Welcome And Guest Introduction

Chris Norton 0:00

Welcome to Embracing Marketing Mistakes . I'm Chris Norton and today my guest is one of the co-hosts of the Artificial Intelligence Show , one of my top three podcasts that I listen to every week . Now , this guy began as a journalist in Cairo , lived through the Egyptian Revolution , moved into agency marketing , which is why he's on the show , but now he helps brands apply AI at SmarterX . He also shares his own mistake . He catches himself thinking that bit too small every single day , even though AI knowledge is his job . Today's guest is Mike Kaput . You're gonna love this one . This is one of my favourites . Enjoy . Mike Kaput , welcome to the show .

Mike Kaput 0:48

Thanks for having me, Chris.

Chris Norton 0:50

I am so excited . Right . Um , right , for the listeners who don't know , if you don't listen to it , um the you uh Mike and Paul Rate Roetzer , is that how you say his name ? Yep , that's it . Both host the uh one of my top three podcasts that I listen to . Um and I've listened to the show for I think three years . I'm a I'm a bit of an AI geek , love the show . I listen weekly , every week . I've listened to some of the Thursday ones uh as well , which we'll get on to . So on a Wednesday , Mike posts a show with uh Paul , and I'd say that you know your show is more uh current affairs in AI . And you sort of but what I like about it is , and for our listeners who've not heard it before , so our listeners um we're number one in Apple in marketing today , and have been for I think about six months . We're doing we've been doing quite well in and we're about we we we jump around in business rank ratings in the UK from like eight to ten . Now , weirdly , you guys are always uh and The Artificial Intelligence Show is always somewhere around as well . And it's because you you you come at it from a marketing perspective , which is what I I quite like . I know you're trying to spread it out of what you're doing at the moment , but I I hear you like um you you you sort of reference back to marketing and your experience , which is bizarre , because Paul hosts it owned a PR agency . Hello , which is what I do . So I host um we I hold I own Prohibition with my business partner , which is uh a PR agency uh in the biggest state in the UK , which is uh Yorkshire region . Um is the big so we're we're the biggest PR agency in the biggest state in the I'm not saying the biggest PR agency in the country , but we're yeah , we're doing okay . And uh so your the fact that you he hosts he hosts the show , it's about AI , which I love . I've got some I've I'm not a Doomer , but I've got I'm getting I'm getting to the point where I'm a bit worried I might be a Doomer . There is a question soon , don't worry . Uh I'm just giving the audience some context . And yeah , so um you you basically do a current affairs show , which goes on quite a lot , like and you read your news , and I'm not you're like a couple of BBC newsreaders going through , and then you give your opinion . No politics , because you don't want to get penalised by Donald . And um I think it's fascinating because you cover so many different areas in marketing , artificial intelligence , and you bring it back to like how it can practically be used , which I think is really really useful because that's how I do my training in the UK when I'm training clients or we're working with clients . So I found it really interesting , and the fact that you are an ex journalist , is that correct ?

Mike Kaput 3:26

That is correct.

Chris Norton 3:27

So, like, there's so many, it's like, yeah, I was like, I've got to I've left it two or two and a bit years. I was like, and I don't want to do it when the two

From Cairo Journalism To Marketing AI

Chris Norton 3:35

arch pod was too small, but I thought I need to get these guys on the show because I've I think I'd like both of you coming at the different angles. So, how did you how did you get into this, Mike? From first of all, you're a journalist in is it Cairo where you were a journalist? I mean that that goes straight into AI, doesn't it? Journalist in Cairo, straight into AI. Do you want to explain it?

Mike Kaput 3:55

Yeah, Chris, it's a wild story. I've had a it's certainly a winding path to where we are today. But um, you know, I in college studied political science as kind of almost on a whim. I took up learning Arabic in college. I think we had to like take a language credit or something when I signed up. So I had burnt out in high school on uh French, which I did not like and took for years in high school. And uh, you know, when it came time to say, hey, what language do you want to take? I just rolled the dice Arabic. I figured what how bad could it be? Do it for a semester or two, and ended up really enjoying it. So, long story short, when it came time to study abroad, which is something I'd wanted to do, uh, at my small school in Ohio, one of the selections was Cairo, Egypt. And so went there for a semester, fell in love with it, moved back after I graduated college, and fell into a job as an editor and a journalist at a couple English English language magazines. So it's like an associate editor, wasn't doing anything, you know, I wasn't like in war zones or anything, though I knew plenty of people who were, which was interesting. But I was, you know, mostly writing and editing in English and living in Cairo for a few years. And unfortunately, my time there got cut short by the Egyptian Revolution, which I lived through some of and left the country. It was quite the experience. And, you know, after I'd returned to the States, I actually got pretty quickly introduced to Paul because Paul owned a PR and marketing firm in my hometown of Cleveland, Ohio. And a mutual acquaintance who actually works for us now and is one of my very good friends, um, introduced us. And at the time, Paul had just kind of this a long time ago, Paul had just uh discovered the power of this wonderful software platform called HubSpot, which at the time had only a few dozen employees, I think. And Paul made a very big, very prescient bet on HubSpot and inbound marketing.

Chris Norton 5:59

And as a big Paul didn't pay off, did it?

Mike Kaput 6:01

Not at all.

Chris Norton 6:03

Wow is this?

Mike Kaput 6:04

Yeah , this isn't the first time Paul has seen kind of around the corner , but I was lucky enough as part of that . Paul was making a heavy investment in content marketing and was looking for people who are very good at writing . So I'd had a writing , business writing and journalism background . I made the shift to the a world of marketing agencies . And long story short , spent almost a decade or so in that world , kind of becoming a professional marketer , for lack of a better phrase . And within the walls of the agency , Paul started getting geeked out about AI and looking into applications of it . This was way back . Paul first started his journey in like 2012 , maybe , looking into AI . And so in about 2014 , I helped Paul write one of the chapters of his second book that he was writing . And it was all about um , it was called the Marketing Performance Blueprint . And it was all about marketing and more data-driven analytical marketing and marketing technology , but there was a chapter on the early applications of AI . And from that , Paul and I started really seriously talking about how AI could be applied to marketing . Figured eventually Paul was like , let's start sharing what we're learning . He launched something called Marketing AI Institute in late 2016 , which was basically at that point just a website that was under the banner of our agency . It was a separate brand , but everything owned by Paul . So it's kind of moonlighting while working at the agency on this kind of Marketing AI Institute project . And from there , we had built up a pretty decent audience , just interested in what we had to say as marketers and business people . That was kind of the unique approach , like you alluded to . We were just tech savvy , but non-engineers in business and in marketing , trying to figure this stuff out . This was well , obviously , before ChatGPT . And fast forward to today , our company SmarterX uh Marketing AI Institute is a subset of that . It's a sister kind of uh where the SmarterX is a parent company , so to speak . But today we are an AI transformation company that helps brands and businesses better understand , approach , and apply AI , not just in marketing , but across many different functions and industries , every function in industry . We do that through education , events , and media . So we kind of grew out , got our start in marketing specifically , but grew out of more this really powerful need for real-world applications of AI and real-world use cases of AI in business at large that was like understandable and approachable by your average person

Building Marketing AI Institute And SmarterX

Mike Kaput 8:50

, not by a data scientist or engineer .

Chris Norton 8:53

Yeah. So do you get like because when I listen to your example , you do an example every week , right ? Where you're saying this is my uh use case this week . And weirdly , Paul has over the last sort of two or three months , he was he always had like a loose one he'd said , I've been jousting with it on strategic sense , whereas you'd go right into what you were actually using it , I'd be like , My God , that's what I'm doing with it . And you were like giving interesting um examples of agentic as well . Like , because a few years ago , that was not a thing . This this last year , last 12 months has been about agentic , and like because I think I started using Claude Code in January . Oh , Claude Cowork , and then I moved into the Claude because I was on ChatGPT , and then it's it's like at the moment , it's this I've got and to be clear , I've got ChatGPT , I've got Gemini , I've got I've got Claude , I've got Copilot , I've got I'm I'm I am AI over overloaded , and actually , I was we were having a bit of a joke with someone the other day . AI is supposed to release all this uh extra capacity , and I find myself working weekends to do put to give it things to do , to to use the tokens and to because I've got ideas coming out of my my head , but I've got I've got a question for you in there , which is like because owning a PR agency , and the guy you that's what you guys did . You guys were in a in a PR agency as well . Are you do you like the value side of what a humans giving now ? Because you guys have so he's as Paul sold the agency , right ? He sold the agency . So there's a big problem with human value now , and every week you're covering the jobs and situation and whether it's affecting jobs , and Peter P that's all people seem to bang on about oh , what's it gonna do to jobs ? And I get it because we've I I've heard your latest um that Anthropic report that you covered last week . There was a big report saying that it sees that in the next three years loads of jobs are gonna go , and it's like that's quite doomer-ish in itself , which is quite shocking when it's about to do an IPO . Yeah , but as a business owner , like in the UK , I'm looking at it and going , okay , this feels quite a way off to me , even now , like because everybody's got the technology already , but not everybody's using it . The adoption is not huge everywhere

Human Value Through AI Architecture

Chris Norton 11:05

. How do we add keep adding value as humans ? And how what where do you think the value is ? What and how are you using it to ensure that you're adding your value rather than just clicking Claude , do that for me ? ChatGPT , do that for me . Gemini , read my Google Doc . You know what I mean ? It's like , am I just clicking buttons and doing more work ? Where's the value now ?

Mike Kaput 11:28

Yeah , that's a great question . I think one , everyone is trying to figure out the answer to that question . I don't know if there's necessarily as hard and fast answers as some people would claim to have found , though . Hey , if you found the answer with a capital A , let me know , please . I would like to know . Um , as long as you're not trying to sell me something , I suppose . But uh I would say what I've found is that , and I work a lot in content these days still , and that's like ground zero for this . It is very possible if you decide to be lazy and too reliant on the tools to click a button and just have your day be full of clicking buttons and having AI do work for you . And there's nothing wrong with using AI for plenty of things , but I find increasingly that I am drawn towards what I would call the architecture and the orchestration aspects of the work . So it has never been more important to sit down with the skills I have , the expertise I have , the perspective I have , and the context I have , and actually architect what I want AI to do , what it should actually be focused on , what direction generally I'm even trying to go as a team , as an organisation , as a business unit , et cetera . All of these very high-level strategic decisions , what I might call like the origination of like what are we even doing to begin with ? This is absolutely critical because AI , at least today , can scale up really dumb decisions as well as it can scale up really great decisions . So , like what you actually work on is probably the number one hardest question and most valuable thing to solve because you can do when you can do anything , doing the right thing is kind of the most important piece of this . So I struggle a lot to spend time on that , but I also think it is the most important thing we can do . So I spend a lot of time on what I would call like the architecture part of it . And that may even be as simple as stepping away from the computer to start and just writing down on a piece of paper . Here are all my thoughts about this thing I want us to build or a project I want an agent to go complete . It genuinely may be that mundane , that manual , um , because that part matters so much . There's obviously plenty of parts where there's a human in the loop with the work being done and reviews and oversight . But I think that architecture part is a huge piece of it . And then there's the orchestration piece , which is understanding how to fit together different tools and capabilities and crowds of agents at some point to get things done in the right way . And there's a lot of like not just science , but probably art involved in that as well at the moment , at least in you know , marketing sales and service , that I think there's still a big role for humans there . I would say to the kind of the agency point , when I think about it's an interesting time for agencies to say the least , like I would love to get your thoughts on that . But like I look back at the stuff we used to do as an agency , and this was a long time ago . So it doesn't apply to what agencies are doing now , but I look at what we used to charge clients for a decade ago , and basically all of it would be done by AI today . Um , that doesn't mean we would still be trying to sell clients that same stuff , but it's a pretty clear indicator that , oh , okay , like the market and the agency world has changed . I do the thing that gives me a lot of hope for agencies is I can tell you , every single customer we talk to , we fill this role in a little bit in the education space . Like every single customer we talk to , every client , every partner , everyone is craving guidance . They need help . And it doesn't mean they need every single thing solved for them , but even people a few steps ahead of them in the journey that know how to communicate and guide and educate and consult are in one way or another invaluable . Now , how that translates into your service mix , your revenue model , that that's bigger questions . But I think that being seen as that trusted advisor is extremely important and potentially a huge opportunity for agencies that are not fully seen that way today .

Chris Norton 15:58

Yeah , I mean , well , I mean , we I have just had a look at my HubSpot . Uh yes , we're a HubSpot agency . Have been in HubSpot for 15 years since I started the agency , uh , 16 years nearly now . And um , I can see that AI search , well , our AI search traffic's doubled because of the content we've been doing , and so we're doing the right things . But I mean it the traffic and everything dropped off a cliff . So I've been doing uh um years , you know , when the AI Overviews came out , so I've been doing lots of webinars on uh GEO and AEO delivering them , and clients are genuinely interested , they want that guidance on how to appear in visibility , but visibility doesn't mean business as well . Yeah , and then the the problem I have with AI is AI just tells you what you want to can you can you SEO optimise this page for me ? Absolutely , Chris . I can do that , I'm on it . And then you look at it and you go , hang on a minute , that's not quite you haven't thought about that , have you ? Absolutely , Chris . You completely go . I haven't thought about it . Well , why didn't you think about that ? Yes , that's my fault . And you're like , you end up in this loop of having a conversation with somebody , and you're like , you're like , why can't you just think of that ? And that's what you're talking about . The contextual side of it is really , really difficult . And if for somebody that hasn't got the experience or the context , and you set them off , um , they could make a lot of mistakes , is the problem . But thinking they've got the right result , and I've seen a lot of legal cases where people have gone into courts with the they think they've got a an ironclad case and and they haven't . So I've got another question for you because if if you're um if you are a graduate in so my son is 20 , he's uh 19 , right ? He's not gone to university , he's gonna do it , he's

Graduates And AI Literacy In Marketing

Chris Norton 17:40

gonna do an apprenticeship , which I think for the first time in a long time might be a good idea . Didn't someone say that everyone should be plumbers ?

Mike Kaput 17:46

Uh right , plumbers for data centres , maybe even more specifically .

Chris Norton 17:52

Yeah , Optimus starts doing the plumbing . Um I I don't know , yeah . So my question's this as a graduate entering marketing and comms , what what should they be doing now , do you think ? Is there something they should be doing to ensure that they are because they're not got the experience , and this but all of a sudden you've got it at your fingertips . What's what's the what's the help to get them started ?

Mike Kaput 18:17

You know, on one hand, this is an extremely tough environment for entry-level work because of what you described, because the things that are precisely valuable when working with AI, the experience, the context, the judgment, the real world background to know what is good and what is not is precisely what they haven't learned yet. So on one hand, frankly, it really sucks. And I don't have like a good answer to like how it can not suck to be in that position. But on the other hand, I do see, at least with the businesses that we talk to and the market as I see it, a huge opportunity. And this will be a very imperfect analogy, but I do kind of see a little bit of you know, this rhyming with the history of like social media and digital, which is like you would typically, not always, but sometimes in the past, you'd say, okay, like this young person right out of college understands this technology or understands social media in a way that like I just don't get it, or I need help, like understanding this. We can teach them everything else. Let's get them in to help with this piece of it. I think there's probably an opportunity similar with AI in the sense of using AI is not going to replace all the marketing judgment and expertise and background that you are going to need to learn on the job. But there are, despite what you may read on X, there are 99.9% of companies are not building their own agent swarms. They are not using AI for everything. They barely understand this stuff. So if you as a younger person can say, hey, look, I have built assistance or solutions or prompts or whatever it looks like for you, different tools and solutions and approaches using AI to common marketing problems, showing that kind of work, which is thankfully never been easier to do, is really, really compelling and I think opens doors because it shows that you are actually AI literate, which will be actually absolutely critical. The trap you cannot fall into is just assuming that everyone out there is totally ignorant of AI, you cannot be clicking a button and submitting AI-generated work, which plenty of young and older people are doing these days, unfortunately. It's a nightmare. And we it's a it's a real endemic problem. And but beyond that, I do think if you can come to me and say, hey, I have at least domain expertise in marketing that I've been taught in school. By the way, here are the ways in which I am able to use AI to solve real marketing problems. I think that's very, very compelling to a business to at least get your foot in the door and start learning all the things that AI cannot do or cannot teach you right out of the gate. But also, AI is a fantastic learning tool as well. There's plenty of scenarios where, frankly, if your AI skills are really, really good, that may be more important than being a total dyed in the wool expert in marketing in a lot of cases. Like that, there's a weird thing happening. I'm not saying it replaces that expertise, but the models are getting good enough that like that uh it's not guaranteed that the deep expertise will always be the advantage it is today.

Chris Norton 21:39

You mentioned you mentioned AI swarms there , right ? So um , so well , first of all , it was oh , let's create a load of custom GPTs . Well , they've gone , they've they've been binned off , haven't they ? Yeah , yeah . And then Claude invented skills , Anthropic brought skills , and now everybody's talking about skills . Now we've got a genit

Agents At Scale And Cybersecurity Risk

Chris Norton 22:00

. So you can have multiple agents . What's an agent swarm then ? That sounds really scary . It sounds like you're gonna you're gonna send me a load of like North Korean hackers onto my website to do some , you know , to do what what what what is an AI swarm ?

Mike Kaput 22:16

Well , I apologise . I hate that I even use that term . I think Sam Altman was on a podcast recently . We talked about this on our podcast episode that comes out actually today , I think . Um where he was he Yeah , he was he was on a podcast and cat and he was like , I hate this term swarm . It just means like a group of agents . And the but he kept using the term because it's become such a you know turn of phrase . It's just it just means many agents , a group of agents . That's it . Um , you do hear about it a bit sensationalised in exactly the types of scenarios you just mentioned , which you hear about it with like the Hugging Face incident with OpenAI , where it's like the dozens or hundreds of agents being deployed towards a common task . Often the way we hear about that is through malicious cybersecurity stuff , because that is where the frontier of this stuff is happening . That's what you might call an agent swarm .

Chris Norton 23:13

Yeah, I thought it was interesting how you and Paul explained the fact that which was really really you broke it down not in this today's episode that you've issued, um, but um uh recent a recent episode which was about the fact that people are getting um hacked. Um and and basically that uh the hacking is happening over is there's like this period of the next 24 months, which we're really at at risk of everyone being hacked because the technology um today's technology can be cre the software can be prepared by AI and it can be protected for future proofing. Whereas you've got a Samsung fridge that you bought seven years ago that's connected to the internet, it might be a hacking problem, and that's literally what's happening everywhere, every single Wi-Fi device, and you've got this stuff happening, and then the UK in the UK, the government have warned us to get to to start buying things in case the the internet and the infrastructure is hacked. Well, that's one of the things. I mean, that's what they're telling us. What else are they are they not telling us? You know, that they're they're basically saying prepare for the end of the apocalypse. And I'm not even a doomer. I'm like, why is the government telling us to prepare for this? And it's partly, I think most of it is AI hacking over the next 24 months. We've got this period, like you say, you guys described it well, 24 months where we're at risk because the system it's gonna take 24 months for them to patch everything up. Is that is that fair? Is that what you guys were talking about? Do you think?

Mike Kaput 24:45

Yeah, I would this is yeah, this is not our kind of original idea, but it sounds like based on we're not cybersecurity experts by any means, we do not claim to be. But from what some of the commentators out there seem to be saying with these high-profile cybersecurity incidents, is look, like all software to date that underpins the entire world has for the most part been built by humans and vetted by humans. That means it is limited by how quickly humans uh can patch security vulnerabilities. It is also limited by how well human attackers enabled by in the past things like automation at scale and their own brains. It's limited by how well they can attack those systems. Well, now AI changes the equation, right? Where now the agents have proven out as being actually viable, which is, like you said, happened in the past year, I think. Um, it's basically pre, there's pre-AI and post-AI. None of the software is built for post-AI. None of the security vulnerabilities. They're discovering new ones every hour of the day post-AI, because AI every exactly AI can discover this at scale. So what some people have said is look, there will be this like 18 or 24 month period where it's going to be really, really fraught. Because until cybersecurity experts and software companies can use AI, frontier level AI, to make these systems secure, other people will be trying to use frontier level AI to destroy and break these systems. So at some point moving forward, you will hit a tipping point where software will be just built natively by frontier AI to be relatively secure, though obviously we'll never be a hundred percent secure. But there's this like window of real danger, is what it seems like people are saying, and we are in it, I think.

Chris Norton 26:44

This

Starting An Agency By Asking AI

Chris Norton 26:45

is what I was supposed to be a positive AI chat. Yeah, right. It literally is that's what so okay, let's go positive. So let's say, let's say you decided to start a marketing agency tomorrow. I don't know why it did. How would you how would you if if you if you uh I gave you fifty thousand dollars, off you go, start? I don't know how much we needed more of it because you need to you're on better money than we are in the UK. Uh here's a hundred thousand dollars. Go and start your own marketing agency, Mike. How what what what architecture would you how would you go about doing that today in 2026?

Mike Kaput 27:22

You know, I'm going to have a cop-out answer here, but it's it's real. Like this is really what I would do. I would record exactly what you just said, and I would crank up GPT 6 Astra to Max or Ultra and drop in that exact question. I literally would not determine how to do it myself. I would have the best, I'm not kidding you. I would have the best frontier model on the planet or multiple ones, stress testing each other's work, starting from scratch and say, given what we know about the current trajectory of models and where they're going, here are the parameters. How would you build this? I would use my own perhaps judgment as the person who is now involved in having to execute this plan. Of course, I would work with AI, but I'm not kidding you. That would be my first question. And the reason I would do it would that would be my first step is because I actually think if you are starting today, you basically are better served by throwing out anything you think you know and starting from scratch and then mixing back in what you think you know as it makes sense, because things are so uh transient, things are moving so strangely and quickly.

Chris Norton 29:23

You've blown my mind with that. I didn't think because I've seen a lot of people joking, I've seen a lot of spoof videos where people are going, um, Claude, what do I think about what do I think about the weather? Or you know, that's clearly planned question because nobody can think for themselves anymore. Um literally, we're talking about a really serious issue, and you would first go to and and to be honest, I do do that with a lot of things. Do you think it's this is an honest, honest question as well. Do you think it's worth turning them up? Because that's one of the things I find really confusing. Is is and I'm not I'm not over-exaggerating. I think Claude's the one that I go to the most. Um I've got Gemini Argon and all that sort of stuff. It's got about 14 different models. It's got 14 models and five different settings. Does that really matter? Do you think?

Mike Kaput 30:14

I do think it matters, but it is so confusing that I'm not sure how much it matters to go deep on it. I would say if you're a knowledge worker that has that you're not doing coding regularly and you have enough usage, like you're on a pro plan or something, crank it up if you can wait, because you can get really good responses. Now, doing this across an organization and figuring out token usage, how much intelligence versus speed you are willing to trade off, those are all questions that are a bit above my pay grade. I would say they're very contextual to the organization. But for me, if I have a hard question like the one we just posed, yes, I think using the best model possible and cranking it up for a question like that 100% would make a difference. Would it make a difference in writing your emails? Probably not enough of a difference to make it worth waiting however long you're gonna wait. But I would say for big strategic questions, it does make a difference. And I do just want to be clear about that kind of hypothetical. I I don't think necessarily that scenario would be having AI think for you. I think it would be more about the exercise of saying that what I think I know could be wrong. And I haven't considered exactly how weird the future could get. So I'm not saying I would take the AI plan. I well, I've thought a little bit about it.

Chris Norton 31:42

You've thought about it much more than 90% of our marketing audience. Like everybody listening to this, everybody that listens to this show is either a business owner, a marketer, or a comms person. And they everybody's heard they've they listened to me talk about AI, they know I use it a lot. I mean, it's I'm in 80% of my day, pretty much. Uh, and I'm doing what you're doing, which is jousting with it. But you guys have thought about it. You guys are running surveys and in the US about how people are thinking about it. You do the regular weekly polls, and I I listen to a lot of you the and that's what I like about the show is you cover the data and and and what's interesting about it. So I would say you and Paul have actually thought about the few because Paul keeps mentioning the fact that there's nobody out there asking the questions about what do we do when when all the jobs dry up and we're all sat around going, Great, I've just paid for our own job market to vanish. It's not necessarily extinction, uh our extinction, it's the extinction of jobs. We've just got loads of time on our hands and no money by the sounds of it.

Mike Kaput 32:43

Yeah, yeah, yeah. That's the scenario that I wonder if people should be taking a little more seriously.

Chris Norton 32:52

I think everybody , yeah , and I don't think everybody listen , I think the people that listen to your show will be in the top 10 % , you know , like 90 % of people are just like , Do you know what ? I went I used ChatGPT , it was G GPT , it was rubbish . Yeah , and I don't use it because it gives and they don't know what they're doing with the property .

Mike Kaput 33:09

Right, right, right, right.

Chris Norton 33:12

Um, right, well, this show is about mistakes, uh, Mike.

The Daily Mistake Of Thinking Small

Chris Norton 33:15

So um we share marketing mistakes. Um, have you made any mistakes in AI or in marketing that you'd like to share?

Mike Kaput 33:23

You know , I make mistakes in AI probably regularly , like often . Um , I will say like the category of mistake I make in AI most often , and I still make it and have to catch myself all the time , is kind of a I don't know if it's unique to me . It's not unique to me , but it might be just a function of like what I do . I'm always looking to push the frontier , right ? And say , okay , I don't just want to use AI to get more productive . I've done that . It's great , it is extremely valuable to businesses . We teach it all the time , but I'm always looking for not just AI for productivity or performance , both of which are critical to our business , but AI for innovation . And that's a lot harder . So Paul has a really great thing he often says in talks , which is like productivity is getting 10 % better , right ? That's transformative for a big business . Nothing wrong with that . Innovation is doing something 10x better . Um , that that is way , way harder . And the mistake that I have often made in AI is not thinking big enough . I am still often in the old way of thinking , where I say , not exactly like , oh , I should do this myself , why and not using AI for it , but more like , why are we doing this in the first place anymore ? Like we've , as a small example , like we are exploring and building and strategising around how all our systems are connected , all our AI tools rather are connected to HubSpot and what we can do with that . It's pretty straightforward , kind of a very big use case , right ? A very big use case for AI . But then suddenly these things come up where I'm like , well , wait a second . We went halfway into a full conversation about like what reports we need from HubSpot , what dashboard someone should build . And then I'm like , well , wait , why did I even let this conversation happen ? Can't we just ask these tools ? Like , what are we doing here ? Like , we don't need to do things the old way . You just go ask the question you have to that incredible machine intelligence we have hooked into . You know , so it's that uh call it mindset approach . I'm not sure . I am I am obviously like better at it , I would think , than a lot of people because I do do this all the time . But I consistently catch myself saying , wait a second , we have all these capabilities that I know about because we talk about them every week on the podcast . Why am I not taking a breath , taking a step back , sitting back , frankly , like I said before , getting out a piece of paper and just gaming out what is my dream outcome here ? Instead of , oh , let me try to do this thing a little faster or a little better or a little more efficiently . So challenging myself to dream bigger or big enough is honestly a mistake I fall into every day .

Chris Norton 36:32

Oh my god , that answer is blown me away as well because I that's literally what I'm doing all the time . I'm literally exactly what while you were saying that I didn't want to interrupt you , but I'm literally because I've listened to your show and I'm and like I'm a HubSpot user , I pay $ 20 odd thousand dollars , uh thousand pounds , it's even more than dollars a year to have multiple seats . And currently what I'm doing , I and let I'll let everybody into what we're doing . I'm currently assessing every bit of SaaS software that I've got in the business , and I'm but I've been building SaaS products , um , and I've I've I work with a partnership . I haven't even announced this anywhere . We're about to do an announcement , but we we've been working with an AI specialist , not your average chump organization , proper AI development company that builds proper proper bits of kit . And so I've been doing like uh prototypes in Claude Code or in um in Codex , we'll build prototypes for software , we'll then take the the a working prototype that I've tested for security , give it to the AI company , they stress test it and then put it into our Microsoft environment after it's completely safe . Now I'm not a web developer , but I have developed Twitter tools and stuff when back in the day because I'm a social media listening person , that's what I was into in in the days of early social media . But if I can do that , everybody listening to the show can do that . You have to be a bit of a geek , like we are . You have to be geek because you enjoy that stuff . But the fact that we can now manifest apps that do like I'd go , oh , Mailchimp , it's expensive . I'm spending thousands on that . I might need a different email provider on top of HubSpot , you know what I mean ? And and these looking at ways to reduce costs , that's what I'm doing . Like everybody everywhere is running business , must be doing that , right , Mike ? Do you think they are or are they not ?

Mike Kaput 38:24

I'm sceptical , everyone is doing that .

Chris Norton 38:27

Right , okay . So when you're when you because I my MAICON , that's what it's called , isn't it ? Yes , I've got a couple of questions here . So first of all , Paul's book sounds far sounded fascinating . So for those of you that Paul came up with it , he's written a couple of books before , I think , two or three books before . Yeah , but now he's been when I said earlier he was keeping things on the back burner , he wasn't sharing them . Uh but basically he was working on a book , and but the beauty of working on a book in 2026 , 27 is that if you're a podcaster like we are , you can generate how much how many spoken words have you got . Right , right . Exactly . Right . Uh so he learns your voice as well , but also he's been jousting with ideas and he's created a book . The book sounds fascinating , but I've got a bit of a flaw with the book . And the my flaw is this you'll like this . I registered my interest before my MAICON in the book . I never received nothing . So the the system has fought AI sister has fallen down somewhere . There's a marketing mistake . I I signed up for a book , so I thought , oh , I'd love to do that . It's about AI transformation . I think I'd love to read it and then I might share it with a few of my senior team . But as I registered my intro , it said register before , we're gonna have it ready for my MAICON . You guys talked a bit about MAICON , but you obviously you people that there , and I was like , where's the book ? Where's my email ? Where's the where's the damn thing ? I want to read it , and so and I don't

MakeCon Lessons On Change Management

Chris Norton 39:55

really know what what the biggest issues were that came out of MAICON . So what what what what were the most interesting bits that you've not talked about on the pod that came out of MAICON ?

Mike Kaput 40:04

Well , this maybe this is all related because uh I don't know if it was an AI problem , but it was probably a human problem on our end rather than a technology problem . But honestly , the the thing that came out of MAICON , one of the trends that was about kind of this idea . I had much more nuanced conversations with folks about what AI should be doing , what humans should be doing , and kind of the tension uh involved in that . I mean , I was I recall explicitly I did a panel where I interviewed three more mid-market um business leaders , either CEOs or CTOs , basically , who are basically in charge of figuring out how to use this stuff in a real-world business context and are running the day-to-day business , not like sitting around vibe coding all day . Um and a big thing they were dealing with was like AI is amazing . We are deploying it for all this cool stuff , but like I'm getting AI slop from people . I'm having people push back on AI being used for things . I'm having people just click a button and assume and you know , outsourcing their thinking to AI . They these were not people who are anti-AI at all . These were people who were in the real messy middle of deploying the technology . So I guess kind of related to your the email faux pas we had here is I think like the human element is such a such a question mark here , where it's like people , this is a big theme that we've hit on is like the technology is just this small piece of it . Giving your people the tools is obviously an important if critical step . But there are so many other pillars to AI transformation and to actually getting this stuff to work for you and with you , not against you , that have much more to do with change management , human beings , and human capital and management than just knowing all the technical ins and outs of AI . So I think that was a big thread through throughout MAICON .

Chris Norton 42:24

Yeah , I mean , yeah , because it used to be , I mean , um it used to be like , oh , here's 30 , here's 30 Copilot licences , away you go . That's AI done . Oh , and here's how to use here's how to use Copilot . Now off you go . Um I I've got a bit of an issue with transformation in the fact that and then you you guys touch on it sometimes , it's token usage , it's a nightmare . Like Copilot does has co-work now . We've got so to give you another insight into the business , I've got several people using Claude , I've got several people using ChatGPT , everybody has got access to co to Copilot because of the the the fact that it's in the Microsoft environment , we're in a Microsoft environment , I know you're in a Google environment , but I've also got Google as I've got all of a different goal . And I I often I often do what you said , Mike , which is stress test one thing again , and then ask an AI to review what another AI has done . So it's it it it fascinates me . But the problem is , you're right , like people aren't using it . So a lot of people don't use it well , and to use it well , you've got to get right in the weeds of using it , haven't you ? Yeah , yeah .

Mike Kaput 43:32

It's it's get not only consistent usage, but just the literacy around it, and also changing processes, workflows, approvals as a result of it. I mean, there's so much around not just getting into the technology and throwing in a prompt. There's much, much more to it to actually get value out of AI, I would say.

Chris Norton 43:54

And what does a human-centred approach actually require a business to give up ? That's not a question that I've written . That was a question I got suggested . So I want , yeah , what do you think ?

Mike Kaput 44:07

Well , my sense would be that a business is going to perhaps give up some tempting short-term financial incentives in some cases . If we are truly focused on being human-centred , we may have to make some hard trade-offs between what is uh not not economically necessary . If you're talking about the survival of the business , it's kind of probably all bets are off realistically . But if you're talking about , oh , hey , we could have an incredible quarter versus a pretty good quarter , and the pretty good quarter is going to be due to the fact that we're choosing not to lay people off and we're instead doing the hard , necessary , important work . Of reallocating humans towards better higher value work because of , if we have done our jobs right , AI is doing a lot more in making us more productive . That might require a long , longer horizon , longer-term view of the purpose of people within an organisation . So I don't know . There's no easy answer to this stuff . But I think as Paul has talked about a lot , a human-centred approach may have to be resisting this temptation to just say AI is going to help us do more with less . So as a result , we're going to get rid of people . We are going to have people doing 10 times more than they should or can do and just kind of treat them , yeah , you know , like output machines enabled by AI . Yeah , I'm sure some places , plenty of places , will go that route . I don't think it's particularly sustainable or advisable , but I think you're going to have to be really , it's hard , but you're going to have to be really diligent and intentional about how you see your people's role at the organisation moving forward beyond just the tasks that they do in a given day .

Chris Norton 45:56

Yeah, I th I well, I'm what I'm thinking is that the savings should be given back to the clients, right? So saving time, we're doing things a little bit quicker. Let's give clients more time. Like in the UK, we could we can actually have proper face-to-face meetings. We don't have to, I know you can in the US, but you have to fly a lot of places. In the UK, nobody's more than five hours drive, really. So, but the a lot a lot of Zoom calls are still a thing now, you know, even after the pandemic. But the question is, who does get the savings? Is it is it the right? So is it the marketing agency or the PR agency? They're saving money because they've got everybody trained up in AI, everybody's operating at brilliant. So they're they get this the the agency owners, myself, like you know, oh yeah, we take excellent profits. I can tell you that's not happening. I can tell you that's not happening. Yeah, uh, I am investing a lot of money in the the the having the best tech. We've done lots of training. I think I was talking to you a while ago about maybe getting a couple of my staff on some of your training courses as well, yeah, just to

Pricing Models And The 2027 Marketer

Chris Norton 46:55

see if it helps, you know. Um, but the the where do the where do the savings go? Because there's a big question in the uh that I've seen in the PR press, which is um day rates, they're a thing of the pattern. Day rates are dead. Yeah, charging by the hour, is that dead? Or I'm not you've got to charge by the output, haven't you?

Mike Kaput 47:14

Yeah , I mean , I you know , I've been out of the agency world for a few years , so take it with a grain of salt , but I would have a hard time on the other end of the equation understanding charging by the hour , just because I would look at it and immediately say that it would invite a conversation that would not be kind to an agency . Even if I was wrong , I would be immediately saying , Well , why did it take this many hours ? And if you say to me , Well , I'm paying for the expertise and that's expressed in those hours or the labour , whatever , and it can't be done elsewhere , I'd be like , Well , bet . Let's try . Like , you know , I'd be like so I yeah , I mean , I'd be interested to uh how confident someone is in that logic . Um , yeah , outcome-based price or output-based seems like the way forward . I I don't know how to navigate all the nuances there , but I just , I mean , I've uh relatively recently seen small for small things like bills for services we have paid for that have been itemised by the hour , and it was small enough and inconsequential enough stuff that I didn't make a stink over it . But like I immediately internally was like , wait a second , we're still doing hours here . This is crazy .

Chris Norton 48:33

Yeah , well , I've I I there's a law firm , right ? In the UK . Uh , but we we used a law firm when I set the business up . Me and my business partner , we used a law firm . Uh , I'll keep I won't name them to do our contract for a shareholder agreement because the Prohibition was established . I was I've been running for two and a half years before my business partner Will joined . Um , so he I did a shareholder agreement . That legal law firm , which I uh uh a couple of weeks ago has gone into administration . And I'm like , is this the AI because law firms don't really go into administration , but actually legal . I'm like legal now . Do people just go for a personal , like small , because obviously we were just a start , we weren't very big at that point . We're not big corporate , we're not using big corporate lawyers , we're talking businesses that help , you know , people with personal legal situations . Those sorts of companies now people are just going , Claude , or judging me too , hold me together a legal contract for a podcast interview with Mike Kaput off the Artificial Intelligence Show where he can't disclose anything . You know what I mean ? Is that what people are doing now ? And they're just whistling it up .

Mike Kaput 49:37

That I don't know. I mean, legal is an interesting one because that's probably something I'd be less hesitant to screw up on my own because I don't know what I don't know, right? If I ask Claude, it sounds amazing, but I could be liable for all sorts of nonsense. It's a little different, though, in marketing, right? Where it's like, okay, there's no question that I could ask Claude for a marketing strategy or an output, and it's not as good as what a talented agency would do. That's very possible. But like, what is the real downside? It's like, okay, throw it out, start over, you know. Whereas the legal one, I'm like, ooh, I could get, I might get sued. So my only point here is not that Claude or whatever can do what a marketing agency can. It's just like the bar is a little higher, where if an agency came to me and said we're charging by the hour, it'd be like, no, no, no. Whereas a lawyer, I kind of am not happy about that, but I am, I don't know what else to do, I guess.

Chris Norton 50:35

Yeah, exactly. Well, exactly. And I'm not sure if that's because we're marketeers, though.

Mike Kaput 50:40

Right. That's a good point.

Chris Norton 50:41

Well, if we were both lawyers, and this was the uh this was um a a legal part embracing legal mistakes, then maybe maybe it would, you know what I mean? It'd be like right, but I think because we're both marketeers, we don't feel comfortable. I would I know what you mean. Like, I would say that what it's done is enabled, like they there shouldn't be, and this is not true, there should there should not be shit marketing anymore. Right. But there is, there's tons, there's absolutely tons in there.

Mike Kaput 51:10

It's fascinating. It kind of comes back to what we've talked about a little bit with AI slop and things like that. I think it's really eye-opening to see how people use these tools, and there's a lot of debate and discussion around how these tools should be used and what is and isn't good or bad or ugly, but it never ceases to amaze me what people think is acceptable. And so I kind of sit back and say, I'm not super worried the cybersecurity stuff aside, I'm not immediately worried about AI as some nebulous technology. I'm always worried about the people using it, right? Like and there's such a wide spectrum. I I I struggle sometimes to be like, well, a human should always be doing this because have you seen some of the crappy output from humans? Like, so I don't know. It's it yeah, it's it challenges a lot of my assumptions on a daily basis, I would say.

Chris Norton 52:14

Yeah, and so okay, um, we've got about five minutes. So I just wanted to ask you, um if this is a this is a question for the next the future in AI is a nightmare, let's be honest. Three years. Also, you've got a terrible cough, haven't you? You're managing it.

Mike Kaput 52:30

Yeah, I'm getting I'm I'm getting over something, yeah.

Chris Norton 52:33

You're managing to keep it off camera, which is uh keep it off mic, which is genius. That's a pro. Um, but uh uh yeah, so the future 2027, if I'm a marketing professional, whether in-house or in agency or whatever, what do I what's what do what's the future for me in 2027? Do you think? What do I need to be aware of? What should I be what should I be staying on top of? You know, because there's a lot of noise around this people are sick of the term AI. In fact, I believe we're using the wrong term now, according to Donald. So we need to actually you know what I mean. So what do you think what do you think we should be aware of for 2027?

Mike Kaput 53:13

With the caveat that I think it is impossible to predict with how fast things are moving , I would say it would be a really good idea if you're a marketer that has only used chat-based tools and not agentic-based tools , even if it's something like Chat GPT work or Claude Cowork , totally fine . Um , if you want to go look at a Codex , a Claude Code , et cetera . If you have not touched any of those tools , now is the time . You need to understand what is possible and start understanding how these tools are similar and different from the normal chat that you're using . If you have only been using AI for basic use cases and not sitting back and understanding how you can approach AI for strategy , for bigger picture business thinking and planning , I would be really , really focused on that . I would be asking again , I would be asking AI your question , Chris . I'd be sitting there and saying , look , here's my career trajectory , here's exactly what I do , here's exactly what my background is based on how scaling laws appear to be going . Do a little research . What should I be focused on in 2027 ? I'm not saying it'll be the right answer , but it'll give you some idea , it'll give you some ideas . But yeah , if you have not done anything with agentic AI yet , that is absolutely your priority . You don't have to like go burn the midnight oil , like building all these agents . Just go try out ChatGPT work or Claude Cowork . Start giving it access to tools that you have approved to give it access to . If you can , if you cannot do anything at work , connect personal email and drive or document , you know , productivity suite accounts to start seeing what happens when you can say , not just let's go back and forth to refine a document , but here's an entire workflow you can do yourself . Go try , have um one of these systems help you create a skill to do a repeatable bit of work on your own and then reuse and refine that skill . That's helpful . I I don't know how long skills will stick around , but that process is valuable no matter what . Because what you're gonna do is start understanding that you're no longer just chatting , you're delegating work to autonomous AI . And there's a lot of opportunity there , and there's a lot of open questions you're gonna have about when I should be doing this ? What type of work ? Is it secure ? How do I make it secure ? Good . If you have all those questions , you're going down the right path . Are there easy , comfortable answers to those ? No . So welcome to the world at this point . But you're going to start thinking exactly how you need to be thinking as someone that is seen in 2027 as a valuable agentic marketer who can help contribute to the conversations and campaigns in your company that are using AI agents to get things done , should those be happening . And if those aren't happening , then great , you're at the forefront .

Chris Norton 56:19

Yeah. I mean, the thing for me is when I've been talking to people and showing them, they've actually co-work and these the different where the agentic side is they're just like, I've actually got to think differently now, like thinking differently about how you're thinking. That's the weird bit. Um, it's very different. Um, Mike, you've been on the show now. Uh, this is the question that everyone always goes if you'd

Who To Invite Next And How To Connect

Chris Norton 56:45

listened to an episode, you'd know. Um, you've been on the show now. Um, if you were us, who would who would be the person that you would invite on our show next and why?

Mike Kaput 56:56

Oh , that's a great question . Um , who would I invite on your show next and why ? You know who comes to mind ? And obviously , this is not uh me saying I have no idea if they do podcast appearances or what , but you know , top of mind for me is our good friend Chris Penn , who is just an AI wizard at everything related to marketing and analytics . And I read Chris's newsletter without fail every Sunday morning . Um and yeah , and it is just uh I I realise how far I have to go . Yeah , me too . He is geek level 12 , isn't he ? I I'm pretty sure he's making the AI itself at this point .

Chris Norton 57:39

I think he maybe he's not real. I don't know.

Mike Kaput 57:41

He might be right, right.

Chris Norton 57:42

Yeah, he's a he's he's a he's a he's an unsynthetic version of himself. Um yes. Okay, cool. If anyone wants to get hold of you, Mike, what's the best way to get hold of you?

Mike Kaput 57:51

Yeah, feel free. I'm pretty active, you know, ideally on LinkedIn, so feel free to connect with me. Like, shoot me a request, happy to connect, shoot me a message, or you can always contact us through smarterx.ai.

Chris Norton 58:05

Cool. Mike, thanks for coming on. Really appreciate it.

Mike Kaput 58:08

Chris, thanks for having me.