Preparing for AI Agents with John Munsell of Bizzuka & LSU

Welcome to the AI Chat Podcast. I'm your host, Jayden Shafer, and today on the show we have a phenomenal guest, John Munsell. I wanna tell you a little bit about him. He'll give you a little bit of his background. I was excited to do this show today because he is the co-founder and CEO over at Bazooka. That's an AI consulting firm. So he helps a lot of businesses adopt and scale AI. This is something I know a lot of the listeners are interested in, so I got a lot of good questions queued up. Another cool thing about John is that he's been doing marketing, software development, and sales for about 25 years. And he is the adjunct instructor of artificial intelligence over at LSU. So this is an absolute legend. We're excited to have you on, John. Welcome to the show.

A legend in his own mind, maybe. No, I appreciate it, Jayden. Hey, it's great to be on. Good to meet you. I like, I like the work that you've done, like we talked about earlier. So I'm excited to be on. Hopefully I can add some value to your audience.

100%. Yeah. So, tell us a little bit though because I'm curious, I'm sure others are, what got you into this space? How did you start? You know, how did you become the expert in AI that you are, you know, teaching at university? What got you interested and involved in all of this?

Yeah, great question. Well, I started a software company back in '97. All right. So, back then, the web had just come out. And I could see the writing on the wall. I was like, man, this is the industry to be in. And my idea was I'll sell the picks and shovels to the guys entering the gold rush. So we started building websites for people, and, and back then there was no WordPress, so we built our own web content management system. And, um, that was the problem that we solved for people for years. And then WordPress came out, and all of a sudden, you know, what became our competitive advantage became our albatross because everybody else had something free that they could use and we were supporting a CMS. So by then we had gone from charging an average of $175,000 for a website down to about $18,000, you know. And then gradually, I guess it was what, 2020 when we started messing around with CopyAI and Jasper, which I think was called Jarvis back then. And at the point I was like, okay, well, this is kind of interesting. You know, it gives you a head start, but it still talks in circles, you know. But I thought, okay, if this is the worst it'll ever be, you know, we could be in for some interesting stuff. So I had decided, you know, I was getting— my wife and I were getting close to being empty nesters as our kids were graduating from college, and I thought it would be cool to lean out the company instead of having 40 employees, shrink it. So I sold off the agency side. We gradually grew to be a digital marketing agency and a software development side. I sold that part off to, um, figure out what problems the market wanted solved now. And we, you know, we'd started to use AI to help us, um, have these roundtables. We started off with a thing called the CXO roundtable to just have meetings with various C-suite and talk about, you know, what problems they wanted solved. The idea was that I would help them solve marketing problems. But as time went on, I was like, this AI thing is getting kind of interesting because we were using it to write our copy and all that stuff. And then, as you know, ChatGPT came out and hit the world. I think it was version 3 and we were playing around with that. I'm like, they took the gates off, right? So now we got a GPT without guardrails. This is a whole lot more fun. So I changed the name of the group because everybody was interested in AI and I changed it to CXO AI Roundtable. And so every Friday we'd meet and we'd teach them new stuff about AI. We would compare different tools that were out there. I would have vendors come in and talk about their tool. And, you know, again, the idea was what problem does the market want solved? And eventually I figured out what the problem was. And we had gotten to be really good at prompting to make it sound, you know, less robotic, less ChatGPT-ish. And we developed a unique framework for doing that. And I would teach it in the roundtable. And then I thought, okay, well, I've been doing this for free for so long. I need to hurry up and monetize because there's only so long you can work for free. And so we did some training for people. And as we were doing the training, I realized people weren't getting a certain concept across. And the concept was what we now call scalable prompt engineering. And it's the idea that instead of the way you're traditionally taught to just type big giant questions or requests into AI, to break them more into components or modules, if you will, so that you could swap those out. And that way it becomes scalable inside your organization, right? So somebody could look at your prompt and go, oh, if I swap out the persona for this, I can target a different audience. If I swap out the style of voice for here, I can write a press release instead of a sales letter, you know, those kinds of things. So as I was teaching people, I realized they're, they're there was a, I don't know, some kind of void or something. They weren't quite getting it. And then it hit me that I needed to give them a visual diagram of how this worked. And years ago, I had a guy on stage at an Inc. 5000 conference talk about the business model canvas. I don't know whether you're familiar with it or not, but he wrote a book called Business Model Generation. If you're doing startups, you should go get Business Model Generation, that book, go get the business model canvas, and it'll teach you all about the various aspects of your business model. Well, that's what we used to teach our employees when they would come on. We'd like, this is our business model. These are the people that we serve. These are the suppliers we deal with. It's our revenue sources, our expenses, et cetera. It's a cool thing. But as I thought about it, I was like, wow, that's actually the perfect framework for me to teach people this principle of scalable prompt engineering. So I designed it all up. And the next meeting of our training, I went through it all with them. And man, the light bulbs just went on for everybody. And that's when I realized, okay, we're clicking on all cylinders now. So I taught a few more cohorts. And then the provost of LSU came over to see me because he'd heard what we were doing. And I showed him and he's like, whoa, I've never seen anything like that. Do you have the prompt structure to match? And I showed him that. And And that's when he's like, well, that's next level stuff. Would you be interested in teaching our honors students next semester? And I was like, I don't know that I want to become a professor and I don't know that we can have the impact, but I think we could probably have a better impact if we went to employers and tried to upskill their employees so that they had a fighting chance of staying in business with the onslaught of AI. So that's where that partnership came. And so we teach stuff through LSU's continuing ed now. Okay.

Love it.

Amazing.

Well, I mean, that sounds like a really cool kind of journey to, to getting into this. Obviously you had your finger on the pulse from a very, uh, from the very early age, uh, stages. I think that's something that I, I see with a lot of marketers was, uh, even myself, uh, you know, was using Jasper before ChatGPT came out. And, uh, to be honest, like it had so many limitations, but I was trying to completely maximize it. I had 2 virtual assistants that were on my account. 24/7 generating articles. I think I was on their $600 a month tier just because I was so happy with how much content. And of course, when ChatGPT came out, that pretty much nuked that because then I was getting the same thing for $20 a month. So I hope Jasper is still alive and kicking out there.

They're doing well. They've— they've— they've been good.

Yeah, that's good. They have a lot of cool integrations and stuff. So yeah, they're doing great. But yeah, amazing, amazing journey. And it's cool to see that you were kind of on that cutting edge and that's what kind of got you to, to jump in and, and make it happen what you're doing today. So my question for you, based off of what you're doing, you know, what are some of the common challenges that people are facing when they start using AI tools for, for business, for work? You know, you consult with a lot of different companies. What are some of the common hurdles that people are needing help overcoming?

Yeah, great question. I guess it depends on the size of the company. So, let's just take a small business to a medium-sized business, some company that has, I don't know, 10 to 200 employees. And I'm writing a book about this. In fact, a lot of the people that have gone through our training are now editing my book for me. And it should be, with any luck, I'll publish it next week. But, and it's a legit book, you know, it's not a pamphlet, but it teaches people how to, or businesses how to build what we call an AI-first culture. All right. The problem they're running into is that the majority of people are all self-taught. Okay. So they all pick up a tool like a ChatGPT or a Claude or Perplexity, and they got to figure it out. And so they watch a couple of YouTube videos or they go buy a 1,000 prompts for $9.99. And, you know, those are all— you could just ask ChatGPT to create those prompts. Eventually you figure out they don't work. But the bottom line is they're all self-taught. They all have different methodologies. Or if they've seen any training, they've all seen different trainings. So there's no common framework that they're operating from within. So that means it's not necessarily scalable knowledge. It's not shareable knowledge. And so the expertise kind of gets stuck in one person's brain. And what we try to teach people is, look, you know, if you really want to have an AI-first culture, then everybody needs to understand it the same way and needs to operate it the same way. It's like anything else in your operation. You have to have SOPs, you have to have standard operating procedures, and they have to be documented that way. If you have turnover, so somebody else can come in, instantly pick up and use it. It's just like writing code. If you've written code or whether you're using a no-code tool, somebody has to know what the code actually says, right? So when you're writing good source code, you put notes inside the source code so people can come in after you, right? And fix it. So we teach people how to write prompts like that. The interesting thing is people would think, oh man, that's gonna take so much time, et cetera. We actually can write prompts with 90% fewer words and get significantly better output. And that's what we teach people is, look, let's get everybody on the same page, understanding it from the same methodology, and now create a system where people can share it. Like I can show somebody a marketing prompt in HR and they can see how to hot swap some variables out and use that same prompt to do something different in HR. Without having to rewrite a whole lot of it from scratch. Does that make sense? I don't know if that answered your question. I got way the hell off track.

No, man, that makes a lot of sense. That's super interesting. So obviously this is something that you help people out with. You have your agency Bazooka that you are— you do a lot of consulting with. Something that I would be curious— I was looking through your website and I saw you made a claim there And I'd be curious your, your take on it. But that's kind of the concept of upskilling, upscaling, and skilling employees versus hiring new talent. This is probably something a lot of employees are concerned about. So this, whatever, you know, your, your take is on this might be interesting for people to either share with an employer, but yeah, what's your take? Is it better to hire a completely new team that's an AI-first team, or is it better to train your current employees and upskill them?

It's, it's far better to train your current employees. And here's, here's why. I mean, having run a business for the last 25 years, I can tell you that you invest a lot of money in an employee. You invest a lot of time in them to get them productive. It usually takes somewhere between 18 months and 2 years before somebody is really cooking on all cylinders. They know your culture, they know your processes, they know other people, they're connected, they become part of the family, part of the team, part of everything, part of the heartbeat of the company. To rip that out and replace it with somebody just because they know AI rips the heart out of your company and the soul out of your company, and it deflates the morale of everybody in there. So, and you still have to teach that new person your culture, etc., assuming you don't destroy it by ripping everybody out. But if I can take an A-performer employee or a B-performer employee or even a C-performer employee, because everybody's got a mix of all of them, if I can take that person and in 6 weeks, teach them how to operate AI at an incredibly efficient level. I've taken a C employee and I've made that person an A employee. And I haven't shifted anything with the culture or the personality of the company. I've stuck to my guns in terms of keeping people who are loyal to me, right? It's a whole different ballgame when you do that. But like I said, the minute you say to somebody, look, you don't know AI, but, you know, I got this, this person over here who's, who has been playing around with AI for 18 months, you know, I'm going to hire them over you. I mean, keep in mind, Jayden, most people think AI was invented in December of 2 years ago, right? Right. So how much better can somebody be if they only have 2 years of experience at AI, right? They would have to put in thousands of hours of work. But if you can teach them the right way, somebody in your organization, they don't have to learn how to build AI agents. They don't have to learn how to integrate all kinds of things through Zapier or Make or any of these tools. All they need to know is how do they do their job, the, the tasks that take time. How do they use AI to do that better? How do they use AI to essentially help them think through problems without bothering their boss? You know, how do they populate the AI safely with enough information to where they can use AI as a thought partner to make better decisions? That's really the foundation of what you need to teach the majority of your employees. It's always good to have a couple of people that know how to build automations for sure. But so many other people can just use ChatGPT or Claude or Perplexity, those three, to really do stuff really, really well. You know, you're not going to have somebody in finance that needs to create motion graphics, right? So they're not going to go hop on Midjourney if their job is to analyze spreadsheets.

So Totally. Totally. Okay. I love that. I think that's a great take. That is definitely the way to continue scaling your company in a sustainable way because yeah, your employees, they already know everything going on in your company. And sure, you could hire someone that's technically an AI expert, but you're gonna have to train that person on your company. So you're training someone on something. Either way, you probably have a huge leg up because anyone can learn these AI skills and then you don't have to train them on, you know, what your companies are proficient at. Something that I, I was looking at on your website is the scalable prompt engineering. This is something you teach and you talk about. Tell us a little bit about what that is and what the core concepts behind scalable prompt engineering, what they are.

Sure. Well, I'm a big spreadsheet guy. You know, you ever built a spreadsheet?

I love them.

You played around with Excel a little bit. Course. Yeah, I mean, who hasn't, right? Um, in a spreadsheet, like, if you want to— let's say you've got, um, 12 columns of information and you're tasked with saying, hey, what happens if we increase expenses by 10% next year? You're not going to take your monthly, uh, rows and columns of, you know, January, February, March, April, May, and you're not going to hardcode into a cell times 1.1 and figure out what that 10% growth is, right? That's hard coding it inside the cell. Now, if your boss says, hey, I need you to change that to 8%. Okay, well, now you got to edit each one of those cells and you're not going to do that. What you're going to do is you're going to take that 10%, you're going to put it up here in its own unique cell, and you're going to create a formula that says, I want this number to increase by that percentage, and then you're going to carry it across, right? That's all scalable prompt engineering is, is figuring out what those components are to a prompt and break them into variables or containers, what we would call them. So that when somebody looks at it, they're like, okay, I can just take that container. You know, if the container happens to be a big description of your target persona, you know, swap it out and put in a new persona and everything else will be the same. So you might have a container for persona, you might have a container for the style and brand voice. And inside the style and brand voice, instead of it just saying, I want you to talk in a nice conversational way with a little humor, well, there are ways to calibrate those. So instead of saying with a little humor, you can actually calibrate humor in a level of 1 to 10 or 0 to 10.. And so you can just say humor equals 1/10 or humor equals 10/10 and it will dial it back, right? We had a gal go through our training and then she was building a chatbot for an ophthalmology clinic. And the whole idea of the chatbot was to help teach new employees how to sell ophthalmology services. And she comes into one of our office hours sessions. She says, John, I need some help. I can't get our chatbot to stop talking like a pirate. And I was like, what do you mean a pirate? You know, like, yeah, a legit pirate, like arrr, matey, and all that crap. I'm like, oh my gosh, let me look at the prompt. So I go look at the prompt. And sure enough, she had used sentences to describe the tone of voice as opposed to a calibration method. But not only that, There's what we call prompt conflict. When you use paragraph prompts, you frequently get into prompt conflict. And what she did was she told the AI it had a role, and then she told it how to speak. Those things can conflict with one another. And what she did was she told it that she wanted it to speak with a little humor, kind of like that funny uncle. Okay, funny uncle now tells the AI it's talking to a child. And it's supposed to be humorous. And AI, for some reason, thinks all children think pirates are funny. And so it literally started talking like a pirate. That's hilarious. So we had to dial that back. But it was interesting because, you know, all she had to do was turn it into parameters, test it out, and say, hmm, it's being a little too stodgy. And so we just dial up one of these parameters and dial back another one., and it sounds perfect.

I love that. That's, uh, that's great advice and that's such a funny story. Something that I'd love to hear from you. I know AI use cases are different for everyone, um, but I, I feel like it's always really interesting to hear what your personal, some of the personal ways, um, some of your personal favorite ways to use AI today for your workflows, for your projects, for your tasks, stuff that you actually do. Like what is, What are the AI tools and what are the things that you're currently using it for?

Well, I'll tell you what, let's keep it at the very basic level. I mean, I could go into very, very complicated automations, but let's just talk about basic because the majority of people out there just want to use it at a basic level. I create a ton of Claude projects or custom GPTs, and all those are basically, you know, prompts that you would stick in something and then you just execute against it. But your project or your custom GPT can also hold a lot of information, you know, documents, et cetera. So I have one that I call Bazooka In-House Legal Counsel. It's got all the information about our company in there. It's got all of our tax IDs, our addresses, our board of directors, our shareholder makeup, and all that stuff. So So if somebody sends me a document, I can have the legal counselor look at it. Then I can also have it fill in all of the things that are blank on that document and send it back, right? I have one that I can take with me to like trade shows or conferences. Somebody gives me their business card. I take a picture of the business card. It immediately turns it into a dress book entry that I can put on my phone. Cool. I've got, well, heck, I've got a bunch of them that I use to write the book that I'm writing. Every time I modify a chapter, I upload that. That becomes more source material for me to continue writing. I've gone through, I don't know, 5 different versions of that. And I'm trying to think of some others. I love transcripts. So, like every meeting we have, we're doing it on Zoom and we're usually using like a MeetGeek or a Fathom to record it. Shoot, I don't have it with me. Well, this is a a wearable version of it. Are you familiar with a PlaudNote? P-L-A-U-D-Note? Uh-uh. It's an AI note-taker. It's literally the size of a credit card. It's twice as thick as a credit card, but it is the exact dimensions of a credit card. And it'll record meetings and then it'll digitize them, the transcripts, and it'll analyze them. I've created my own GPTs to do different things with those transcripts. A lot of times it's just, hey, this is a meeting transcript, so I just Instead of telling the AI, hey, I want you to take this transcript and do X, Y, and Z, I literally just paste the transcript into that GPT and hit go, and it does everything I want it to do. Sometimes what I want it to do is take that transcript and craft a case study out of it. Like if I'm interviewing a client or a student or something like that. So I take the transcript and it will instantly write up an amazing a detailed case study based on the transcript. I don't have to do anything with it. Blog posts, all that stuff, you know. So there's a million different use cases. Like, we have 9 different styles of blog posts that we write, you know. You got listicles, you got thought leader posts, you got News Jack posts, you have all these different things. Um, as you know, you can't tell any AI tool, hey, write me a 1,500-word blog post about topic X. It doesn't, it doesn't work. It doesn't know how to count. Uh-huh. If you know what the right workflow is and you know how to run it through the workflow, it'll write a 2,400-word blog post and it'll be really good and it'll inject it with all of your thoughts, et cetera. And it'll be written towards your persona and it will retrieve current information to inject in it. So those are just some of the use cases.

Oh, I love them. Those are fantastic. Uh, I bet that, that'll give people a lot of, uh, get their wheels turning on a lot of interesting use cases. Something I'd be curious to ask you about because, you know, you upskill a lot of people, a lot of companies and employees, but like suppose someone today listening is in the, in the position where they're looking for a new job. What do you think the most, what do you think the best way on a resume to showcase that they do have AI skill sets and perhaps even in an employer right? They've upskilled all the employees, but they still need to hire more people. So they're looking for people that would be good at AI. So this question could be for both sides of this, but like, what are the things people should be looking for on a resume or putting on a resume, the titles or signifiers that someone they're hiring already is, is kind of knows what's going on with AI?

Yeah. So I mean, that's a really good question. There's several ways to go about it. One, if they go through one of our courses at LSU, LSU offers them a certificate that they can put on their LinkedIn profile, you know, for whatever that's worth. The other is, look, when, when you're trying to get a job, your resume and your LinkedIn profile need to talk about what you have accomplished and how you've impacted your employer. They don't need to look like a list of tasks you performed. So the first thing I would do is I would rewrite my resume, or at least the last job experience, and say how I I specifically used AI to impact the employer. And if I hadn't used AI, then I would say, hey, I went through this course, this training, and I'm now certified to do X, Y, or Z. Or you could just say, I've put in over 3,020 hours working with this product, that product, and the other product. But, you know, somebody needs to be looking at it and go, wow, okay, this person has done something, has put in some hours and should know something, right? If you're trying to get into the consulting game, there are businesses out there that'll train you how to build AI automations or how to be an AI coach or consultant. Are you familiar with chiefaiofficer.com? No. So Chief AI Officer, they have a certification where you can become a chief AI officer. So you're certified at that level. We offer a certification in our process. It's Ingrain AI. And so we teach people how to run strategy meetings, how to run interdepartmental collaboration meetings, and how to do scalable prompt engineering. So that they can take that and they can turn around and coach other businesses. So those are, you know, some of the ways that you could do it. At a minimum, you need to be able to describe to somebody how you use AI in multiple circumstances to simplify something to where that future employer can say, ah, this person is thinking AI, knows how to execute it, and has done something that simplified something that will translate to me as an employer.

Okay, I love that. Yeah, no, I think that's really valuable, actionable, uh, advice. Um, something I would love to ask you about— you've had your finger on the pulse for quite a while. You're an early adopter in this, um, in the space, as we kind of talked about. Based off of that, what are some predictions you would make about AI and what we're going to see in the AI landscape in the next year that maybe people aren't thinking about or talking about that people listening should look out for?

So I think, you know, I think 2025 is going to be the year of the AI agent. It's going to start to explode towards, especially towards the latter half of the year, which means to me in the front half of the year, if you really want to make some money in AI, learn AI security because that is going to have to be at the forefront of everybody's mind. How do we make sure if we let AI loose inside of our desktops to where it has access to our passwords and can connect stuff and actually execute stuff, we got to make sure that's pretty secure, right? That's the part that spooks me right now. But I think, you know, agentic workflows are amazing because it's more than just a logic tree, you know, a decision tree. When AI can actually intelligently interpret stuff and take actions, that's pretty cool. The scary part is like if I were to say, hey, if I receive an email from such and such, I want you to go ahead and put it into QuickBooks. And then I want you to go look at QuickBooks and see which bills are due, and then I want you to turn around and pay the ones that are due now through ACH, through our accounting and banking system. All right. So now you've basically given an AI agent the keys to the kingdom. And if you get a rogue person, like, for instance, 2 weeks ago, I can go back in history. 7 years ago, somebody sent us— this is pre-AI— somebody sent my accountant an email masquerading as me. It was clearly a spoofed email, but it says, hey, I need you to quickly wire transfer $19,000 to this, this, this company so we can do this deal with them. Let me know what you need. She just hit reply instead of looking at it and going, this is not John's address. She hit reply and she goes, okay, send me what I think. And then he sent the wiring instructions. She wired out $19,800. Oof. Boom. That's pre-AI. Okay. Uh, we were fortunate enough to get the money back because the scammer's bank accounts had been closed right before our wire hit and our wire bounced. It took about 10 days, but we got our money back. Fast forward to 2 weeks ago, somebody hacked into my bookkeeper's email account. Took a— sent an email to one of our clients, said, I need you to change our ACH payment instructions. Can you send me what I need to get that done? That person replies back— well, this person had hacked in and created rules to where all that email was hidden from her view, but whoever this person, bad actor's view, wasn't. Bad actor takes the ACH form— it was a PDF form— attaches my actual signature to it. So clearly that person had access to something inside our network. I don't know what. Attach my signature to it, send it to the client. The client makes the change. Once again, we're blessed in that bad actor's accounts got closed. So the payments bounced. Client calls us and goes, we're trying to send it to the new address you sent. My accountant says it, sends it over to me. Hey, do you know why this is bouncing? I'm looking at it. I'm like, yeah, that's not our bank account. Oh my God. We tracked it out. We figured all that. Okay. Now imagine in the world of deepfakes, right? I don't know whether you were familiar with the deal that went on. I think it was in February of last year. Hong Kong guy wires out in a Zoom meeting. A deepfaked Zoom meeting with the CEO on video gets 5 guys to wire out $10 million each. I mean, okay, I wish I had that kind of a problem, you know, if I had $10 million, if 5 guys could go wire out. But still, that's crazy to me that, you know, they've literally faked the CEO on a video conference. So crazy. Yeah, so imagine, you know, you get AI operating your actual equipment and making transactions for you and you get a bad actor in there. I don't know how you fix that. That's why I say, I think if I were going to be making money, I'd figure out how to get into the security game for AI.

Yeah, that makes a lot of sense. It's going to be a really, really interesting, an interesting game because yeah, there'll be whole scams and people setup just to try to trick your AI bookkeeper, and everyone's like, oh, I got an AI bookkeeper, it's fantastic. But there's going to have to be a lot of levels of security on that for sure. What are some of the— I'm talking about AI and agents— what are some of the most exciting use cases, um, of AI agents, uh, that you're excited to use or you're excited to kind of see out in the wild?

Yeah, um, God, there's way too many to even ponder right now. But, you know, since I'm primarily in the marketing space, The thing that I like most, I like the idea of being able to just dictate my thoughts into something and then having those thoughts then be analyzed by the AI. And then the AI decides that I just tell it to go write a blog post, that I just tell it to write an email to so-and-so, or did I just tell it to, you know, book an appointment with somebody? And then it logically will branch out and execute all those things for me. So literally, it's an assistant for me. It knows what I just told it to do, and it will then interpret and execute and take care of all that stuff. And it'll sound like me when it's done. You know, I've got friends of mine that are recording videos. Well, actually, I just pinged a buddy of mine. About 2 hours ago, I was scrolling through YouTube in a mindless moment of my life, and I saw one of his posts. And I was like, I'm watching it. I'm like, okay, I think he did this with HeyGen or, or I forgot what other tool. But you kind of tell it sounded like him. It looked like him. It moved like him. But there were these little parts where you're like, he wouldn't have, he wouldn't have made that movement with those words. So I just, I took a screenshot of it, sent it to him. I said, like, is it you or is it AI? He goes, nah, it's totally AI. I was like, it's great. So they literally, he showed me his workflow, massive agent workflow. But he literally just types out what he wants and it goes into You know, it goes into Descript, it goes into ElevenLabs, it goes into a couple other tools, produces this video. And next thing you know, he's got a YouTube Short on there that promotes something and it's him talking with a screen behind him of something happening. Crazy. So that's crazy.

Yeah. Very cool. That's amazing.

Yeah.

Marketing is going to be such a phenomenal, phenomenal area for these agents. No doubt. What's your thoughts on— because I know a lot about what you do today is helping to upskill people with AI tools. When we start talking about agents, I think a lot of people are concerned about job security. I mean, you just gave a great use case of an agent becoming an assistant. I have a lot of virtual assistants currently that work for me that do kind of small tasks that aren't super complicated, but they're tedious and repetitive. And I can imagine AI agents could do a lot of those tasks. I already come up with an SOP or a training on how to do it. I hand it off. So that's not much different than perhaps prompting an AI agent. What do you— what's your thoughts on the impact to employees, employers? Are we going to see a lot of layoffs? Are AI agents going to take a lot of jobs? Are there going to be more jobs created? How do you see the landscape of the job environment with something like this?

Yeah, I mean, you know, if you want to span out 5 years, you know, it's like I was telling our team this morning, we had a mastermind meeting this morning and I was telling them, I think, I mean, the main reason I'm writing this book is so that the employers that engage right now and try to develop an AI-first mentality, they're going to be the ones that protect their margins and they'll do a lot better. But the more businesses do this, yes, you're going to see layoffs because you just, I've got that my opening introduction to the book talks about it. You do something and you identify some workflow and all of a sudden what used to take you 2 weeks takes you now 4 hours. Okay, cool. Now, what do you do with that employee's time now that they only take 4 hours? Well, you could do a couple of things. Do a lot more, you know, in that 2 weeks. So that person does, you know, 8, 10 times as much stuff now, but you got to have the ability to go sell that stuff. Or you could just keep that employee on staff and then teach that person to do other tricks, right? There's a period where you're going to have a nice, healthy profit margin until competitor B sees what you're doing and they do it and they say, well, we're not going to keep the margin. We're going to cut the price. So now you got to compete against price. You've already got your margins and you're like, okay, well, I'll cut price too. Well, then this guy does the same thing. The next thing you know, now you scale that across the landscape and you have a whole lot of people cutting price, which means eventually they're going to have to cut people. The more people you cut, the more you actually shrink the economy because there's a lot of disposable income that's not getting reinvested the economy. And then the question is, when does it come out? So the question really becomes, how soon does the trough hit? How wide is the trough? And how realistic is it that it comes out? A lot of people saying, oh, you're going to have more jobs created. You know, in, in every disruption, that's, that's true. This one's a little different because this, this disruption, um, the faster it grows, you know, you think about AI agents is one thing, but robots is another thing. Self-driving cars is a simple example, but I don't know whether you saw, what is it? I forgot what, all these $30,000 humanoid robots that you can buy. It'll really go around your house and do stuff for you. Okay. The more you do that, I really don't have a prediction on it. I know that there's going to be layoffs. The question is, What are the new jobs? I think in the short run, and this is why I'm trying to teach this as fast as possible to people, I think in the short run, the more you know about AI, the safer you're actually going to be because you'll be able to take the knowledge and expertise that you already have and you'll be able to make sure that the AI agent delivers that, right? You know, I always say, look, if you don't know what excellence looks like, you're always going to get average or below out of ChatGPT. You know, you got to know, like when it comes to copywriting, if you don't know what excellence looks like in copywriting, you're going to just settle for what it puts out there and then you're going to sound like everybody else. And at some point that becomes nauseating. That's why I'm scared for marketing for two reasons. I'm scared for marketing because so much of what can be done can be automated. So it makes a, the role of being a marketer very difficult to sustain in the view of automation and all that stuff, right? But on the other end, if you're a marketer that knows what excellence looks like, then you can make sure you're the one overseeing the agent and overseeing the output and confirming that it is outstanding. Same with finance, same with HR. You've got to have some chops. In order to make sure that what you're asking AI to do is really good, right? The question is, how many of those people do you need in an organization?

You know, I think that's, that is the critical question of the time. So it's going to be really interesting to see. There's so much innovation coming this year, next year. We got an exciting few years ahead of us, no doubt. John, you ready to wrap up? The last thing I want to ask you about is what is one piece of advice you feel like you could share to people listening when it comes, you know, to preparing for the oncoming wave of AI, when it comes to, you know, upskilling yourself and becoming better with these tools? What's a piece of advice you feel like you could share with people?

Man, the most that I could tell you is you gotta dive in headfirst and you gotta do it now. You know, you've got to do it. The thing that you can't do is be distracted by what I call the parlor tricks, right? You know, the, oh, look, I can take a picture of what's in my fridge and ask it what to cook for dinner. You know, that's not got a business use case for it. So you have to think about what problem you can solve for your employer or future employer. And you need to dig in and figure out how can AI help me with that. And at a minimum, you need to be figuring out how to use a tool that would impact your current role. Now, you know, if it's marketing, there's a myriad of tools. If it's HR, you know, some of them will be embedded. But at a minimum, figure out how to use Claude, how to use ChatGPT, Perplexity, and Gemini at a minimum. From there, then I would go and look at— make sure that you know how to use the AI note-takers. For sure. They saved my life a number of times. You can get into image generation. There are lots of the tools that make your life easy out there. But I would probably also say if you can find a mastermind to join where somebody boils all that down for you, that would be cool to find one that's in your space or in your zone. You know, if it's a marketing one, cool. If it's a copywriting one, cool. But, uh, it's hard to drink from the fire hose, right? You know, you just sit there on YouTube and spin through it all day, you'll start to panic. Um, if you're in a group where there's somebody's doing that drinking for you, it, it helps.

100%. That's some phenomenal advice. Thank you so much for sharing that, John. If people are interested, or if companies are interested in having you, um, consult with them, help upskill their employees or get in contact, what's the best way for them to get in contact with you?

Oh, well, thanks, Jayden. I appreciate it. Um, so the best place— I'm on, I'm on LinkedIn, JW Munsel, uh, or bizuka.com. Bizuka is B-I-Z-Z-U-K-A.com. And I tell you what I, I'll do for your listeners, um, and I, I only have a few left that I can do this for, but if you go to bazooka.com/ingrain, I-N-G-R-A-I-N. You can get on the waiting list for the book. And I'll— if they mention that they heard me on Jayden's podcast, then I'll send them a book for free.

Oh, amazing. Very cool. And yeah, give me that, give me that link after on— I'll— to the, to the listener, I'll have that link in the show notes. So you'll be able to click and go there directly. John, thank you so much for coming on the show. This has been phenomenal, a great conversation. Really excited about your book, everything else that you're working on. And thanks so much for sharing all your advice and insights with us. We appreciate it. To the listener, thank you so much for tuning in to the AI Chat Podcast. Make sure to rate us wherever you get your podcasts. Give us a like and a thumbs up over on YouTube, and hope you all have a fantastic rest of your day.