of your coverage, whatever your title. My grandmother used to always brag because my aunt was a top person at a record label and my dad was a senior person in this hospital. My grandmother loved talking about having a
**The Relevance of Agency Operating Systems**
(1:52) daughter who no, my dad was a chief and my aunt was a president and she talked about that to no end. Senior editor is pretty good too. I'll take it. It's been crazy like you said week. We're here to talk about the OS, and that story I think was two or even three weeks ago, but it's still going to
**Ad Age Article on AI in Advertising**
(2:16) be relevant going forward. I'm going to bring this up, and I'll put the link in the chat for those who aren't subscribed to Adage. You can even get some of this stuff with the free registration. It was one of the best summaries I've seen of what's going on in the space. I love
**Origin of AI Platform Names**
(2:42) did you come up with the names yourself? Did AI come up with these names? This is amazing. I could take you behind the scenes. As I've been reporting on AI, I am using AI a lot more in my reporting, not to write these things, but to brainstorm as people say. Let me analyze some of these platforms, do some of the research, deep
**AI's Role in Brainstorming for Reporting**
(3:04) research on ChatGPT, Gemini, cross-reference them, fact-check, do all that. I was getting a vibe of each one as I was communicating with the LLM. I had to do a lot of tweaking to come up with these names, but these are my names. AI assist. I love it. So what do you
**Agency Differentiation with AI**
(3:32) what's going on here? There's so much more in here, but are these holding companies and agencies successfully differentiating with AI? That's why I did this story. Out of CES, I think we saw so many of these announcements. We've seen these platforms that the media holding
**Evolution of AI Platforms in Agencies**
(3:59) companies and agencies are launching probably for the last year or two. Publix with core AI for years. Omnicom has had Omni AI, I think it made it Omni. Dentsu's changed from Mercury as its key branding to Dentsu.connect last year. These have been coming up for a little bit, but I think we saw a
**CES and "The Machine" as a Breakout Moment**
(4:24) breakout moment around CES. Stagwell, "The Machine" was what set it off. When you get an email about "we're launching the machine," you think, "what is this? What are these platforms? What do they do?" I tried in this chart, where you see seven different categories, to compare them where they do
**Differentiating AI Services and Tools**
(4:46) differentiate, and I think they do. I think there are different reasons for each of them, different ways brands will interact with them, different marketing messages, business strategy. So I think there is some difference, but a lot are coming up with the same services and tools too. AI audiences, say, every agency needs AI audiences where you're creating personas. That's
**The Rise of Vibe Targeting**
(5:08) the hot thing to do in advertising: create target audiences. Vibe targeting, that's a term I tried to coin, tried to help coin last year after Brian O'Kelly first came up with it, but I latched onto it. So vibe targeting is a thing. All these platforms help with these new-age advertising formats.
**Unique AI Use Cases**
(5:31) Was there any one of these things where you thought, "wow, that's a use case I hadn't seen before"? I have to think back because I've done a few. I've looked at Horizon Blue in the past and done a few stories on where these are coming up with interesting "oh, wow" when I look at the
**AI Platforms Making Sense of Data**
(5:55) platform it's making sense. Different platforms make sense of data in ways that when you're not in advertising every day, you're looking at the data. These platforms are relying on sales data, sales data coming from everywhere across the country, products being sold at cash registers across the country. You're like, "Oh, wow. Surveillance
**Brands' Hesitancy to Discuss Data Usage**
(6:17) advertising is real and it's all coming into these platforms." Some of the details when I get into these systems, they do come up with interesting things. I would like more of that in my reporting too because brands are hesitant to discuss this stuff. They don't want to discuss how they're using this data. They don't want to discuss when the LLM comes up
**Credit for LLM-Generated Campaign Ideas**
(6:37) with an idea for a marketing campaign, they don't want to credit the idea to the LLM they may have brainstormed with. It's hard to get, but those are the great stories. If you hear the chocolate maker found the best idea for its ad through talking with LLM or creating audience personas, it found a target demo it never thought of before. Those details are the things I would
**Agency's Fear vs. Enthusiasm for AI**
(6:59) like to get more of. When you talk to folks on the agency side, especially at some of the major players, what's your sense of the fear-to-enthusiasm spectrum for all this? The fear of AI is definitely real. Another behind the scenes after the
**Creative's Skepticism on AI Platforms**
(7:28) story went up, it got a lot of attention on LinkedIn. I think that's where you saw it, David, and came to me. I also got a DM from someone in the agency world on the creative side who wanted to downplay how relevant these platforms are. Anonymously, I know who it is, but anonymously said, "No creative in New York even knows what
**WPP's Open Pro and Self-Serve Marketing**
(7:51) these platforms are. No one's using this stuff. It's not even real." I have some sympathy for that message. But then I talked to WPP about them developing this platform Open. They have Open Pro, a self-serve tool which could bring brands to use the agency on their own. Who do they need if they're using the platform to create
**Agency-Wide AI Adoption Push**
(8:16) their marketing? I'm talking to WPP, and they say everyone's using this stuff. Everyone in the agency is using it. No one's not enthusiastic about using it. I would be surprised. I would like to dig in a little more about the climate in agency land about this stuff because there's a message from up top to adopt AI, use it all
**Innovation vs. Efficiency in AI Messaging**
(8:43) around. I think unenthusiastic workers about that might get a stigma. As far as the messages they're trying to send to the market from these holding companies, how much of it is about innovation versus efficiency? Efficiency is pretty scary when you're working on
**Agency Restructuring and Downsizing**
(9:10) the agency side. I was a publicist for a few years, and I can't tell you the number of reorgs I was part of in just a few years at one holding company. Always pairing down. We've seen it play out. I don't cover the agency world as much as our agency reporters and what's happening Omnicom
**Mergers and Disappearance of Agency Brands**
(9:32) buying IPG. I read the headlines, and it's this classic hundred-year-old agency losing its name and now it's agency plus. With this it's RIP JWT came from the chat right now. Oh man. Has anyone in this room who's
**Fear of AI's Impact on Agency Brands**
(10:05) been on the agency side not had one of their brands on their resume disappear. I can see a few folks here where it's like, "yeah, 360i MRy, all my old brands, all my old t-shirts, they're collector's items now." There's some legitimate fear that AI isn't going to help that. But then is it
**Race to Commoditization in Creative Work**
(10:34) just a race to commoditization when we all have access to all of this? Yep. How much does the creative become the same? You see it happening in some of the commodifying tools that are, "oh, let me create a banner ad for a retail site" or "let me create a quick CTV ad with a little animation." These brands
**Blurring Brand Identities with AI Tools**
(10:58) blend together. You can hard to tell what's luxury, what's common brand. It's equalizing the look of some of these brands. So I think brands are going to have to watch out for that as they adopt these tools. With some of the innovation happening now, are you seeing some of this democratization of innovation
**Democratization of Space-Age AI Tech**
(11:29) and access now thanks to how easy it is to get what I consider space-age tech that a few years ago was unfathomable. It was unfathomable. That's a fact. It's there. No one. It was something that you knew was being worked on in the basement of Google but not available, and then it came out and is publicly available. So it is
**Brand Distinctions and Super Bowl Ads**
(11:54) a great leap, and it does democratize. I think you can see it; there are still distinctions between the brands. Pepsi has a Super Bowl commercial, say, and there's a polar bear in that commercial. I don't know if you've all seen they're taking on Coca-Cola, showing the polar bear moping around because it likes Pepsi more.
**AI's Role in Polished vs. Sloppy Ads**
(12:16) I don't know if that Pepsi polar bear is AI. I should mention Pepsi is not my brand I cover, but I could see it being AI if it's not now. At some point, AI has a hand there. I'm sure it's CGI AI. So maybe AI comes there. That ad is a mix maybe, or it's polished. You see a major brand use AI in a way that's still cinematic and polished, and then you see
**Generative AI for Brands like Kelshi**
(12:45) other brands come out, and they'll quickly create a generative AI commercial. Kelshi came out with one last year. That was a crazy one. That was leaning into the fact that AI is still a little sloppy. It's clearly not polished, but that's okay for a brand like Kelshi. So, different brands are going to have to
**AI and the Uncanny Valley for Luxury Brands**
(13:03) figure it out. If you're Prada, you're not going to mess around with your logos and icons in any way unless you have it perfectly captured. It's a good point. I was talking with someone about the movie Polar Express which brought the phrase "uncanny valley" into public consciousness and
**Future Perception of CGI as AI**
(13:28) I even had to stop myself from saying it was an AI-generated movie. Obviously that technology was not available then. With what you're saying about the polar bears, from the viewer's perspective, a few years from now, we're probably going to call all past CGI AI, right? It all
**Generational View of Non-Real Imagery**
(13:51) looks like my kid's going to call Jurassic Park AI. If it's something that isn't real, she's not even going to know if it's animatronic or CGI or AI. It's going to be, "oh, someone did something that maybe was cool." It's all AI now. I think it must be some of the platforms that
**Generative AI Production Studios**
(14:16) are coming up that aren't agencies. I hear so much about Brandtech Group and Pencil, Secret Level, a production studio. These are all companies we've covered coming up with generative AI in production, and this stuff is getting better. There's a subreddit on Reddit, "Is this AI?" Nobody can tell on a lot of
**The "Super Bowl LLM" and AI Presence**
(14:38) these. Now getting into the Super Bowl this year. Is this the AI bowl now? What are you seeing? I'll coin this with you now. Let's see how this flies. Super Bowl LLM. Can we start spreading that? I want to be the first stamp on that because Anthropic
**Anthropic's Super Bowl Marketing Strategy**
(15:03) just said it's in the game today. We had heard rumblings that they could be, and then we got some word that they're an F1 sponsor. So they're clearly ramping up marketing. Anthropic is an interesting case because they're an enterprise B2B business. What are they doing in the Super Bowl? They want to be mass adopted to consumers, obviously. That's what we've seen of
**Anthropic's Anti-Ad Stance in Super Bowl Commercial**
(15:26) their Super Bowl commercial. It's to appeal to consumers. Fittingly, they're taking on ChatGPT with that commercial, saying, "We don't have ads in our platforms." Again, it's all about ads. It's all about AI. I'm in a sweet spot here. At the same time, it's becoming trendy to make fun of
**Alexa's AI Super Bowl Spot and Anti-AI Trend**
(15:51) AI, like Dollar Shave Club did. We have a Super Bowl Alexa. He's in Alexa Plus. He's in the game too with an AI spot. Amazon's leaning on AI too. Sorry I interrupted. No, I am wondering if anyone here beyond glancing at it
**Amazon's Goodwill and Alexa AI**
(16:14) the day the story broke actually used the Alexa website for anything when they're AI-powered. I think Amazon at least earned a lot of goodwill from fans by putting Michael B. Jordan in an Alexa ad a while back. So, I guess they can stay in the game. They've done some pretty good ones so
**The Anti-AI Backlash from Brands**
(16:44) far. That backlash to AI, which I'm seeing on two fronts. There's the Dollar Shave Club that came out with their new ad that's entirely making fun of AI. Then there are all these brands, even Wikipedia and others, trying to come out and say, "we're anti-AI." It's
**"Real People" vs. AI Messaging Traction**
(17:10) going back to, "we have real models, we have real people, we're real, we don't want to use this stuff." How much traction do you think that message even gets? Especially as soon as the message gets out there and everyone's saying it. That's my Super Bowl story coming tomorrow: Super Bowl LLM.
**Brands' Strategies: Against or Criticizing AI**
(17:35) Look for that. I'm trying to look at what the strategies are. One of the topics is who are the brands that will lean against AI or criticize AI. As AI is becoming mainstream, there's an equal reaction to going another route because that becomes a strategy. "Oh, you're going this route, I'm going to go that." Pepsi, Coca-Cola uses AI instead.
**Rivalry in Anti-AI Advertising**
(18:00) Pepsi criticizes Coca-Cola for AI, or like you said with Manscaped or Dollar Shave Club. Are they competitors with Manscaped? They're both shaving brands. Let's say they're in the same category. Same category, men's grooming, but they're not direct competitors. Manscaped, if anyone
**Manscaped's Ill-Received Generative AI Character**
(18:24) recalls, a few months ago, had a generative AI launch of a new character called Mr. Balszac. I don't know if anybody saw this character, but he was I don't think he was well-received. No, he was not. I wouldn't look him up. He's not going to be making a Super Bowl appearance, I'm sure, because it was not great, but he was
**Opportunistic Anti-AI Messaging by Rivals**
(18:46) AI generated. If you're seeing Dollar Shave Club now with a very anti-AI ad, there could be a subtle dig there. These rivals are setting each other up. When one rival goes into AI hard, the other sees an equally opportunistic message to go anti-AI. So maybe it is which one wins out, but hard to see AI losing out, right?
**Advertising's Push for "Artistic, Human" Content**
(19:10) Earl, is there any more to add? I would say that it's going to be the push for "artistal," "artistic," "human," "not AI generated." That's going to be the push in terms of advertising. "Our stuff is made of real people versus assisted by AI bots." That's what you're going to differentiate yourself
**Hypocrisy in Anti-AI Stances**
(19:32) which I also think runs the risk of being hypocritical. They're like, "okay, we have real people in our ads," and then more stories come out of all the other ways their organizations are using AI. Who isn't? Who can be? It's like being an electricity-free
**The Impracticality of Being "AI-Free"**
(19:57) company. "We use smoke signals." I don't want to hire the accountant who uses pencil and paper. I think that's what will come off. I don't think TurboTax will say, "hey, we have abacuses and we're better for that for some reason." I think if we had to guess on this one, it will be about sentiment.
**Public Sentiment on AI**
(20:21) What is the temperature on what the country thinks about AI? I think it's mostly negative, which is coming probably more from TikTok than anywhere else. But TikTok is not a reflection of the general audiences. G, where are you seeing where the winds are blowing these days?
**AI's Inevitable Future**
(20:41) I'm always torn until it's settled. Everyone said, "you can't." This technology is going to be here. No matter what shakes out, in the future, we're going to be talking to an LLM to search the web or engaging an assistant to book us a hotel room in some capacity. Whichever company wins out, it will be the question.
**Garrett's Personal Use of AI in Reporting**
(21:05) Let's go down to a micro level for a bit. I think it's fascinating for someone like yourself in your role: where do you use AI? What do you like using yourself, and where do you specifically not use AI? I think media is a good example because we're always going to be the
**Caution in AI Use for News Gathering**
(21:34) most careful. The best reporters and editors will be very careful, and newsrooms will be very careful about how it's applied in images. The standards of everything in actual news gathering have to be impeccable, unimpeachable. We, the serious ones of us, live by that. But
**Avoiding Wikipedia for News Background**
(21:59) that doesn't mean I was never going to be like, "Oh, I'm not going to Wikipedia." I don't use Wikipedia ever. There's an example of a tech that I wouldn't go there and try to get information for background or context for a story. There are still some ways you are keeping it pure, but you have to use this. It's like using
**Using ChatGPT for Earnings Analysis**
(22:21) Google. I'll go to, for instance, say, "who's going to give their earnings tomorrow, Google?" They're giving earnings this week. You could take their earnings report, put it through ChatGPT, take last quarter's, put it through ChatGPT, and really do an analysis. That's where it's coming up; that's where you might be using it to run analysis and
**Gemini vs. ChatGPT for Google Earnings**
(22:45) then fact-check it. It would be very funny if you're comparing Gemini versus ChatGPT on Google's own earnings and Gemini tries to massage that. We could see how it positions it, although with Google, there's never anything bad to say; make another hundred billion dollars tomorrow.
**Personal AI Play for Super Bowl Research**
(23:09) but they were supposed to make a hundred trillion, and now it's only 99 trillion. Disappointing. Are there any things that you especially like playing with yourself? I liked for the Super Bowl, looking at, going back, doing some
**Gemini for Super Bowl Brand Mix Analysis**
(23:32) research on Gemini. "Hey, what was the mix of brands over the last five years in Super Bowl: auto, travel, retail, tech?" And "give me a chart over the last five years." It can spin up a chart that's relatively accurate. It didn't look that great. Gemini now Pro, I'm only on the free tier of Gemini, so I don't know what the enterprise is getting, but they're
**Gemini's Multimodal Capabilities**
(23:57) promising multimodal, right? Graphics, charts, images, all this stuff. So, it could spin up these graphics, which is a good way to condense information. "Show me this in a chart," and I can read it. Hence the chart we opened the podcast with, which was the chart for, "show me these agency platforms and what they are, lay out
**Notebook LM Discussion**
(24:20) the landscape for me so I can visualize what it looks like. Notebook LM someone mentions here is the one I hear all the time, and I have to get into it. I'm the proponent. I'm literally the unofficial salesman for Notebook LM. Well, you're going to have to rival Daniel for the sales job there. We're going to fight on this
**Exciting AI Trends for the Year**
(24:44) one. Educating myself on AI topics. It's gotten really good, G. What are the other trends you're excited about this year? What are things you think you're going to be covering way more now that you weren't a year ago? What will pop up this year? I think
**The Frenzy Around Ads on ChatGPT**
(25:09) obviously it's the ads on ChatGPT. We knew it was coming, but we were waiting for that announcement. You saw the frenzy that set off, unleashing all the same debates we discussed: AI versus non-AI, who needs AI, do we need ads in AI? How bad will that make AI if ads influence the responses? That sets off that debate. Also, ChatGPT, OpenAI
**OpenAI's Need for Revenue and Ad Strategy**
(25:36) needs money. It's like looking at Facebook or Google when their startups are not generating revenue. How are you going to do it? They turn to ads. In Google and Meta's cases, you're talking to hundred-billion-dollar ad businesses. Amazon turned on its ad business now, making 60-75 billion a year in ads. Can OpenAI do that?
**Standard Startup Monetization Models**
(25:59) It's a smart play though. Every startup started with a freemium model, then they started with an ad-based sponsored lower tier, then a paid subscription model for regular users, and then enterprise users. It's pretty standard at this point. This is not surprising. Absolutely. It always interests me how you
**Google Search Ads and Netflix Ad Revenue**
(26:20) stumble on the right formula. Google search ads must be super valuable. But it's also, I was reading what Netflix, they had a decent year with their ad business last year, but it was 3% of revenues. So if the Netflix ad business went away, some
**Netflix's Subscription-First Business Model**
(26:46) advertisers might miss it, but it's not like "there goes Netflix," right? That was never their business model. Their business model was the subscription model, which is what they succeeded on. They only added the ad-supporting tiers to get people who couldn't afford the regular subscription model. It's pretty straightforward if
**Netflix's Ad-Supported Tier and Crackdown on Sharing**
(27:07) you think about it. I have a sense that probably people who are submitting to Netflix or getting a new subscription or renewing their old, there's a difference that five a month versus eight a month. I think they cracked down on sharing at the same time they released their ad-supported model. Coincidence? I think not.
**Poor Ad Experience on Netflix vs. Prime TV**
(27:29) and the ad experience is awful. That is a fact. Yes. Which has to be effectively. Sure. Yes. In terms of Prime, I don't mind it on my Fire TV. I did my Cyber Monday shopping, and now I'm getting bombarded with all
**Omnipresent Advertising Across Platforms**
(27:50) the Prime ads. You wanted the Fire, and in the other room, I've got the Telly. I've got these CTV ads going on 24/7 here. Everywhere. Out of home, on your mobile, on your TikTok, on your CTV. Everywhere you're being messaged. No matter where you are, no matter where
**Poor Targeting and Irrelevant Ads**
(28:15) you look, where your eyeballs are, advertisers and brands are going to be there. My biggest problem with any of these ads isn't so much the ad experience; it's that Telly thinks I'm next in line to join ICE one of these days. I seriously The targeting is low. That's
**Opening the Floor for Questions**
(28:35) the targeting issue here. That's just targeting. I think it's or lack thereof. Fair enough. Who else has questions? We got a great bunch of people in the room here. Who's got questions for Garrett? What he's seeing, stuff he's covered? Who wants
**Tamika's Question: ChatGPT Privacy Concerns**
(28:55) to question the questioner? Tamika. Hey, Garrett. How are you? I have a question along the lines of privacy with ChatGPT and the sentiment piece, partly because with chat, people are being way more intimate with their bots than they would. I have a friend who talks to it like it is a therapist. So, are
**Intimate Chatbot Conversations and Regulatory Implications**
(29:23) there any implications from a privacy and a targeting standpoint that people share things with their chatbots that they probably wouldn't share with anyone else? Have you heard anyone thinking or talking about the privacy implications or what type of regulation might be thrust upon OpenAI from a targeting perspective?
**OpenAI's Personalization Opt-Out Strategy**
(29:48) Good. Obviously they talk about that when they launch the ads. Stratd wants to get ahead of that. "Here's how you turn off personalization," but they're being a little cute with it because you could turn off personalization. First of all, you're still collecting all that data. And they're saying the other part is, "we
**Decoupling Data Sales and Audience Targeting**
(30:08) don't sell that data to advertisers," which I think is the Meta stance. "We're not selling you the data, we're selling you the audience that data is based on." I don't think there's any way to decouple that you're targeting using this data. Personally, I wish privacy was more of a concern. It never seems to be. And
**The Sisyphean Challenge of Online Privacy**
(30:34) you can't use the internet without giving your full credentials in ways that are able to be tracked across the web at all times. It's almost a Sisyphean thing. You can't beat it. That's why privacy is tricky. Certainly there will be users concerned about it, a lot. In Europe, this is a much bigger deal. That's where it will
**Europe Setting the Tone for Privacy Regulations**
(31:01) mostly play out. I'm sure in Europe they set the tone, and then hopefully some of those regulations get adopted in the US. I share your sentiment too because someone I know very close, same thing, talking to ChatGPT about their girlfriend, about the problems, about "what do you do, what's my personality issue with this?
**Chatbot Advice and Double Standards for Privacy**
(31:25) What are you getting advice from ChatGPT about this stuff? People are... It is also, one of the things writing my Quadbot piece for tomorrow, it's great to see a lot of these privacy and security concerns come up. It's also funny that the standards we set for the companies that most of us
**Trusting Big Tech vs. New AI Companies**
(31:48) trust? Apple, Google especially? The amount of info that Google has on us across email, browsing history, search history, files, documents that Google has. People aren't too worried about that showing up on the web somewhere? You see some of these new companies
**Lack of Trust for New AI Companies**
(32:17) trying to get access to all this from scratch, and they have not earned that trust. Maybe they're not worried because they haven't seen the consequence. What is the consequence? I challenge you, Karen. What is the consequence? I don't know. If you're looking for a job and
**Potential Consequences of Data Usage**
(32:39) then they take that data and use it against you, or you're running for office, whatever, there could be all kinds of consequences. People are more trusting. I don't think a lot of people give a lot of thought to that. Let's keep the questions coming for G. Derek, you've been waiting. Let me ask. Let me go to Derek
**Derek's Question: Agency AI Use Cases**
(33:03) first because he's been waiting. Questions anyone can answer. It's cool. Derek, why don't you go for it, and Karen, we'll come back to you. Thanks, David. I wanted to ask Garrett or the others on the call, where are the larger agencies? Where
**Focus on AI Personas and Audience Targeting**
(33:20) what are the use cases for the LLM operating systems that they're building? Where are they putting the most effort and time? Is it creative media planning and buying? Where are they putting their efforts? I'm seeing a lot of AI personas, the targeting to build audiences based on data you collect. You go out in the field, you get human research, panel
**Customizing LLMs for Brand-Specific Strategies**
(33:45) data, and then you feed the LLM. I think they're trying to customize the LLMs with bespoke data for each brand. They want each brand to have their own unique place to go and brainstorm these ideas, media strategies, creative ideas. You have to train the LLMs to know the brand's colors. If you're doing generative AI, it needs to know the colors, the fonts,
**Broad AI Application Across Agency Functions**
(34:10) the style. I think across all these areas is where that is. I don't see one standing out over the other, except for the ones that are easier to grasp first. Persona AI personas are a lower lift than deeper training of AI for other purposes. Media buying is coming into play, agentic AI is coming into play. All these things they're doing that I
**Understanding Agentic AI Pitches**
(34:38) want to see the results of before I say it's done. That's what they're doing. When someone gives you an agentic AI pitch, are you seeing substance behind this? Are you able to understand what they're saying? I try to translate it in those stories. When Pubmatic comes to me and says, "Oh, we
**Agentic Trading in Programmatic Advertising**
(35:01) did the first agentic trading in programmatic." You're like, "Okay, what does that mean?" At least they put a brand to it. Using Claude, the brand uses Claude, the agency uses Claude, defines a media strategy, sends it to Pubmatic. The idea is that they don't have to go into Pubmatic's interface to execute this campaign.
**Agentic AI as an Incremental Efficiency Leap**
(35:26) That's where it is. I still think it's a basic leap of efficiency. It's not like bots are handling trillions of dollars of trading every day. It's still an incremental step. But there is a lot of tension about what is truly agentic, what is just using an LLM to give you some help. Hey, on the same vein as the agency
**Independent Agencies' Window of Opportunity**
(35:52) topic, you mentioned Horizon. A lot of your analysis or views on the hold. I'm curious if you have a thought on if independent agencies have a window of opportunity here. I say that knowing that the holding companies have acquired a lot of data assets and things like that that give them somewhat of a
**Leveling the Playing Field for Independents**
(36:24) proprietary advantage. Generally speaking, when it comes to things like creative or planning, or even some of the more nascent stuff like agentic buying, do you think there is a window of opportunity for independent agencies to level the playing field, or does that stay the same? I say that knowing that a lot of the
**Challenger Moment for Independent Agencies**
(36:51) holding company power in the past has been derived by scale and power of the network. Does that change or stay the same in your mind? No, I think you definitely have a challenger moment. I hear Brandtech raised a lot as the model that some of these are going after, where you're creating a
**AI-First Holding Companies and Vayner Media**
(37:15) holding company of different AI-first services and agencies. Brand has emerged a lot. Stagwell is doing what it does. We mentioned Horizon, Vayner Media. I don't know if they've put AI into a product as they're not trying to AI. I have people inside. They're building it.
**The Ubiquity of AI and New Opportunities**
(37:39) and they're building it. I like to say we have AI everywhere. AI is two letters. Why not? They'll have something. This is the general move that people are going, and it will open up opportunities and things we haven't even thought
**Earl's Question: Client Awareness of Agency AI Use**
(38:00) of in advertising services will emerge. Thank you. Thanks, Justin. Earl. I forgot what I was talking about. I want to confirm that G is right in terms of agencies using them either knowingly or unknowingly without knowing what's happening, and I think that clients don't know. So, if
**Ingesting Sensitive Client Data into AI Systems**
(38:34) you're going to build a comprehensive AI system, it's going to have to ingest all the different sensitive data from all the different clients they work with. I don't think they understand AI point of view on that. It's a very good point. Something we have to watch out because brands want to know, and this is
**Karen's Question: API vs. Consumer AI Privacy**
(38:56) being written into the contracts: "my data to train your AI." Awesome. Thanks, Earl. Karen. I asked Gemini and got the answer. My question was if there is a difference between API privacy and general consumer privacy, and it seems like there is. So if I have my own app and I use the API
**Data Privacy: API vs. Consumer Platforms**
(39:34) to connect to the cloud, that data seems to be more private than what you do on the consumer platform. Have you looked into that? Do you have any comments on this? You're stepping into technical territory. I'm going to stop technical
**Custom GPTs and API Data Privacy Choices**
(39:57) and legal, technical and legal. I would be interested to know more what the differentiation is. What do you mean by who's using the API where they would? For example, I have a product called Curator, and you can have your own API and create your own custom GPTs. The data that is generated there, you can then choose in your API whether you want
**Agencies Using Proprietary APIs for Data Privacy**
(40:24) that public or private. I'm assuming a lot of these agencies are also using their own API so that their data will be proprietary and private to them. For sure APIs do mediate that, I know that much. It is technical for me. The other example you were sharing is where which other
**Security of Paid vs. Free AI Conversions**
(40:48) way would it be giving data fully up? If you paid $20 for OpenAI or Gemini, that conversation would probably not be as secure. No, it just knows you're conversing with it and keeps track. But even if you have your own API, it still has a
**API Cost and Custom Access to AI**
(41:11) context memory. I think you can use that or feed it back and use it for your activities. It depends on how much you spend at the API level. The more you spend, the more custom access you get. ChatGPT for $20 a month or now for $8 a month if you use ChatGPT Go. You can't opt out of
**Opting Out of Training Data Collection**
(41:37) training data collection with the $20 Pro account. With the Plus account, you can opt out of it. Anthropic doesn't do it at all. Gemini makes it almost impossible to opt out. Anything you put into Gemini is going into the Google training machine. All of the companies have been asked this question so many times that they're very secure.
**Government Use of AI and Security Requirements**
(42:01) If the CIA and the FBI and the three little government agencies are using Anthropic and Gemini and ChatGPT for government, most likely an advertising agency will have the security requirements, and all of them have reporting. There are a couple companies that do SOC compliance and all of that, so you can get that trust report for any of the
**Walt's Question: Agency Integration**
(42:24) companies. Let's hope. Thanks, Daniel. We got more questions coming in. Let's go to Walt. Hey there. Thanks for joining us, Gareth. This is a great discussion. I'm old AF, so I've been around long enough to have started at full-service agencies. The thing that
**Lack of Integration in Agencies**
(42:46) I've seen that's been left behind is a tight integration between creative and media. First we had creative and media. We used to call data "research" before it became data, and then technology came in. Now you have four things you're juggling, and they're all splintered into so many sub-disciplines. I understand that the holding companies have to
**AI's Potential for Integrated Agency Functions**
(43:07) over-operationalize, so they don't mess it up because they have 50 people touching every campaign start to finish. Are you seeing anything in our new world of AI where we might return to really combining media, creative, data, and technology together in integrated discussions across teams versus the
**AI's Role in Converging Creative and Media**
(43:26) silo effect we've seen happen over the past 25 plus years? I feel it converges a bit with creative AI being able to go across all these and inform, using media to inform your creative and creative informing media, and this is all coming together. I would imagine that's the way it's going.
**Reality of Cross-Team Collaboration**
(43:49) Not really. If you ask the people on the individual teams if they work across the aisle, the truthful answer is not really, not to the extent they could. That's interesting because I talk to the people who are pushing this technology mostly, of course. We have ears everywhere, so not saying we
**Technology's Promise of Walls Coming Down**
(44:12) don't talk to the people inside using it too. But the people pushing the technology would say, "Oh, I can be a media buyer and get the research that a data researcher used to have to provide just by going into the AI platform." So, they're talking about the walls coming down. In reality, that could be something totally different, but the idea is, "I'm a media
**AI's Impact on Roles and Skillsets**
(44:36) buyer. I can make an ad now using these tools." Whether that happens and can you really make a good ad? David's talking about developing his own apps using Claude. That's an example where he wasn't a developer, now he is. So, I haven't read any deep articles on closer integration as a result of all this
**Identifying Leading Integrated Agencies**
(45:02) yet. I'm wondering who the first holding company or large independent is going to be to break out and kick everybody else's ass. My reporter sense, what would you look for? Who would be the best source for something like that? Because I'll go get this story. That's why I do these interview
**Importance of Great Ads vs. Data-Driven Targeting**
(45:20) these talks. We'll go write that story. Who would you like to hear from on that? There's also the bit that could be a whole other discussion here, maybe too long for what we have now, in terms of how much great ads matter versus if you've got the data, if you're vibe targeting, if you're vibe creating and
**Debating Standards for Ad Effectiveness**
(45:45) all this stuff. If it's matching the right thing to the right person, then who cares where it's running and if it's targeting who you say it is? Even if your brand messaging is perfect, if it gets someone to click and gets enough people to buy something, are some of our standards too high? This is not something I
**New Territory: Data Insights and Ad Standards**
(46:07) necessarily want to espouse, but I think these are going to be conversations and debates we're going to have to have as we get more data and have the potential to get more insights from that data. So we're talking about a lot of new territory here that we'll need to get into. Adam, what do you have for us? Very quickly, somewhat
**Adam's Question: Agency Team Morphing**
(46:29) dovetailing on the previous thing about how all these teams are going to morph. That article is great about highlighting what each holding company is doing. But I feel like I'm getting some PTSD from back in the early days of mobile and use of data and custom graphs. I was part of the early team at Jumptap, one of the mobile app
**Past "Centers of Excellence" and Dispersed Budgets**
(46:58) platforms, and WPP Digital was an investor. We worked with the early Zaxis team. Then we went over to Anallect, and I thought they were the coolest things on paper because it was a center of excellence within the holding company. Then you talk to the people at the center of excellence, and it turns out all the IO's, all the dollars, are held by all
**Future of AI Tools in Holding Companies**
(47:21) the different agencies out in the field. Garrett, putting you on the spot, do you have a crystal ball as to what happens to all these things agency holding companies are talking about? Ultimately, in theory, everything that is AI
**Cynical Take on Agency AI Development**
(47:44) meshes up with an employee, with a user, digital experience and journey. Do all these AI tools for the holding companies become a very thin layer over whatever frontier model the agency's deploying in its stack? I want the cynical take on what you're doing right now, which is not
**Cynical Outlook: Agencies vs. Big Tech and AI Ubiquity**
(48:14) a knock against your article, but what's going to happen in 18 to 60 months with these things? My cynical take on it would be that they can't take on Google or Meta in terms of what they're trying to do in some ways. These tools will become so widely used that you may not even need an agency to do some of this
**Need for Deeper Investigation into AI Platform Effectiveness**
(48:44) stuff. That would be my most cynical take. As far as their platforms, there's more digging to do. I appreciate the question because I'm writing about other platforms. I want to dive into more specifics about which sides of these platforms actually give results, or are brands actually using, or are the creatives within the
**The Role of Prompt Engineering in Agency AI Use**
(49:07) agencies actually using? So there's a lot more. And any agency or account that is actually using it, how much are they having to prompt engineer like crazy to the point where why do they need to give up? That could be something too.
**Peter's Question: Client Demand for Agency AI**
(49:22) when they could probably vibe-code up a middle layer or a series of OpenAI or Claude code skills. We'll see. That's why this story was hopefully the first, and I was happy. Great start. That's awesome. Peter, one quick question. Sure. I was wondering about the
**Client Readiness and Agency Investment in AI**
(49:45) other side of the coin: client demand. The agencies are investing heavily, at least the big holding companies. A report last week on AI and marketing showed that on the client side, there's low readiness; they're not doing much. Is this the expectation the agency is going to
**Clients' Willingness to Pay for Agency AI**
(50:10) handle this, and does the brand actually want to pay for this to make up for what they're not doing themselves? I think yes and no. They don't want to pay for it, of course, but they will, or maybe it cuts costs, and that's why they're pushing the adoption. "Hey, save us money in some capacity." Some of this stuff does cost money. Person--I hate to
**Brands Expecting Free AI Services**
(50:36) lean back on AI personas, that's on my mind, but we're hearing from agencies, "Oh, that should be table stakes and free." Brands want AI personas for free. "You should give that to us. Your market research using AI should be free." That's one area. The other area is brands are pushing adoption or forcing agencies to adopt because
**Brands Participating in the AI Hype Train**
(51:00) they want to see, "oh, we're so AI forward. We're doing it." It's a hype train, but they're participating in the hype train. Garrett, appreciate you walking us through what's hype, what's real, and all your ongoing coverage. It would be fun having you back sometime and see how things are changing.
**Closing Remarks and Future Events**
(51:22) You're welcome here anytime. This discussion, I think we had a new record with the number of comments during the chat. Gary, if you missed anything, I can send you all that so you can see the whole likely sidebar that's been brewing. This has been a lot of fun. Appreciate everyone participating. We got a lot of terrific events coming up. Enjoy the Super Bowl
**Thanks and Farewell**
(51:45) everyone. Look forward to your comments in the Slack on the AI ads. We will see you all very soon. Garrett, thanks again. This was terrific. Thank you, David. Take care, everyone.