Mike Kaput | Structuring Your Company for AI - An Update From the Front Lines

0:00Mike Kaput:The CMO Confidential podcast is a proud member of the I Hear Everything podcast network. Looking to launch or scale your podcast? I Hear Everything delivers podcast production, growth, and monetization solutions that transform your words into profit. Ready to give your brand a voice? Then visit IHearEverything.com. Welcome to CMO Confidential, the podcast that takes you inside the drama, decisions, and choices that go with being the head of marketing. Hosted by five-time CMO Mike Linton. Typeface has completely changed the process of large campaign executions. Everyone knows AI can help with images and headlines, but the real impact is making the leap from a creative brief to a live multi-channel large campaign at speed.

0:52Typeface just announced its marketing orchestration Engine, the first platform built to automate campaign workflows. You can roll out campaigns that used to take months in just a few hours. Typeface uses agentic AI to orchestrate the entire process, taking one campaign and instantly orchestrating it into thousands of personalized experiences across ads, email, and video. Major brands like ASICS and Post Holdings are already transforming their marketing with Typeface. See how to move from brief to personalized campaigns in hours, not months at typeface.ai slash CMO. Welcome marketers, advertisers, and those who love them to Chief Marketing Officer Confidential.

1:40CMO Confidential is a program that takes you inside the drama, the decisions, and the politics that go with being the head of marketing at any company in what is one of the most scrutinized jobs in the executive suite. I'm Mike Linton, the former chief marketing officer of Best Buy, eBay, Farmers Insurance, and Ancestry.com, here today with my guest, Mike Kaput. Today's topic, structuring your company for AI, an update from the front lines. Now, Mike is the chief content officer for the Marketing AI Institute and host of the Artificial Intelligence Show, a podcast that has been running since 2021.

2:21The Marketing AI Institute, where he works, was founded 10 years ago and hosts the largest AI event for marketing conference in the U.S. right here in Cleveland, Ohio. So Mike, if anybody, has one of the very front row seats for the AI revolution, which he has occupied since well before the introduction of ChatGPT. His company recently completed a very large study on AI usage, and he's here to share the results. Welcome, Mike. Hey, Mike. Thanks for having me. Super excited. Yeah. Well, I think this is really a fun topic. And you surveyed more than 2 ,000 companies on how AI is actually being used.

3:06Tell us, why did you even undertake this study and how you went about selecting the companies? Yeah, of course. So for the last five years at Marketing AI Institute, we have run this thing called the State of Marketing AI Report. And that's basically year-on-year research where we've surveyed hundreds and then eventually a couple thousand marketing leaders and business leaders, not just marketing, but mostly marketing for that particular study. And we collected all this benchmark data over the years because for a long time, our work was solely for marketing. Marketing AI Institute, pretty self-explanatory.

3:43But in the last couple of years, we've actually expanded quite a bit. We have a parent company now called SmarterX that our CEO and founder, Paul Reitzer, started. So we're one of the brands under SmarterX. SmarterX is not just for marketers. Marketing is a huge part of what we do, but it's about helping businesses of all stripes, all functions, all departments adopt AI responsibly, because we started out as marketers. We have deep marketing backgrounds, but we found that everything we were teaching and learning and sharing in marketing was also highly, highly relevant to non-marketing functions.

4:16So long story short, we did all this - You got your talking points in there, Mike. That was as good as anybody's ever - Oh my gosh. I'm getting the talking points in the first. I mean, that is an A plus. Wow. Well, that makes me feel good about the rest of this. So, you know, just remember that when I say something dumb. Yeah. So we have been collecting this marketing data for years on AI adoption, education, barriers. And because of the expanded mission of SmarterX, we decided this year, okay, we're doing the state of AI for business report. So we've kind of taken our existing audience, our new audience, some of the adapted the questions from the last several years, added some new ones and basically said, OK, let's interview as many company or survey rather as many companies as possible.

5:05Some in marketing, some not. We actually only have about a third of its marketing this year in terms of job title, function, et cetera. But we've got now 2 ,000 plus respondents at different companies across finance, HR, legal, operations, et cetera, C-suite, all companies, all industries, a lot of different sizes. Like, you know, we're still kind of parsing the data right now. It's still an early, we just finished up. But yeah, really good representation. We basically chose the companies by surveying everyone we could in our existing audience. Our database has almost 150 ,000 people across every possible function.

5:40We have thousands of learners in our AI academy, many of whom are not marketers. So it really came from kind of our existing channels and audience. And we were just looking for the widest cross-section possible. So if I parse that out, what I say is, all right, these guys had a really broad field and they surveyed the whole field. Yes, that's exactly it. And so what we're getting here is input from what you guys, the broadest field you could reach, which is one of the broader fields most people probably could reach. So give us the headline conclusion of what this field says. Yeah, a couple big things to note here.

6:17We can unpack whatever we want to unpack, but really two big things kind of jumped out to me from the preliminary data that we're kind of analyzing at the moment. So first is there's quite a bit of disconnect between individuals and organizations when it comes to AI adoption, integration, and where everyone's at. So individuals are actually kind of outpacing organizations. So when we ask people, how would you define your personal state of AI adoption? There's a number of options, two of which are really important in terms of later phase stages, which would be integration and transformation. 53 % of people said they're in those later stages.

6:57Whereas when we talk to organization, when we asked at the organization level, hey, are you understanding piloting or scaling AI? Only 25 % of organizations are at the scaling stage at the moment. And on top of all this - And when I say, when you say scaling, what does scaling actually mean? I'm rolling it out to everybody or I'm just - Yeah, scaling means you've gone beyond initial pilots. You've used AI before in the piloting phase, perhaps done it for a number of things, but scaling is not only is it rolling out to everyone, you are expanding your use of it across every possible facet of the company.

7:30Obviously that encompasses quite a wide array of companies, but that's the most mature stage we would say at this stage. And that is, that means I've rolled this out and it's in the toolbox. And I kind of expect everyone to open that toolbox. 100%. Yeah. Yeah. So you'd be the most mature organizations, we would say, would be in that scaling phase. And so, you know, only about a fourth, which is still good. That's better than before. But only about a fourth are at the furthest or as far along as they could be, whereas almost half of individuals are quickly moving in that direction. So, you know, again, it's still, there's a lot of nuance to the data.

8:05but individuals are essentially outpacing their organization. If only 25 % are scaling, what are the other 75 % doing? They're either doing not enough or they're in the piloting phase. So a big trend is we've seen a lot of people, most actually, so about 46 % to 47 % of all the people surveyed are in the piloting phase. So they've started using AI for select use cases. They may be using one or more tools, but it is not widespread across the company. And we see this is really common. I don't want to say kind of to our team, I told you so, but I felt like I was pretty confident this would be the data because this is a challenge everyone has.

8:48And wait, one of the things about pilot is I can have a real pilot or I can have a pilot that gets the board and everybody else off my back. How many, and I don't know if you can answer this from the survey, but when you look at the pilots? Are the pilots real? And if the pilot goes bad, do companies actually not scale? Yeah. So I think that we probably don't have any conclusive data on what is real or not in terms of this. I think the vast majority of people that answer that they are piloting, I would say they believe they're real. So I'm sure there's some diversity there, but I think they believe it at least.

9:29I would say that when we encounter organizations piloting, the vast majority are doing some real things with it. But in terms of how far and how fast that's moving, sometimes that could be a little overstated. And are there any companies that are still just hanging back waiting? Yeah, I think you'd be kind of surprised. We were actually surprised. So in years of doing this, pretty recently, I think we've come to the conclusion that a lot of places are a lot further earlier in their journey than I would have anticipated, just because there's so many actual challenges to figure out, especially bigger enterprises.

10:06There's a lot of legal, a lot of IT, a lot of logistical things to figure out that people are not moving very fast to figure out. So plenty of widespread adoption. The hype is real. This is happening, but the gap between people doing stuff and not doing stuff, I don't think has ever been bigger. Well, and some of the hype about we're an AI first company or we are on the cutting edge, you know, you're now seeing huge amounts of advertising saying, you know, we're all over it. Yep. Is that a lot of the advertising is out of the actual implementation or is that in line? Yeah, I think that there is a tendency to overstate things sometimes.

10:45I know we're the worst sometimes. Yeah, I think there's some hype going on. Again, there is real technology, real use cases, and real results behind the hype 110%. But I think a lot of people are talking a big game too at the moment. Before we go on, I should have asked this in the beginning. What actually is the chief content officer do at the Marketing AI Institute? Like that's such a big title, chief content officer. That is an excellent question. So we kind of think of the business, both Marketing AI Institute and SmarterX as a whole. It's basically an education event and media company. So, you know, we've got online education, in-person virtual events, and there's a media engine that powers all of it.

11:31We've always been a content shop. We have a content engine that where we teach, publish research, publish content weekly through the podcast, newsletters, written content. So basically my job is I own that content engine and make sure that we're publishing across all the right channels, all the right stuff our audience needs to know if they want to accelerate their AI literacy. We are taking a short break from this show for a word from our sponsor, Typeface.

12:04To orchestrate workflows across channels, meet ArcGents, your AI teammates handling complexity so you can focus on what matters. All with a reimagined workspace where you and AI create as one. Spaces works the way you do. Create anything, from documents to videos to entire campaigns, at scale. Our agents help you craft campaign storyboards, powered by your brand hub, giving you the perfect starting point. Transform one asset into countless variations for every audience and market.

12:45Join industry leaders already reimagining their content lifecycle with Typeface. Welcome to Marketing's Next Chapter, where every story finds its voice.

12:58Now, back to our discussion. Okay, that was a sidebar that I took us down. I'm going to get us back on the regular show. So, you know, I want to go back to this say-do gap where we just talked about the advertising where what I say is maybe not what I do. How can you give us some examples of actually say-do working in the marketplace? And then how do you even know as a leader if you have a good say-do gap? You know, I'm saying I'm AI. I'm telling the board I'm doing AI. I got all these pilots running. But am I really doing AI or am I just saying I'm doing it and not really embedding it into the infrastructure of the company?

13:44Yeah, I think there can be a big gap there. And what we see sometimes happen is so that gap first starts with what you just mentioned, which is what you're telling the board, your boss, all this great stuff we're doing with AI. And it's true. I don't think people are lying about it. They're starting to run those kinds of experiments and pilots. But when you actually sit down and say, okay, well, what's happening each week in and out, there's even wider gaps between what is being said and what is being done. And I think a lot of that happens because the execution is really, really hard. It's one, very hard sometimes to put pieces in place to actually monitor AI usage.

14:22A big other piece that I'll probably talk about again is number two, AI literacy is very often ignored by companies or underinvested in. So we're shotgunning tools at people and saying, hey, we bought all these licenses, go use this stuff. And your average knowledge worker might not fully understand what AI even is, what it's capable of and where the technology is going. And so you say like, oh, we did all this stuff. We bought all this stuff. We have a plan. And then it's kind of crickets sometimes in organizations. And how do I measure true adoption and true literacy from a real like, okay, I really want this to work.

15:01How do I actually measure that? Particularly if I'm a big, humongous company and I don't really, I know all my workers can't absorb this all at once, as much as I want them to. How do I measure all this stuff? Yeah, so this is an ongoing challenge, and there's no perfect answer just yet. But I would say we see the most successful companies doing a few things, which is one, sounds very simple. But at a very baseline level, you need to be looking at whatever analytics or metrics are available in the AI tools you've bought. You can see usage. You can typically see who's using what and how often it's being used.

15:39That's a part that people really need to pay attention to. because if people are not using the tools, the rest of this doesn't really matter. So if they're not using the tools, okay, why is that? Now, it may be that education or literacy problem, and that's kind of one of the things we built our company to do is basically not only provide the education platform that companies don't have and need for AI literacy, but also you have increasingly when everything's in one platform, you have ways to actually track over time how much better people are getting with AI, how much educational content they're both consuming and using.

16:17So I think being able to also track that piece of it, the literacy piece, would be really smart and wise for companies to start doing. And if I read all this hype, and I will say, I think I'm a heavy AI user. I use it. I teach college. I use it there. I use it for the show. I use it a lot. But then I always think I'm super behind. How do I know if I am a laggard or not as a person, individual, and then as a company? I mean, really, no. Yeah. So as without hiring you guys, because I know you're going to get the talking points in. So I'll just get them in for you. I love it. I want to do a self-assessment.

16:57How do I do it? You know, at the individual level, I think it'd be wise to consider just how often you are even using tools to begin with. Are you a daily, weekly, monthly user? If you're not using AI every day for at least something, I would say you might want to consider further use cases for AI. I think seeing what you're using it for is the number of use cases you are leveraging AI for going up or down. Pretty simple question, or staying the same rather. You probably want that number to be going up because that's a plateau you can really hit where you get geeked about a few different use cases.

17:31use AI for them, you see some results, but you're not continually expanding and reinventing your workflows using the tools. Now, companies is a little bit of a bigger challenge, but I would say there's some kind of obvious boxes to check to understand if you mostly if you are a laggard, right? So like, if you don't have any really formal useful AI policies, that's probably an issue. If you do not have at least an approved tool, and even if it's not your ideal tool, something that has already been approved that people can use and are taught how to use. If you don't have that, you're probably lagging behind.

18:07If you are finding yourself stuck in that phase where only one department owns this or only a couple are experimenting with it, I don't think you're that far behind, but you might be lagging a bit. That's a huge issue. Like 40 plus percent of people said that silos and like inconsistent deployment. Yeah, it was a huge issue. That is almost half of the companies are not expanding this past the silo. And to toot the horn of our audience, that's probably low because we probably, hopefully, if we've done our job, have a little bit more of a forward thinking or at least further advanced audience than your average company.

18:46I could probably guarantee that's at least over 50 % at your average firm, if not much higher. Hey, so Mike, one of the things a bunch of our our guests have said before is you should have more than one tool in play yeah i i i'm certain you probably believe that but how many tools should you have in play there's like you could have at least five in five of the big guys in right now i always say the right answer is whatever you can get approved especially like working with bigger enterprises so all you would eat can eat buffet the right answer is how much can you eat honestly i would agree with that i would say you You can never have enough redundancy.

19:24But guess what? Even if your organization is a little more conservative and you only get one tool approved, that's better than zero. But I would say you need at least two or three, as I would argue, as your daily drivers. Look at what's happened in the news recently. We just saw Anthropic specifically getting a huge fight that's still ongoing with the U.S. government. They have huge usage problems due to this now where they're trying to buy more compute. You and I were talking before we started recording about all the need for all these data centers and compute. And there's outages and usage problems.

19:56And I don't know about you, but I'm at the point now where if my AI tool of choice goes down in a workday, it's like if the internet's going down. So we're not getting a huge amount done. So I'd say redundancy is important. At least two tools, probably three would be kind of my sweet spot personally. Okay. I have to ask if I am trying to catch up, say I'm in the middle of the pack or I'm in that bottom of the set. Yeah. What are the biggest mistakes companies make when they try and catch up? Because, you know, every no, I haven't talked to anybody, maybe one or two people that think they are truly on the front end of this.

20:37Yeah. You probably talk to a bunch more people, maybe on the close front end. but almost everybody feels it's changing so fast, it's hard to keep up. If I want to catch up, I know I'm behind. I acknowledge I'm behind as a company. What is the biggest mistake I can make here? We've talked about what you should do right, but I'm sure you see a bunch of people step in it when they do this. Tell us. Yeah. If I really had to choose, I would say some version of the following mistake, which is, okay, let's say you've kind of had your your come to Jesus moment over here and you say, okay, I know I'm behind.

21:10AI is super important. We're all in. Great. That's amazing. But it's very tempting for companies to then say, okay, we're going to get all the tools. We're going to get all the latest technology. We're going to give it to people. And then we're going to go. And we're going to not only catch up, but get ahead. That's an amazing sentiment. There's nothing wrong with wanting to be on the bleeding edge like that. But if you think you're just going to hand people tools, and especially as we're getting into AI agents, which we could talk about. Oh, we are going to talk about that. Yeah. As you hand people these tools that not only they might not understand the full power of, but also are increasingly autonomous, it can lead to a huge amount of problems.

21:48You cannot just turn this stuff on and expect your team to get it. This year in the data, for instance, lack of training and education is still the number one barrier that people say when it comes to adopting AI effectively in their organization. It's not about access to tools. A lot of them don't really have issues anymore justifying getting AI tools in 2026, which is a great thing. We've asked a question around, what is your top barrier to AI adoption? It's been training and education for six straight years. Hey, Mike, and if I don't want to hire a firm like you, if I want to say, I have to do my training and education, or I'm a small company, is there something I should get?

22:33because there's all these tools I can choose from. Is there a standard training anywhere? Not really standard training. We often point people, we see a lot of education solutions as super complimentary to what we do, but there's plenty of good free training through Coursera, LinkedIn Learning. Honestly, I think one thing people don't do enough is if you don't have a budget or you want to get more experience and catch up and find decent training, go ask AI. Not for training to take, Like have it teach you. I mean, half the time when I need to figure something out, I ask Claude or ChatGBT or Gemini, hey, teach me this concept.

23:11What is this about? How do I start implementing it in a real way in my business or in my work or my life? So that's a really valuable outlet. I would say, though, that's the issue right now. One thing we've tried to help solve is like there is no centralized. There's no one-stop shop for this stuff. I mean, our solution is beginning to become that. And there's plenty of others like it. But yeah, there's no one single authority where you say, you do that, you'll be up to speed totally. And then let's talk about the advanced edge, where I now have a bunch of AI agents maybe reporting to me, working for me, that I have created.

23:50Tell me what are the do's and don'ts, or the Hall of Fame and the Hall of Shame in managing agentic agents that you have created? Yeah, so it's still very, very early on a lot of this stuff. So we're still even kind of feeling our way around it. But I think big issues you want to be considerate of is one, what does your agent have access to? Because a lot of people don't always very carefully look at what permissions they give agents for different folders, file systems, accounts, that alone. For example, by grade, the mic could put agent for all my stuff, and I don't actually box it in right. It actually is all over everything I have.

24:34It can be, depending on the tool, yeah. There's been people that, for instance, OpenClaw is a big buzzworthy open source agent at the moment. You'll hear these horror stories of people giving the agent access to their bank accounts, their password. Not a good idea to begin with. So no surprise that things go wrong. But yes, agents will try to go do things for you by figuring out the steps to do those things. There's a lot of unintended consequences. It may think that the best way to fix your code base is to delete it. Because if you delete the code base, there's no problems, right? I mean, it's not going to always be that blatant, but there's all this unintended stuff where agents could go wrong really, really quick.

25:13Or you come back and your agent has sold your house on MaltBot to another agent. Yeah, exactly. Right. So any tips for doing this right? You have these agents. If you have any examples of how to do this right, say you're right on the edge and you say, all right, we're going to make the next jump. We're going to do agents. And how do you integrate them with people correctly and manage them correctly? Any thoughts on that? Yeah, I think that's such a bleeding edge area. I think everyone's still trying to figure that out. But for me, it always comes down to some combination of like fully understanding your actual workflow.