552 episodes | 520K+ downloads

Supporting image for Scaling AI Strategies for Business Success with John Munsell
John Munsell Scaling AI Strategies for Business Success with John Munsell
AI Agent

In today’s fast-paced world, Artificial intelligence isn’t just a buzzword—it’s a necessity. But as AI continues to grow at lightning speed, many small businesses are struggling to keep up. I sat down with John Munsell to break down the truth about scaling your business with AI. We explore how implementing AI without a clear strategy can lead to confusion, inefficiencies, and missed opportunities. Ready to get your team up to speed?

Scaling AI Strategies for Business Success Summary

Key Takeaways

  • Scalable Prompt Engineering: John explains how to structure prompts that can be used across different departments, turning ad hoc AI usage into a standardized, repeatable process.
  • The AI Strategy Canvas: Learn about John’s nine-block framework that simplifies the development and execution of AI strategies within your organization.
  • Cross-Functional AI Collaboration: Discover how creating prompt containers allows different teams—like marketing and HR—to use the same AI foundations for their unique needs.
  • Cost Efficiency in AI: By reducing prompt lengths and minimizing unnecessary tokens, businesses can save on AI costs while improving output quality.
  • Avoiding Common AI Implementation Mistakes: John highlights the pitfalls of unstructured prompts and how standard operating procedures can help scale AI effectively.

Scaling AI for Business Success: Key Strategies to Streamline Your AI Implementation

AI isn’t just the shiny new tool on the market; it’s a powerful ally that can transform how your business operates—if you know how to scale it effectively. Yet, many companies dive in without a clear strategy, creating more chaos than clarity. Let’s break down the key steps to scaling AI across your organization so you can make smarter, faster, and more efficient decisions.

Why Scaling AI Matters

The true power of AI lies not in isolated experiments but in creating repeatable, scalable processes that enhance operations across all departments. The problem? Most companies jump into AI without a strategy, leading to inconsistent results and wasted time. A structured approach ensures that your AI initiatives are more than just one-off projects; they become integral parts of your business that drive growth.

  1. Scalable Prompt Engineering: A Standardized Approach

One of the biggest challenges in AI adoption is prompt engineering—how you craft instructions for your AI models to follow. Most businesses create prompts on the fly, which works fine if you’re flying solo. But the minute you need to integrate these prompts across different teams, like marketing, HR, or sales, things get messy. That’s where scalable prompt engineering comes in.

Instead of treating each prompt as a standalone task, you should structure them using variables that can be easily swapped out. This method, often called prompt containers, allows anyone in your organization to adapt a prompt for their specific needs. For instance, a marketing prompt can be adjusted by HR simply by changing the target audience variable, turning a blog post prompt into one for writing job descriptions.

  1. The AI Strategy Canvas: Your Roadmap to AI Success

Scaling AI isn’t just about the prompts; it’s also about having a framework that guides your entire AI strategy. Enter the AI Strategy Canvas, a nine-block tool that helps you plan, execute, and standardize AI use across your organization. This canvas aligns every department by breaking down key elements like target audience, context, and execution strategy, ensuring that everyone’s on the same page.

The beauty of this approach is its flexibility. Whether you’re planning a marketing campaign, developing a new product, or improving internal processes, the AI Strategy Canvas provides a clear roadmap for implementation. It’s about creating a unified language for AI that all departments can understand and utilize.

  1. Cross-Functional Collaboration: Breaking Down Silos

One of the most exciting aspects of a scalable AI strategy is its potential to break down silos between departments. AI prompts built with interchangeable variables mean that what works in one department can easily be adapted for another. Imagine HR leveraging a marketing prompt to craft compelling job descriptions or sales using customer insights generated by a prompt designed for product development.

This cross-functional collaboration not only improves efficiency but also encourages innovative thinking. When every team uses the same structured prompts, it becomes easier to share insights and strategies, ultimately driving better business outcomes.

  1. Cost Efficiency: Doing More with Less

We’ve all been there—writing long-winded prompts that feel like you’re throwing words into the wind. But with AI, every word costs money. AI models, like GPT or Claude, bill you based on the number of tokens (words) used. By scaling your AI prompts efficiently, you’re not just saving on costs; you’re also saving time. Fewer words, more impact.

Scalable AI prompts reduce the number of iterations and manual adjustments needed, cutting down the time and cost associated with refining your AI outputs. Imagine achieving the same—or better—results with a 90% reduction in prompt length. That’s not just efficient; it’s a game-changer.

  1. Avoiding Common AI Pitfalls: Standardize or Stall

The biggest mistake businesses make when implementing AI is not having a standardized process. When every team uses different methodologies, you end up with a disjointed AI strategy that’s impossible to scale. This ad hoc approach leads to inefficiencies, errors, and, ultimately, frustration.

To avoid this, develop standard operating procedures (SOPs) for your AI use. This means creating a prompt library where every prompt is categorized, structured, and easily accessible. Using tools like Notion can help you track prompts, making it easy to refine and re-use them. The goal is to turn prompt creation into a standardized process that anyone in your organization can follow.

Final Thoughts: Making AI Work Smarter, Not Harder

Scaling AI is less about the technology itself and more about the strategy behind it. By creating scalable prompts, using a strategic framework like the AI Strategy Canvas, and encouraging cross-functional collaboration, you can make AI work smarter, not harder, for your business.

Standardizing your approach to AI ensures that your entire team is on the same page, working towards the same goals, and achieving consistent results. It’s not just about implementing AI; it’s about integrating it into the fabric of your organization to save time, cut costs, and drive sustainable growth.

Ready to scale your AI strategy? Start with a plan, structure your prompts, and watch your business soar.

Scaling AI Strategies for Business Success Episode Transcript

Rich: My next guest is the CEO of Bizzuka, a company that has appeared on the Inc. 5000 company list three separate times, empowering businesses to implement AI strategies that deliver two key outcomes, make money, and save time. Oh, those are good outcomes.

Since July 2022, he has guided over 600 business leaders in scaling their operations. His groundbreaking AI Strategy Canvas and Scalable Prompt Engineering methodologies are so impactful, top institutions like LSU have made them mandatory in their AI curriculum.

A serial entrepreneur who successfully built and sold his business, he knows firsthand that strategic AI implementation is the key to scaling in today’s business landscape. His passion lies in helping companies harness AI’s full potential across their entire organization, catapulting their business to new levels of speed and efficiency, while retaining and upscaling as many employees as possible.

So let’s jump into scaling your business through AI with John Munsell. John, welcome to the podcast.

John: Hey, thanks Rich. Great to be here.

Rich: Let’s talk about this AI Strategy Canvas and Scalable Prompt Engineering. What exactly is Scalable Prompt Engineering?

John: Great question. So the majority of people are out there teaching you how to write prompts, which essentially is what I would call conversational prompts. Which is great if you’re just a one-man band and you’re trying to do a few things. But the minute you need to have other people in your organization use those same prompts or use the foundation of those prompts, you need to learn how to scale that, which means they need to be structured in a really simple yet understandable way.

So you could do what we call as hot swap variables. So we create these things called variables, we create prompt containers and prompt stacks, so that anybody could follow in behind me and they know exactly what I’m trying to accomplish with it, and they could look at the prompt and say, I could use that for a different purpose. In other words, somebody in HR could look at a marketing prompt and go, oh, I like what you’re doing here. I could use it instead of using it to write a blog post, for instance, I could use it to write a job description. All I have to do is swap out these variables. Does that make sense?

Rich: Yeah, talk to me a little bit about those variables. Because I can understand where having different members of the marketing team I’ll use the similar prompts, but all of a sudden taking it over to HR seems like this huge jump to me. So what do these variables look like?

John: We created a framework that we call the AI Strategy Canvas. Now, we use that canvas, it’s got nine blocks in it. And for all your listeners, I’ll show you how you can get a copy of it at the end. But the nine blocks, first of all, you use them to discuss the development of AI strategy in your organization. So it helps everybody get on the same page and have a good conversation, and they know all the ingredients they’re going to have to take place and all the areas they have to discuss.

Then you go into the interdepartmental discussions or the cross functional collaboration to actually build out an AI initiative, and you use those same blocks to have that discussion. But then when you go into executing, and somebody has to build the prompts, those same blocks come into play.

So for instance, block number one is the target audience in marketing, we would call that a persona. All right. In HR, it’s still a persona, but it’s a different one than say, marketing. Marketing might be targeting CEOs of businesses with more than 500 employees that are in the oil and gas service space. Marketing, on the other hand, is going to have a persona if they’re turning around and they’re trying to recruit, let’s say they’re trying to recruit a new accounting manager. That persona has problems, has issues, has expertise, has, Things they want in a job. So anything that HR would create in terms of a job description would need to really appeal to that person.

So you build out that. We teach people how to build those blocks out. And sometimes they’re literally just stacks of variables that talk about what are their goals, their fears, their mistaken beliefs, their processes for making a decision. Those variables turn into a block or what we would call a prompt container. And that prompt container could be used in any number of circumstances. But when somebody sees how that’s structured, like I say, going from marketing to HR, marketing to sales, or marketing to finance, they go, oh, I have a container already built for target audience. I’m going to swap out the target audience. But your request, which is block number nine, has me executing something. I can do the exact same thing, but I’m talking to somebody else. I don’t know whether it’s easier to see it than it is to describe it.

Rich: What would you say are some of the benefits of using this approach compared to maybe the ad hoc approach that people have been using till now?

John: Yeah, so there’s several. One is that it may if everybody in or in your organization understands how to craft prompts the same way, it becomes far more efficient.  But if you have everybody in your organization learning a different way from a different instructor using a different framework, different methodology, now you got a big set of chaos on your hands rather than a set of efficient rules and regs, right?

It’s just like any other business. You have to have standard operating procedures if you want to really scale up the business and run efficiently. So that’s one of them where it helps people all get on the same page and create a methodology that’s easily replicatable and easily understood.

The other is that I can teach anybody to come into my organization and I can teach them this methodology. And the way they get results is far more effective. And then if you really want to get into it, it goes down to cost. The number of words you put in a prompt is like what they call tokens. We can get into the technical explanation of tokens. But when you start getting billed by ChatGPT or Claude or whoever for using tokens, you want fewer words, fewer characters to mitigate that.

Now, granted, the costs are going down as the models improve, but if you think about if it takes me 600 words to get the AI to do something, but I could do that same thing and get better results in 60. It’s not just the cost of the token. It’s the cost of how long did it take you to write 600 words to get the AI to do this?

I literally can take it and do the exact same thing in 60 words rather than 600. So I can reduce the amount of words it takes to process and get better results. I can reduce it by 90%. So those are the benefits.

Rich: And those are obviously strong benefits. If people haven’t seen your AI strategy canvas, the AI skills builder and all this, is this its own AI or does this sit on top of popular AI LLMs that we have today?

John: Yeah, it’s a framework is all it really is. It’s a foundation where we explain people a better way to structure prompts and a better way to think through what you’re doing so you don’t get… you know if you’ve ever wrestled with any kind of an AI tool, whether it’s ChatGPT or whomever, you’re going back and forth and back and forth and back and forth and asking questions. At some point you get the answer that you want. You get it styled up, whether it’s a blog post or some kind of ad copy or a job description or a research report, whatever it might be. You eventually get there. You don’t want everybody in your organization going through all of those iterations. But once you get there, you can go back through your string and convert that into what again, what we would call a scalable prompt. And so now the process gets reduced to a couple of steps instead of 15, and now it becomes repeatable.

And there are other ways that we would put it together. For instance, I’m writing this book. The book is called The End Grain, and it’s about building a culture of AI first in your organization and helping people understand to go to AI first and then scale it. Well that book, the first thing I had to do was write an outline for the book, and then I had to put meat on the bones for the outline. I would put that outline inside of a GPT and in ChatGPT, but in Claude, they came out with these things called projects, which is the same thing.

So I put the outline in there and then I would have it start writing against the outline. And then I would put in there a copy of the AI strategy canvas and some documentation that I wrote on that. And then I would put a few other documents in there and then I would have it. Query all of that inside of this project to continually write the book. And then as I would write more chapters in the book, I would add them to the GPT or to the Claude project, which becomes this growing knowledge base. So the more I layered it up, the fewer prompts I had to use to generate a chapter or a lesson or whatever it might be. Does that make sense?

Rich: Absolutely.

John: Yeah. So that’s the idea. It’s just over time you end up with something that becomes so simple and so scalable that it just makes far more sense for your organization.

In our organization, we have a prompt library where we have probably 400 prompts that we share. And if we’re writing a social media post, they know which one to go to. Like for instance, one of the blocks in there is the style and brand voice. And so we teach people how to calibrate that. Not just by saying, use a friendly conversational voice. We actually have variables that we use, like 50 some odd variables that we can calibrate from 1 to 10. So instead of saying, use a little bit of humor, I can just say the humor level should be 2 out of 10. If I say it’s 8 out of 10, but you get the idea. The idea is to really calibrate what you’re going to say so you only have to do it once, and you don’t have to go back and forth with the AI.

Rich: What I love, because I was looking through your notes before we came on the show, is that you’ve got a sarcasm meter too. Mine personally would be off the charts, but I thought that was a great variable to include.

In your AI strategy canvas, you outlined several components critical for AI implementation. Can you talk to me about a few of these components, and particularly how context and role aspects influence AI strategy?

John: Sure. So the first thing that you have to do is the target audience. So that’s primary. Because everything you do has an audience for it, whether it’s your shareholders, your stakeholders, your employees, your target, whatever it is. That’s the number one thing is first identifying who it is you’re speaking to who’s going to be the beneficiary of the AI.

Context, in most prompts out there, they write a prompt so that every sentence in the prompt is context. But if you break context apart, so you have a target audience, which is context, information about your company is context, information about your products or services, context, right? Resources that you’re going to pull in are context. The style and brand voice is context, and the role that you want the AI to play is potentially context.

What we do is we separate all those out so that now context just becomes the additional information beyond those things that you need to give the AI so that it knows what you’re framing up. So if I’m writing a blog post, I already have a stack that talks about my target audience. I already have information about my company, my products, and services. Context then becomes, what are my thoughts on the subject? What do I think about this? If I’m looking at somebody else’s article and I agree or disagree, I’m going to give it that in a context block.

But it already knows that because of the other blocks, who I’m talking to, how I’m going to roll in my company information, how I’m going to position it against my products, I don’t have to write all that all over again into the prompt. The role is what we’re asking the AI to be. Who are we hiring? Are we hiring a copywriter? Are we hiring a physician? Are we hiring a legal analyst? Are we hiring a financial analyst? That’s the role.

The problem that most people get into is they get really descriptive with the role, and they don’t understand that they just created what we call prompt conflict. Because when you give the AI a role, AI assumes a lot of characteristics around that role. So for instance, if you said, “I want you to be a chemical engineer, and I want you to use a humor level of nine”, it will not be funny because it already assumes a chemical engineer has no sense of humor, right? It’s just going to think this is a serious thing. If you were to say that you’re a chemical engineer trying out for a standup routine, it’ll start to get funny and it’ll create some goofy stuff, but it’ll be on the nerdy side because of the role that you gave it.

Does that make sense? So we want to be sure that we don’t get overly descriptive because the name itself has context embedded in it that we don’t know about. So that’s where we have all these other variables.

Rich: It’s always a challenge on whether you’re giving enough information to AI, or you’re not giving enough information to AI. So I can see how that would be helpful. And the prompt conflict is a good way of approaching this.

You talk a lot about your prompt libraries, and I’m just curious, because I’ve attempted to do this myself with my own company with mixed results. Are you using more of the, I guess on ChatGPT is the custom GPT’s and on cloud it’s the projects. Are you primarily creating custom GPT’s to use them, or are you pulling in that content every single time that you want to start a new conversation with ChatGPT?

John: Great question. That scale is starting to shift. It’s probably 90% of the time we’re just pulling in a prompt and rolling with it. And 10% of the time we’re building a GPT. But as we start to identify more repetitive tasks that call the exact same information, we’re starting to create more GPTs or Claude projects.

And then the next thing is to basically link those together in what’s called the agentic AI. So it becomes essentially an agent, and the agent knows, based on what you’re asking it to do, whether it needs to call in this GPT or that GPT to process the next step.

Rich: So if I’ve trained it to speak in my voice, and I’ve given it some examples of my writing, some emails, some blog posts, a chapter from my book, so on and so forth, I have it tell me like, what does Rich Brooks sound like, I get all that as a document. How do I keep from having to feed those six documents to ChatGPT every single time I want to have it write in my voice? What are the ways that are most efficient, based on what you’ve learned, to make that happen, whether it’s in ChatGPT or in Claude?

John: So in ChatGPT, there are two ways to do it. One is in custom instructions, where it just automatically is there no matter what session you’re running. And the other is inside of a GPT. You create your style parameters inside the GPT.

I’ll give you, for instance, we have a GPT for HR, we have a legal GPT, we have a copywriting GPT, we have a book writing GPT. If I’m going to write a blog post, we have a copywriting thing, but we have unique post prompts.

So we have one that we call a news jack post. We have a listicle post. We have 10 different blog post prompts. But by running them inside of a Claude project or a GPT, we don’t have to rewrite all the style parameters because they’re part of that. But the way we write for LinkedIn is a little different than we would write for, say, Instagram or Twitter, right? So we have to have all that kind of spelled out in order for it to do it.

And that’s what I’m saying that like the way I write in an email is a lot different than I would write in a blog post. So I’m going to call a different agent. And it may be that I would say, “I want to write a blog post about topic X” – I’m getting really simple, it’s way more complicated than that – but I would write a blog post and then I’d say, when you’re finished, I want you to create a promotional post on LinkedIn, Instagram, and Twitter, and then write an email to the people in our mastermind group saying that we’ve got all this each one of those could trigger a different agent that knows what to write and then spits all that out. So it’s becoming a little more simple and yet complicated at the same time, instead of having to just spit out this giant prompt that tells it to do all this stuff. We separate it out so that it can use a different voice each time that’s consistent with our brand, but it’s also more consistent with the audience we’re trying to reach then than anything else.

Rich: Right. Like I behave one way on LinkedIn than I would on say TikTok. So it makes sense, even if it’s still my voice that I’m going to behave differently.

I have followed almost everything you’ve said, I’m proud to say. But when you talk about an agent, can you give me a definition? When you say ‘agent’, I’m not as familiar with that term. What does that mean exactly?

John: So in its simplest terms, it’s basically a GPT. It’s a GPT that’s potentially connected to other ones using Zapier or something else. So the minute it comes into something that it knows isn’t its lane, it sends the request to another GPT who’s in that lane, and then it waits for information to come back.

So I could do a number of things with it. I could, for instance, if I was just doing expense reports or something, I would connect it to all the things that handle the expense reports, and I would just upload a receipt and have it rock from there.

Rich: So agents allow, using Zapier, allow different custom GPTs to then work together.

John: Talk to each other. Yeah. And sometimes, yeah, there are lots of tools that you could use that’ll do this. And sometimes you’re connecting them with Zapier, and other times they’re going to manage it themselves.

There’s a tool out there you might be familiar with, Make. You might be familiar with Cassidy. You might be familiar with Respell. All of those have different integrations. They’re all essentially startups. Just about everything in the AI world is a startup. But they’re interesting in the way their user interface allows you to connect different apps. They don’t seamlessly connect with every app out there, but they’re usually pretty easy to use connecting with some of the bigger ones.

So if I wanted to roll something into Claude, because I like the way Claude writes better, Claude can’t search the web currently, right? ChatGPT can. Perplexity can. I love Perplexity. And Perplexity now is responding really well to our prompt formulas, and it’s writing some really good copy. So that’s interesting.

But if I prefer Claude, I would create an agent that takes the request that I’m trying to do, maybe it’s ChatGPT. And then it says, okay, now I need you to go out and use Claude to take the information that I just had ChatGPT create, shove it into Claude, have Claude rewrite it and do something with it, shove it into a Word document, and then come back to me and take that document and do something else with it to shove it into a spreadsheet. I’m making stuff up at this point.

Rich: Sure, no, but it sounds Borderline magical. Like when science is so far ahead of where we are now, it just seems magical sometimes.

John: So let me give you a quick example. So really quickly, one of the things that we’re building is how you get all these AI newsletters and stuff. I get a ton of them, just tons of them. What I don’t have time for is to read them all. And not only that, not all of this junk applies to me. I really don’t care to do certain things that come across in these because they’re not in my wheelhouse. They’re not in my business. I don’t really care to make recipes, for instance. I don’t.

But what I would like to do, and we’re almost finished with this, is email comes in, I got a rule inside of Outlook that says, ‘this is an AI newsletter, send it into the AI newsletter agent’. The AI newsletter parses all of the stuff in that, and it knows who our target audience is. And it analyzes all the articles and said, these aren’t going to be of interest to your target audience. These would be. And so then it takes that, and it goes out to Perplexity and says, okay, are these legitimate things? And is there other news that can support this concept? And then it pulls all that news in, and it assembles it. And then it turns that into a compressed newsletter, if you will, that we would then publish into our mastermind group and say, hey, here’s some new stuff that’s going on so I don’t have to go sift through it. It parses automatically for me.

Rich: Now obviously, it took time to set up those rules to get all those things done. But once it’s in place, it just works automatically and then you can zhuzh it as you need over time as your needs change and things like that.

John: Yeah, you have to constantly educate the AI so that it knows when it’s getting it right and knows when it’s getting it wrong. So you’re constantly fine tuning your own little GPTs to get it right. But yeah, that’s the idea.

Rich: Very cool implementation for sure. You work with a lot of businesses. What are some of the common mistakes you see them make when they’re trying to implement AI and automation into their marketing process? And maybe if you can, what are some of the ways that they might avoid those common pitfalls?

John: Honestly, I think the biggest mistake they make is not structuring their prompts correctly. And they’ll go listen to somebody and then they just get into these giant, what I call, word vomit prompts. And they don’t have a process to reuse them. They don’t have a process to streamline them.

So every time they reuse them, they have to read through this whole word vomit again, and they have to change stuff, and then they have to go through the exercise again, or they’ve lost it completely. Because now in their chat history, it’s three months down and they’ve forgotten what it would look like, where did I put that prompt? I know I did it once before, so they don’t find it. So we use, I don’t know if you are you familiar with Notion?

Rich: I don’t think so.

John: Okay. So we build our prompt libraries on Notion. It’s just, it’s so easy. And you can create your own little fields and categories and stuff. Every time we create a prompt we’re putting it, not every time, but we create a prompt that we like, we’re going to put this in here. And sometimes it’s just a request. It’s oh, I need a prompt that’ll do x. And inside of Notion, you can create a stage. What stage is this prompt? Is it under development? Is it a request? Is it approved? Have we distributed it? And you can create a workflow behind a prompt.

But inside of Notion also, we have all the follow up prompts. We have fields for up to seven follow up prompts. So every time you run something, you might want it to do something else and refine it further.

Rich: Can you give me an example of what that might look like?

John: Yeah. I give you a really hairy example. So I created this thing called the Profit Accelerator, and it was originally part of this software called The Profit Accelerator. And they teach you how to interview your prospect and go through what we would call nine levers of marketing. And it could be upselling, cross selling, USP development, blah, blah, blah. And you’re asking them all these questions about their use of these marketing techniques, and if they are doing them right, wrong, or otherwise. If they’re doing them right, great. If they’re not doing them at all, then you ask them, okay, what kind of increase in your revenue do you think would be if we got this right?

So they go through all these questions. Now, what I wanted to do is take it a step further and just instead of saying look, you said on this one, it would increase your revenue 5%, this one, 2%, blah, blah, blah. I wanted to turn around and also say, let me give you some examples if you’ve never done cross selling or if you’ve never done bundling. I want to give you some examples of what that looks like when I give you the report.

So what I used to do is say, look, this report is a blueprint for how you can improve your marketing. The only thing is, it’s $5,000. But here’s the thing, you don’t have to pay me until after I’ve walked you through the entire thing. If you don’t like it, if you don’t think it’s a blueprint, you don’t have to pay me. But if you do, then you can pay me and we’ll see where we go from there.

The prompt sequence, it’s seven different steps. And some of those steps have follow-ups. But what I needed to do is I needed to go out to their website, analyze their website according to these principles, whether it’s products or whatever. So if we’re going to do cross selling or bundling, I needed to know what their product set was, but I also needed to know who their audience was. So I had AI do all of that analytics. I had AI come back and say, oh, we could bundle this product with this product. But I also said, look, if you’re going to tell me I should bundle this, tell me why and what’s the value to the customer.

So each one of those is a follow up prompt or another step in the sequence. And when I’m finished, it creates a 60-page report. But one of the other follow ups, of course, is to turn it into a Word document. Now I can deliver the Word document. That would typically take me a week to do all of that stuff, but I can do it in three hours with AI. And it’s amazing what it’ll prepare. It’s pretty slick.

Rich: I’m sure there are a lot of people who are like, oh my God, I have to start doing this, I’m missing out on this. If you had to give somebody a piece of advice of where they should start, like if they’ve maybe played around with a ChatGPT, they’ve done some generative AI but they haven’t really created any sort of systems. Which I think is what you’re talking about today is creating these scalable, repeatable systems. Where would you recommend they start their journey?

John: First I would buy the pro version of ChatGPT and Claude. And I might even buy Perplexity. Those are my three go to’s for generative AI. I would put off getting into graphic design and all that other stuff later. Video editing, there’s a whole lot of tools that you can get out there with AI. They’re amazing. But if you’re just getting into generative AI and you’re trying to get it to analyze documents, analyze images, or create copy, those would be the key ones.

And then I would say to keep it cheap. I would probably do a little YouTubing and just try to get your head wrapped around what it is you’re trying to accomplish. You might even, I hate to recommend some of these little courses on Udemy, because some of them you literally can ask ChatGPT. That’s how they created them. They just asked ChatGPT to create a course for them, and they’re not that great. But the more you experiment with it, the more you’re like, oh, this is really amazing what it’ll do.

And that’s why I try to teach people how to think AI first in just about everything. Because instead of over the last 20 years, we’ve all become accustomed to Googling stuff, right? So Googling becomes second nature. But I’ve done stuff that’s pretty cool with ChatGPT, if you get the app. Have you got the app on your phone?

Rich: Absolutely.

John: Yeah. So you’d take a picture and tell it to do stuff with the picture. I took a picture, I have a shop vac with a wheel missing and I just took a picture of it. And I’d said, what’s this? And it says, “It looks like it’s a shop vac with a wheel missing. And if you want, I can tell you where to get a spare wheel.” And I’m like, “Yes, please do.” So it says, “Well, turn it this way and let me see the model number.” I was like, wow. And I told this example to our mastermind group this morning. But I took a picture, I have a swimming pool and I had to take a picture of the chlorinator. And I said, “Look, I’m having problems with this. Can you tell me where I can get the manual for it?” And it goes, “Sure. Here you go. Here’s the link.” I just took a picture of the dumb chlorinator I’d long since lost the manual.

The other thing I did is, if you’ve got a swimming pool, you have these little sticks that have all these little cubes on them and you dip it in the water and each one turns a different color and it tells you what your chemistry is, what your pH free chlorine, blah, blah, blah. You typically take some samples of water and you go down to the pool store and they would do an analysis and tell you what’s wrong with your pool. And then they’d say, here, you need to buy 50 of this, 75 of that, 100 of this. And then you walk out $300 lighter. This happened to me because my chlorinator broke, and my pool turned green.

And so I was going back and forth and $600 worth of chemicals later, my pool was still green. So I finally just said, okay, I’m going to have AI analyze my water. And so there’s an app that you can get from Clorox where you can take a picture of your stick and it gives you what the readouts are on the stick. And then it tells you, of course, what you can buy from them, but I didn’t want to do that. So I just took a screenshot of what it told me my little chemistry levels were. And I created a GPT for this, and I just literally just throw the screenshot up into the GPT, no prompts, no nothing, because it knows the size of my pool, it knows that it’s a concrete or pool, it knows that I’m in South Louisiana. So it knows all this other fun stuff. And it instantly comes back and it says, “Okay, this is fine, this is fine, but this is really low. And this is really low. Here are the steps in the sequence. You need to do them.” And it’s sequence was right. The pool store I was going to, the sequences are wrong, which is why I kept screwing up and nothing changed. And the minute I got it in the right sequence, I was able to change everything. And so I built a GPT to do that. All I got to do now is take a picture and roll with it.

Rich: This is awesome. And I didn’t even get to a fraction of the questions that I was planning on asking you today, so I’ll have you come back at some point. But, no, it’s great.

For people who want to learn more about you, learn more about Bizzuka, and the AI Strategy Canvas, and everything else we talked about, where can we send them?

John: Oh, thanks for asking, Rich. I tell you what I’d like to do, if it’s cool with you. I’d like to do something that’s unique for your listeners, because I know you, you do a great job delivering value and I’d like to do the same.

And like I mentioned, I’m writing this book. The book is called Ingrain, I-N-G-R-A-I-N, and I’m hoping, Rich, the doggone thing will be published by the end of September. I have to rewrite three chapters. I’ve got a lot of golf left on that thing, but it’s dangerously close to being finished. But what I’d like to do is I’d like to give probably 50 copies away for free to your listeners.

Rich: Wow.

John: And I think we’re going to probably sell it for $35 to $50 when it comes out. But I’d love for anybody who wants to go to our website and if they just go to bizzuka.com/ingrain, I N G R A I N, you’ll get a little form there and there’s a comments field in there. And if they will just simply write, “Rich sent me” in there, I’m going to put them in line for a free book as soon as it gets up. And what we’ll do is we’ll reply back and get all the information that we need. And as soon as the thing is hot off the press, I’ll send them a copy, and it’ll be a physical copy of the book.

Rich: So awesome. Awesome. We’ll make sure those go into the show notes so people can check it out. And John, this has been great and very informative. Thank you so much for your time today.

John: Thank you, Rich. It was a pleasure. Looking forward to seeing you in the future.

Show Notes:

As co-founder of Bizzuka, John Munsell understands both the importance of AI innovation, and the frustration some feel with how fast it’s evolving. That’s why John and his team excel at helping business leaders understand, optimize, and leverage AI in their businesses with efficient results. Be sure to follow John on LinkedIn. And make sure you grab your FREE copy of his upcoming book, Ingrain.

Rich Brooks is the President of flyte new media, a web design & digital marketing agency in Portland, Maine, and founder of the Agents of Change. He’s passionate about helping small businesses grow online and has put his 25+ years of experience into the book, The Lead Machine: The Small Business Guide to Digital Marketing.