Shelly Palmer on Leading Through AI Change, Identity Threat, and Marketing’s Next Era

0:00It's January 2026, and I'm coming to you live from the Consumer Electronics Show here in Las Vegas. Honestly, the neon lights and the absolute chaos of Vegas feel like the perfect metaphor for the marketing industry right now. It's fast, it's high stakes, and the reward to risk ratio has never been higher. But while some hear only noise, I'm listening for the signals. I see an incredible moment to invent the future of marketing, and I'm betting on marketers to win big. I'm Joshua Espania, VP of AI and Marketing Strategy at Google, and this is Frontier CMO. On this show, I'm meeting the leaders, entrepreneurs, and technologists who are rewriting the marketing playbook for growth in the age of AI.

0:51My guest today is the person Fortune 500 CEOs call when they need to see around the corner. Shelley Palmer is the CEO of the Palmer Group, advisor to some of the world's largest companies, and contributor to our own Think with Google. Shelley has a thesis that hits home for me. AI isn't a tech problem. It's a human leadership challenge. If you're a CMO, growth leader, or executive trying to lead through this AI moment, And this episode is for you.

1:26Morning, Shelley. How are you? I'm doing great. I'm excited here. Excited for this conversation. Excited to be at CES. I love coming to Vegas at the start of the year. So I'm one of the weird ones in that regard. Vegas is my kind of town. Welcome, Shelley. Thank you. Shelley is the advanced media professor for Syracuse University. Professor in residence, as I understand it. I stole that title myself when I put my own new title together. So I'm in residence as well. I like it. You're the CEO of the Palmer Group. You advise many Fortune 500 companies. You are acknowledged as the person that CEOs come to when they want to see around the corner.

2:07I've heard that said. This is for others to say. LinkedIn named you the top, I've got to get this right, top voice in technology. So I don't know that methodology, but, you know, take it. It's a good thing. And then you're the best-selling author. I read your email every day. Thank you. You're the one. And your knowledge is one of the leaders on AI, on marketing, on technology. Welcome to Frontier CMO. I'm thrilled you're here. I'm happy to be here. How are you today? I'm doing great. I'm very, very excited. I'm excited for CES. I'm excited to have a live audience as we record Frontier CMO for the first time.

2:46So we're here in Las Vegas. And a few weeks ago, you and I had a little chat. And you were saying you were kind of underwhelmed about the technology that was coming to CES. And it stuck with me. I was kind of surprised to hear you say that you were underwhelmed by the technology at CES. And this is a city that is a mirage of a city. It's built in the desert, completely artificially, at a trade show where a lot of the technology is kind of vaporware. It's a mirage as well. They show stuff, but it's not real for years, if ever. And it just got me thinking through the holiday break about the mirages in marketing, the mirages in AI.

3:25And I've already had this week, people tell me that they've 10x their production of creative when their marketing campaigns through generative AI and their efficiency metrics are off the charts, thanks to AI. And it just got me thinking about, is the productivity efficiency gains in AI? is this a little bit of a mirage? Because our industry, the marketing industry, has made a lot of pretty crappy ads for decades. Are we just now making more ads at speed, at scale? Is AI actually going to make us better marketers? Or is this efficiency we're seeing, or this excitement about efficiency, kind of a distracting gap?

4:04That is a lot of question in one question. Yes, it is. So let's take these things backwards in order. CES, there are three lenses. what's happening now, what's going to ship in a year or so, and what's never going to ship. And so that's the lens you get at CES. And when I said to you that I was underwhelmed by the technology in context, it was the consumer technology that is underwhelming to me, not technology in general. Everything is an incremental change. And CES traditionally had been about aha moments and bright, shiny new object and something that, oh, did you see the dancing cow or whatever it's going to be?

4:39And, you know, who got the most press releases for being the coolest thing. And that's not what's here. The innovation here is incremental and that's still innovation. And the reason it's underwhelming is that there's nothing to see. So the level of excitement is like, oh, TVs, I know, I can tell you they're going to be thinner, bigger, brighter, like have a slightly better picture, better color, better contrast. It's like, yeah, it's going to be incrementally better. The thing that's important to understand about CES, and I'm going to get to your marketing question in a second. The thing that's important about CES is that once a Rubicon is crossed or once a path of improvement can be calculated, you can make some very, very good predictions about what technology will empower consumers at some point in the future.

5:19So you knew when the RealSense chip showed up that we were going to be able to do product placement in a smartphone in a couple of years, because now we can calculate the angle. We can literally make a 3D image with the multi-lens camera and the RealSense chip. So things like that became real obvious. To your marketing question about the mirage of AI and whether productivity is what it's chalked up to be. If absolutely all AI research and all AI advancements stopped today, as if we just put a stop to it, just put the brakes on, hypothetically, it's a thought experiment, no more advancements in AI.

5:56We're five years away from realizing the efficiencies that can be achieved right now in big corporate, because this is not in any way a technology challenge. It is a leadership challenge. And what I mean by that is, you are giving people, unless you're properly leading them, you find me a human being anywhere in the world who wants to do three years worth of work in one year and get paid for one year for three years worth of output. They're not going to do it. They're going to fold their hands. They're going to figure away the middle manager mafia or the deep state, whatever you want to call that group.

6:27They are not in any way interested in helping you build the weapons to kill them with. They're just not going to do it. If you are a good leader and you can motivate your teams and to enter or become part of a culture of continuous adaptation. And what that means very simply is we know every single morning there's gonna be new technology now. Like wake up, press release, something new. Wow, hey, that two years of coding I just did, they just made it a feature. Wow, that's a pain. Oh, wow, I've been working on this for four months. Now it's part of code. It's done. Now it's part of Gemini. It's like, wait a minute, like, nano banana, minding my own business, trying to do stuff.

7:10All of a sudden, wake up in the morning and I have persistent images and be like, wait, wait, what? We've been trying to work this out with inference and training for six months now. And you guys just made it a feature while I'm not, you know, like, oops. So you know that's coming every day. So how do you lead your team and how do you organize your business around knowing that to be an absolute fact of life? Is this what, because you do a lot of mentoring for CMOs. Yeah. So is this the main thing you're trying to teach CMOs, how to adapt to the technological change, which is coming every day? It's certainly part of it.

7:43The most important part of what we're talking about now is you as a C-suite leader are sitting with other leaders on the LT. So what do you need to do? You go see risk and compliance and say, I know your favorite word is no. You go see the general counsel. I know your favorite word is no. You go see SecOps. You say, I know your favorite word is no. I have some stuff I have to do. I've got technological opportunities I've never had as a chief marketing officer. My organization can be completely augmented and made significantly more efficient. And I'm going to get to what that means in a second.

8:16Significantly more efficient if I can adapt quickly or respond in reasonable timeframes. If you're going to take six months to vet a certain thing, I have to add to my tech stack. Six months from now, I mean, nine iterations later, we just can't function in the world at the speed the world is going. If, you know, you have to think at the speed of YouTube shorts, you can't think of the speed of long form television like that. It's just the way it is. You just described a whole bunch of other departments or other disciplines within companies. So the marketing has a problem with those functions if they're not quick enough, but that's not marketing's problem per se.

8:55Oh no, that is marketing's problem. What do marketers need to fix themselves before they go to those? I would argue that has to get fixed first. I mean, that's my argument. But if you're a marketer, there's a couple of things that you need to understand. First and foremost, agents don't work by themselves. And everyone's going to come in to try to sell you an agent of some kind. Every company, hyperscaler, foundational model builder, you are going to be barraged with agents and agentic solutions and every way you can say the word agency. And the problem is that agents require harnesses. So if you think about what exists right now, the foundational model builders have built models that have some reasoning capabilities and have some level of agency.

9:41Most can already search the web. They can do whatever they can do that feels agentic. But they're self-contained from your perspective. They know what they know. So call that the intelligence and just leave it alone for a minute. it. The harness would be everything around it. What do I need to do to wire this up? I need to give the agent context. I need to give the agent some kind of rubric for scoring. I need some kind of eval. There's a whole flywheel of things that I need in order to ensure the agent is doing its job. So once I have a model chosen for a particular task or goal, and once I've got a harness set up that's going to give me access to my data, the grounding, the pre-prompt, the system prompt, the context, like all the things it needs to have.

10:29That will become the system, whatever that system is going to be called. You'll name it yourself. Everyone's got their own name for their own system. And that empowers agents to be autonomous. Because without a model and a harness to give you a systemic way to give the agency, you're totally doomed. And here's the thing that just isn't out there. What are the permissions tables going to look like? What's the access control? What's the lifecycle management of this agent? Who's looking at it? Who's making sure that it's improved? Where's the human in the loop? So you have to reorganize your marketing department around this idea that you've got human agent pairs or human machine pairs and subject matter expertise matters more than anything ever.

11:15So the more you know, the better you're going to do this, the better you are at it, the better the agents that you are deploying are going to be because you're going to be watching them on a regular basis. And if marketers set their departments up like this, they can absolutely function no matter what tech comes up any given day. And if they don't set it up like this, it grinds to a hole. Everything you were describing there sounded like test, learn, iterate, push boundaries, build up the muscles to actually develop in AI or any function whatsoever. Are there marketers you see who are doing that well today?

11:49Are you prepared to name any names? What are some of the barriers to even just getting into this cycle? I'll go to MNDA jail if I mention names. So I can't do that. But what I can tell you is that everybody is new at this. And this is not only is it new for everybody, it's way harder. Like to say it is one thing. But there are human beings involved. And when I tell you that there are big feelings here, it's understating it in the extreme. There's a thing called identity threat. I didn't know about it. I thought I invented the phrase when I was writing a blog about it. Like, I was so excited that I came up with this thing.

12:27And then I Googled it and I was like, oh, it's been around for 200 years. Damn. It turns out identity threat is basically the fear of automation. And it goes back as far as the cotton gin where people like flipped out that they were going to lose their jobs to a machine. And so here's what's interesting. So let's say you're great at your job. I mean, really great at your job. and you start working with Gemini or Nano Banana or ChatGPT or a Claude code and you do something and it's great. And because you're really good at your job, you know this thing just killed it. And in fact, NetNew killed it in your mind.

13:05The reason is while you've read everything in your field and been in it for so many years, any foundational model has read everything ever printed, ever and seen every image ever available, you haven't. So it's just got to, even though it's a vague understanding of the universe, it's a bigger vague understanding than yours. So all of a sudden somebody will sit down and you're goofing around with Gemini and it writes a couple of paragraphs. Wow. If you are great at your job, you have the ability to recognize that that is really high level and it flips you out. Now, the first time it's a novelty.

13:41And the second time It's like, that was cool. By the fifth or sixth time you sit down and this thing's giving you output you could not have come up with and you recognize it as something you could not have come out with and you're great at your job and you know this thing is better at it than you are. This is where identity threat sinks in hard and no one will say it out loud for fear be labeled a Luddite or an AI Luddite, right? Oh, they're anti-AI. They're going to get fired. You need AI on your resume. You need to control AI. Everybody all of a sudden is like AI something. I am the director of AI for toasters.

14:12It's just like, okay, good. Whatever. Toothpicks. Yeah. I mean, whatever there's AI for. So we are in a really interesting place. All the things we're talking about become sociological and they are the most human issues. And it's funny because the tech, we've never, ever occupied the planet with alien intelligence before. So now we've got intelligence decoupled from consciousness as a partner. And some people say, well, it's just a tool. Use it as just a tool. I don't disagree with that notion. It's not a person. It's a tool set, but it's much more than that. And what makes it much more than that are two things.

14:47One, it's capability and where it's going. If we believe we're getting to artificial general intelligence or artificial super intelligence, we have a different conversation to have. So what is the role of, I guess, branding and building a brand, both in an LLM agentic world, but also with, I just don't know that I could believe that Gemini, as great as Gemini is, is going to be able to build my brand. Why would you ask it to? That's like, why would anybody want to shovel their job function off to a machine? It's like, you know, maybe if your job is shucking corn, you'd like to figure out a more efficient way to do it.

15:21Two kinds of creative in the world, inspired creative and required creative. So let's leave inspired creative completely out of this conversation. Inspired creative. The universe brings me melodies that I want to write down and they become a hit song. Inspired. Doesn't happen every minute. Required creative. I get a brand brief Thursday at two o 'clock in the afternoon and the senior executive of the ad agency says to me, we need you to be able to be in your studio at nine o 'clock tomorrow morning to record. We need to ship by five o 'clock. We need underscores for the 60, 30, 15, 23 plus seven and the five second tag.

15:55And that'd be great. Oh, thanks. That means it's got a budget. It's got to be done. It's an underscore. I've done a thousand of them. I got to get it out the door. And then I need cutdowns. So I've done the 60. I need the 30, the 23 plus seven. I need the 15 and maybe the 10 plus five, which would be like, I need all these cutdowns on the deliverable sheet. I'm not going to ask AI to write the song. I'm not going to ask AI to do any of that. I may use it to enhance my production capability. I may ask it to do some of the gentle compression on the bass drum so I don't have to reach for the knob.

16:26I may talk to my AI and have it do certain things for me, but I would absolutely let it do the cutdowns. Cutdowns, I'm doing mathematically. It's algorithmic the way you cut it down and music has real rules. It's like, okay, I can do this. So inspire creative, require creative, tool calls versus like, if you're laying off your humanity to a machine, you're a moron. Like, you can't do that. Like, that's not what it is. But there are realities in our world. Let's talk about the realities. Production, all production, 100 % of production is a factory business. It lives and dies on the cost controller.

17:03And all the technology I learned, by the way, I learned by building production tools. I wanted to be a more efficient producer and we had a lot of work and I wanted to do more and we had audio work and video work and, and we, I just wanted to be able to do it faster and better. And this is my point. You've got a budget. The budget for the project is 25 grand. I'm making this up. If you spend$26 ,000, you lose a thousand bucks. If you spend $9 ,000, you're making money. It's like, it's the margin as a producer, you're looking for the biggest possible margin you can because you don't have any control over the budget.

17:35The producer walks in and says, you got 25 grand for this. It's like, uh, okay. Now how good of a producer are you? Can you get, can you get their expectations met at a price? Required creative. Can you get their expectations met at a price? AI is all over that. And it's going to be all over that. And it, where it's failing now, it will not fail. You choose your timeline, six months, 12 months, 18 months, sometime before three years and more than five minutes, this is going to be a solved problem. What you're 100 % right about and what can never happen, and I never say never, but what can never happen is you as a human being cannot give your humanity to the machine and then blame the machine for you being an idiot.

18:18That you can't do. So if you think because you just literally don't know how to do something that you can type in, please write me war and peace and get me a Pulitzer Prize, and you think that's going to yield a Pulitzer Prize winning book. That's not the fault of the AI. You just don't know what you're doing with the tool set. But if you hand this tool set to a well-seasoned marketer who really understands that this is a productivity tool writ large, that it's governed like a synthetic employee, which means it is your employee, and you will need to figure out a way to talk to it and to constrain it and to give it rules, which are all done with a set of markdown files, basically for agents, then you can prosper.

19:01And these things can really be valuable. And I can see every single marketing department getting 20, 30 % more productive. Now, you said something earlier, I think is really, really important and we shouldn't gloss over. Does this mean we're going to be making, instead of two terrible commercials, 200 terrible commercials. You know, our BlackBerrys, to date myself, did not make us better communicators. It made us faster communicators. Yeah. So you can, you know, you could make a lot of stuff and then you can make a lot more bad stuff. I mean, you just make a lot of stuff. So whether it's good or bad, that's required or inspired.

19:43Is it going to make a lot of good inspired creative? I don't think it's ever going to make any inspired creative. So inside of Google, we've had a huge amount of conversation about taste and craft and skill. And the analogy I liked is the notion that creative directors are now more important than ever before. If you're flying on a plane, do you want to have the pilot for the plane or the CEO of United flying the plane? Right? I definitely want a pilot every time. And the same with credit, you need people who actually have taste, who have judgment to help set the machines, build the playbooks and do that work.

20:18And it's such an important thing. And I think it's being overlooked. Well, look, the problem is, and you can really, really see this when you get into it with live human beings. Most people don't create for a living. They execute other people's visions. and what the AI tools that we have now, which by the way, I have to say we have now because if there is anything that looks like AGI that happens, everything I'm about to say is nonsense. But at the moment, you have a tool where you interact, you give it a task or a goal, whether it uses agents you're conscious of or whether it just uses agents behind the scene, however it does what it does, it gives you some output.

21:02that output you will score as being relevant or irrelevant to what it is you're trying to accomplish. That's today. And for that, the higher level subject matter expert will do better than anyone else. Like the more you know, the more you get out of this. Let me put you on the spot here. Let's role play for a second here. I'm a new customer. I'm a new client for you, for the Palmer Group. I'm a CMO. I know nothing about AI. I need to get up to speed. What are three practical things as an individual, as a marketer, as a leader that I should be doing every day or every week to learn about and actually train myself about AI technologies and marketing?

21:43Personal projects are my favorite go-to. I don't care what it is. You pick a project. I, my, my project for the, my Christmas vacation, I have a home automation system in my house. And I asked a large language model with its reasoning engine full up. I won't say which one. If you go explore my wifi network and see if there are any open APIs. And it found a whole bunch. I said, well, what'd you find? And it found lighting and it found the security system and it found the cameras and found the doorbell camera and it found all the lights and it found the HVAC system and it started to report back.

22:28I'm like, hey, I hate Control 4, which is the app that the integrator uses for, you know, when you have like a lot of nonsense on your phone. And I had like six different apps. I'm like, I think let's build an app to consolidate it. And of course, this is a rabbit hole that only a geek like me could go down and stay in forever. And so within a few hours, I had a fully working app. And then of course, it's time to refine it. So I'm so in love with the idea that I can control the house without having to call somebody and do it. So every night I'm goofing around with this. Now, what does that do for me?

23:01Every time something happens where they change some code, add a feature, anything happens that would make this work better, or I run into a problem I wouldn't have run into as a thought experiment, but I am running into hands-on keyboard. So you put your hand, one of my clients did something fantastic. She had a kid in travel soccer and she decided she was snack parent and they wanted to do a rotation of like the people who bring like orange slices and, and the Dunkin' Munchkins and whatever you bring to after soccer. So what she did was she built an app that they could sign up. She got Stripe integrated to it.

23:36She built some performance marketing campaigns to get the parents to be reminded that they haven't got their snack thing going. She then built a dashboard with analytics and she was in marketing. And she says, what can I do? I said, what personal project is? And she was complaining that all the parents in the high school don't contribute properly and appropriately. So she decided to like, go for it. And this thing looks like a business. It literally looks like a business. It's got an unbelievable website. It's React over Tailwind. It's sitting up on Vercel as a front end. the back ends in Google Cloud.

24:08It's like, wait, what? And she had no idea how any of that happened or what any of it meant, but she knew it worked. And she knew exactly when, and the Stripe integration was hysterical because the parents, if they didn't want to come with snacks, could send money. Like this was an, and all she does when every time we get on a Zoom, we just talk about like how much she has learned because she's doing a personal project. My advice to everybody is choose something you really care about where you would know what the output looks like and you'd have a reason to continuously improve that project. One of my friends did something with his record collection.

24:43They all have barcodes on the back that tie it to the GraceNote database. And he decided that he was gonna catalog all of his vinyl records and CDs and videotapes in this crazy app with a crazy search capability. And so it doesn't matter what it is, any project. And what happens is if you are deeply into something then you start reading blogs and listening to stuff. All of a sudden you get deeper and deeper into it because you are doing something that you're passionate about. That will translate to work. The other way doesn't. And I would say to every marketer, I'd say to every executive, Tech Tuesdays or Tech Thursdays are a must.