Ep. 277: The AI Operator & Data Ladders - Where You Are and What's Next

Speaker A: M welcome back to out of the Hourglass, the podcast for small business owners and leaders in the trades industry. Do you ever feel like AI should be doing more for your business than it actually is? You're not alone, and it's usually not your fault. Today, Chris Kiefer and Cody Hopkins from paintos, powered by Boolean, join us to break down exactly what separates business owners getting real value from AI, from those still stuck in the basics. We're talking data, tools and a simple framework to figure out exactly where you stand and what to do next. To complete the scorecard that we referenced in this episode, visit PaintOS app AI. Chris and Cody, welcome to out of the Hourglass. Thank you both for being here. How are you guys doing today?

Speaker B: Great. Yeah, Mick, I'm always, uh. I, like, I. We've told you before, but I have my own podcast, so it's really nice to be, ah, getting asked questions once in a while.

Speaker A: Be on the other side. See, I, like, I don't ever want to really be on the other side. I like maintaining control as the host, so I'm happy. I'm happy in the seat, but I'm glad for you that you're happy to be in. In the guest role for. For this moment. Um, Chris, you've obviously been on this podcast before. I think this is your. Maybe third. Yeah, I think third. Um, so it's great to have you back. And Cody, obviously new to out of the Hourglass. Um, you just told me you just had a baby this past week, so it's amazing that you're here. You're surviving literally.

Speaker C: This weekend, it was Saturday. So what is today now? Like? It's, uh, it's been. It's been a crazy couple of days, but it's. It's fun. Uh, it's. I was mentioning earlier, it's our. Our second kid, so that gives you a little bit of, I guess, calm for the storm that is parenting. But, uh, yeah, if it was your

Speaker A: first, I don't know if you'd actually be here for this recording.

Speaker C: I definitely would be.

Speaker A: I think your wife would be, like,

Speaker C: somewhere in the dark, some position.

Speaker A: Yeah.

Speaker C: Uh, oh, also, you're gonna do great, Molly. I know that that's coming up for you too. I'm not.

Speaker A: It is. It is coming up. It is coming up for me. M. Yeah, it's, you know, it's a mindset. We got this. We got this.

Speaker C: Exactly what it is. Good.

Speaker A: Yeah. Um, well, before we. We jump into all of this, um, I'd love for you both just to kind of introduce yourself, um, Chris, you kind of reintroduce yourself to our listeners. And Cody obviously being new, um, Chris, let's start with you. Just give us a quick reintroduction of you know, who you are, what you do and where you're from.

Speaker B: Yeah, so um, I'm a dad, we got, I got five kids, live in Idaho. Um, been in the painting um, world for about eight years now. Um, I have an engineering background and uh, I landed at Webfoot, who is a long time Nolan, um, member and I was at Webfoot for three years as their marketing director. Learned a ton from Gavin and Travis and the team there. And um, yeah the now about three and a half years ago I saw this like before AI was a thing, I saw this like um, the possibilities of automation and integrating tech into um, businesses as something that was more in alignment with my deeper passions than just like being a marketing director solely. And so yeah, in the last three and a half years we worked um, with a number of Nolan clients and um, we now have 26 painting businesses around the country that we are essentially like a fractional CTO for. So, so nice. That's what Boolean does. And that my obsession is like technology and um, integrating it in a holistic way to achieve the, the goals that we have in life and you know, balancing that with the feeling that many people have of like wanting to get off their phone and delete social media. You know, it's an interesting world that we're in. So I'm, I am in the middle of that balance. Mm mhm.

Speaker A: It's a battle. It's a battle for sure. When you get that, that Sunday screen report update of how long you were on your phone, your screen time every day, it's depressing sometimes. Like please don't even tell me.

Speaker C: And I'm gonna elaborate on what Chris said too. For him, like he, all those things he just said about like data integration and automation. He is obsessed with this specifically for painting. And it is, it's funny to uh. Like everyone's first question when they hear about Chris's business is like, well why don't you do it for this or that organization? He's like, no, painting is my jam. And it's just, it's, you know, when someone is obsessed about something because when you talk to them about literally anything like the NBA finals or you know, what he had for breakfast and the conversation always leads to painting, automation somehow it's like, that's hilarious what I was calling you about. But I'm glad to hear your latest thoughts on the subject.

Speaker B: And that's why the, yeah uh, the, the out of the hourglass is like the, the peak of my. The quintessential moments in life because it's like the few percentage of people in the world that care about painting so deeply ah is small and then painting automations it just gets, it's really, it's a really narrow slice of the population.

Speaker A: So it is. You've got listeners here who are very interested in what, in what you have to say and obviously we've listeners who are. And we work with folks who are outside of the painting industry. We work with uh, you know, a variety of trades and a lot of this will apply, apply to that. We won't. We're not going to be painting specific here but it's, it's important you know worth noting Boolean does work solely on. Within the painting industry. Um, um. Cody, introduce yourself. Tell us who you are, what you do and where you're from.

Speaker C: Right. So yeah, my story is interesting. I uh, met Chris like almost 20 years ago now which is shocking and I guess uh, doesn't even look like the two.

Speaker A: You could be friends for 20 years. That's not possible.

Speaker B: That's what I told Cody. I was like we can't say 20. That's like dude, that's like an old person.

Speaker C: Gary the nine. Yeah but I um, I spent the first 10 years of my professional career as a high school teacher. Um, yeah, I won't tell you what subject. I want you to guess at the end.

Speaker A: I, I think I know Chris already told me so that's. I have, I mean I have a leverage up here but I'm gonna forget that I was maybe told and we'll guess at the end.

Speaker C: That's fine. Yeah. So I uh, yeah I taught in high school and I coached with sports and I had my CDL and I drove a school bus and I just did all the things was totally locked into that environment. And then uh, coincidentally enough Chris uh, was in Seattle.

Speaker B: Yeah, it was at the Grand Summit actually. So Cody was like this is past year. Yeah, yeah. I was like hey, I'm in Seattle for this conference. And he was like hey, you want to work out? I was like heck yeah. Where are we going? And so we went and did a workout, went for a run and then I was like hey, you should come grab uh, I'll sneak you into breakfast and get you some bacon and eggs. After we got done running.

Speaker A: That's why my bill was higher than I expected it to Be that extra headcount, Cody, you owe. You owe me.

Speaker B: And then he sat down with Josh Abramson and I and we just had, you know, uh, had a quick bite to eat before we ran up and showered and everything and got ready for the conference. But that was his exposure. He was like. Because he, you know, we knew what I did, but he was kind of like, you know, hey, tell me more about that. And then he was asking some follow up questions and then one thing led to another and it was like, here's a test project, tell me what you think of this. And then, um, yeah, it just, it. Cody is like gone on a exponential growth curve in particular when it comes to AI, which is why I thought he would be a great person to come in and talk to this, because literally I. It is the no exaggeration to say that eight months ago, I mean, what was your. How much did you use AI? Like everyone used AI a little bit.

Speaker C: Yeah.

Speaker B: But what was your. Where were you at eight months ago versus where you're at now?

Speaker C: I had a working relationship with chat GBT on my cell phone.

Speaker A: Okay.

Speaker C: Like we understood, uh, each other. Uh, it made me chuckle, answered my questions, really validated my feelings and that was it. That was the entire extent of my usage of AI.

Speaker B: Yeah. And so he got put in just as like head of operations, you know, managed some people and he had lots of experience with that. And then it was just like a matter of time before basically like Open Claw back in January. I like, I feel like that's when Cody, that's like he became this like Frankenstein of a pure of a person. And just like he's been showing me stuff daily non stop for the last six months. He's built out like a full operating system for us to manage Slack conversations and give context to these AI systems that we're running for clients. But it's uh, allowing us to resolve problems better, faster and cheaper with less staff because we have like everything that we can give access to. We are giving access to it. And then we're also starting to filter what tools we are using based on how well the AI can interact with those tools.

Speaker C: Yeah, and I'll, I'll just add slightly to that and say that when I like first heard about Chris and what he does, it was interesting. It was an opportunity for me to like, want to pursue and kind of migrate, you know, just a change of career, which is like really big and like a scary thing. Um, but, you know, it just kind of came down to my relationship with Chris and like the history that we have and also like the excitement of the thing it is that he was doing. And when Chris is explaining all these like, AI things that I've built out since that point, like those words don't make sense to 90% of people that I talk to about this, which is everyone that I can talk to about this, namely my wife who just blazes over and is like, please, yeah, you, like you're lost.

Speaker A: You lost me at the, at the word AI.

Speaker C: Exactly. And so that's like. And it's such an understandable place to be. And again, most people that I talk to are around that spot and. But Chris didn't send like me on a mission of go figure out how to use AI and become like a 10x AI power user. He was just like, Cody, here are problems. And so Cody then went to try and solve problems and then used AI to help him solve that problem. And then in doing so, in the going in and building a thing and breaking the thing and the thing didn't work, built the next thing to build the next thing to build the next thing, all for the purpose of solving the problem. That was whatever we were experiencing in our business. And this is something I do as a hobby. I just like ask people, what are your problems at work? And let me tell you how I would use AI to help you solve that problem. And you have to identify where a person is, like, uh, in their comfortability. With AI, which is the first mistake that I made was, oh yeah, let me tell you exactly how I would do, like literally with my setup.

Speaker B: Yeah, just create a knowledge base, connect it to the MCP for your CRM. And then, you know, it's like super simple. Yeah, yeah.

Speaker C: So then it's that you take a step back and you ask a person like, okay, what do you like, what's the pain right now with where you like are at work? And have you thought about using AI? What's the last thing you did with AI? And you just take that and you're like, okay, what's like the next level up thing to do? And also what are the things you should try and avoid to do? Like, I'm going to like tell you some of the things that I went through to save you time. Open claw. Namely, like Chris mentioned, if anyone out there is like, in the AI world, they know what that is. Um, it was not a waste of time, but I don't do open claw stuff anymore, um, if that makes sense. And so, yeah, I guess, uh, now we just solve problems all day long. Using AI and are trying to, like, empower other people that, like, we work for in our partnerships to be utilizing these tools and, like, strategically setting themselves up so that they are able to use the tools, which is what we mentioned kind of at the beginning. There's sort of like, two ladders, which we'll talk about later. But, yeah, that's, uh. That's me, the ex high school teacher turned AI problem solver.

Speaker A: I love it. I. And it's so cool. I mean, just to have that. The. The history that you guys have had. But now, I mean, you're only, what, probably then six, seven months into working together, you know, full time on. On all of this. But clearly, it sounds like the Runway from those six months, so much has progressed. And how are you, Cody? How are you? We were talking earlier about, like, AI moves so fast, so how are you? And you, too, Chris. How are you guys staying on top of it all? Or, like, is it even possible to stay on top of it all?

Speaker C: I would love to answer this question. Um, Chris asked me the same question, too, and it's honestly kind of hard to explain. Like, Cody, how did you do this? And I'm like, honestly, I'm. I'm not really sure. But the. If I were to boil it down, it is maximum effort and full immersion.

Speaker A: Um.

Speaker C: Um. Like, the Runway is short when you think about only six, eight, you know, months, about a year, whatever. Um, but that Runway, like, expands when you are doing it for 16 hours a day, because it's just, like, so obsessive, number one. Um, and so, like, for. For me in particular, it is, like, a passion. Like, I can't sleep because it's, like, all these things are working, and I'm just trying to, like, keep the projects moving along in a way that's, like, very exciting for me. And I know not everyone is drawn to that. Um, but in that maximum effort and full immersion, you are doing so much and learning so much through doing. If anyone puts that amount of time to anything that they want to get good at, if it is building a fence or cooking, you know, your homemade pasta or baking sourdough bread, like, the same principles apply if you do it all the time and you. You love it, you know, and you just keep at it. You don't quit. You don't get discouraged, because you're gonna run into struggles and failures and whatever. And, like, the failures evolve so much. Like, in your sourdough journey, you might. The first step might just be keep the starter alive, but they feel this.

Speaker A: I feel this so hard.

Speaker C: Yeah. Like two years later down this road, I just can't get my grams of, you know, whole wheat and whatever. Correct. And my scoring is just off by. You know what I mean? Like, your problems then, uh, like, don't relate to your problems at the beginning at all. But you don't even know what those problems are until you go live it and experience it and try to build it out first. And so that's part one. And then part two is you just have to stay up to date, um, with the news and what's coming out. Like, I blog about this a little bit and I, like, just write down my thoughts about how things are working and going. And I've not even been doing that very long, and I already have like, four articles that I don't even believe are true anymore.

Speaker B: Wow.

Speaker C: And it's just like, how fast things are changing. But I think, like, uh, again, it's not a wasted effort to like, like, take the information and synthesize your thoughts because it shows you the evolution of, you know, where this has gone and where it can be going.

Speaker B: Yeah. So my favorite one thing I would add to that is, like, you're. So Cody's talking about how he got to where he is now. And I. I'm thinking with what you said, Molly, is how do you keep up? So it's like, Cody, Cody, like, got from behind to, like, in the lead very quickly. But now the question of, like, okay, how does someone just, like, stay up to date with what should I be trying and how should I be using this? The two things I would say is, um, like, you have to. You have to have time in the week to just try to improve the system. And I would say it's very similar to. As you guys coach business owners, like, it's hard to work on the business when you're in it. So you have to make sure that you have time in the day to take a step back and be like, okay, on Monday mornings from 8 to 10, I am trying to improve, like, build an AI skill or something. I'm going to try and solve a problem in the business using AI and carve out time for it. Um, that's one thing is, like, you. It's hard. You can't stay up to date if you're not, like, putting in work to try and just use it. Because this is the first technology that's ever existed in humanity where it can tell you how to use it if you ask the right questions. So we've never, ever had something like that before. And I was just listening to. Well, the second thing I was going to say is if you're really hungry to figure out how to adopt this, it would be find a podcast that talks about. And you don't have to listen to it daily, but there's a podcast that Cody and I listen to daily that's called the AI Daily Brief. And it's like, what's going on in the world of AI? And it's crazy. Like, every day there's a lot of stuff going on. It's like, it's crazy, but it's just like, brief. It's like 20 minutes of content that you can listen to at, uh, you know, one and a half or 2x and just like, just a quick, like, hey, there's this, like, right now there's Fable. Fable 5. If you've heard anything in the news, Anthropic got shut down by the government because this is like a national security threat. And now you. We can't use Fable 5. Like, just came out a week ago, and now they're like, at the meeting at the White House. They're, they're executives talking to the White House personnel, and there's all this drama, right?

Speaker A: Yeah.

Speaker B: And for most business owners, maybe that's like, too deep in the weeds. But I would say again, at the. At a bottom line, you should have some periodic source of like, hey, there's these new things that would be useful to you. Cody and I are at a different level of like, trying to be on the cutting edge and experimenting with, like, the model that came out like this. Fable 5 came out a week ago, and it was out, uh, for two days before the government pulled it. And now it'll probably be back again, theoretically. But it's just like, it's. That's what you have to do is carve. Carve out time to use it, and then also make sure you have some source to, like, lightly guide or filter curate. What you should be trying or experimenting with is my opinion.

Speaker A: Well, this, this brings me really nicely kind of into my, My, you know, next phase here of questions for you both. Um, and, you know, Chris, you're talking. You're both talking to business owners on a regular basis, and you're, you're there to solve problems. But what do you guys typically hear? I mean, you guys are obviously immersed in this every day. Um, I use AI tools. I was telling you, Chris, you know, my. When we were kind of prepping for this, I met, like, I can create a skill that's kind of where my, My AI skills kind of tap out right now. I haven't gone beyond that. I know that there's so much more. I've said it on webinars and all these things, and I'm sure I represent a vast majority of our listeners here. And people who probably haven't even gotten to the skill level are still using it as like a Google search or a chat just to, like, answer a question. Um, where do you see that expectation of folks you're talking to and kind of where they're at and where are they stuck? What are they feeling overwhelmed by? Like, what's the general sense from just like, a typical person who's trying to figure this out?

Speaker B: Yeah, I would. I have thought a lot about this, and I think that in general, business owners are. They're caught in this, um, chicken or the egg. There's this, um, uh, it's like where I. I want to invest in where I'm going to get an roi, but I want to know where I can get an ROI before I invest. Right.

Speaker A: Yeah.

Speaker B: So you, like, which one do I do first? And there's like a, uh, an analogy or, like, story I've heard is like, if you imagine it's. It's nighttime and there's street lights, and the street lights are illuminating certain parts of the street, so we just naturally, like, go to where the light is to see if we can find the solution in that beam of light, rather than just going to where the it would actually be, which could be in anywhere along the street. You know what I mean? So it's like, we all. We're always drawn to. And it's. The business owners are smart. They're. They have limited resources and they're trying to be smart with their, you know, their betting and hedging and whatnot. But the reality with AI is like, it's really overwhelming. And there's no, like, there's. There are very few. There needs to be a lot more places that a business owner or employees, um, can go to to learn how to use AI better. And because it's evolving so quickly, most of the courses that come out about using AI, they're out of date the week after they get published because it took the guy, you know, a month to put the course together. And so it's like the best. The best use case or best thing to, um. Like when you. When people feel overwhelmed or they're feeling like. Everyone's been talking about AI for like, four years now. And yes, there's been some stuff that's gotten better, but the only thing like most business owners are feeling like they've gotten incremental gains in like I use it to draft my emails or I mean what like it's like con copywriting type stuff or things like that. And I would say it's because that's where you can see an ROI. Like I could save myself 5 minutes by writing this email 100%. But what you're not thinking about, I think what business owners aren't thinking about and why they're stuck is because there needs to be some level of like play or experimentation, whatever word you want to use to just try like step way back and be like, wait, why, why do we send emails to begin with? You know, like if you ask AI, I'm a painting business and I, you know, if you draft this email to my prospect and it can do that for you. Right. And if you give it more context, it writes a better email. And then if it has the most like access to everything, it'd be the best email. But it's also, you could take a step back and be like, what's the like? And you could even ask AI, why do I send emails to my customers in business? And what would be a way that I could insert you into that problem to solve that problem better for me or what? Uh, you know what I mean? Like yeah, if you ask AI a question it gives you an answer and I can't. I, I feel like I still, I thought this was going to go away, but I still have. People are like, oh yeah, I tried to chat, I tried chat GPT and I told it to write me a bio for, you know, um, this business owner and it just made up a bunch of stuff so it's garbage. And then it's like right off chat GPT because of this really specific use case. And when I hear something like that I'm like, well yeah, new it has no idea who that person is. There's nothing online that could tell you who that person is and you told it to do something. So it's going to give you something that you asked it for, you know, but then if, again if it's like depending upon the problem you're trying to solve, if you ask a deeper question and ah, and maybe even say I don't even know what question to ask you, can you help? Start out by ask telling me what I should be asking you. Like it's just, it's like almost like the matrix where it's like you got to go deeper into it.

Speaker A: Yeah, it feels like too simple. To even start. But that is a great, that is a great place to start. Um, but I feel like, you know, you, you're right, Chris, in the sense of we don't know, we're not thinking, we're not kind of getting like pulling back enough to see, you know, why we're doing the things that we're doing. But it is, it's also that immersion time of the being able to work, to work on, on the learning. We're just not, we just don't always have the time to do that.

Speaker C: Yeah, I think coming from education, all education costs money, cost resources. Like if you went and got a degree, that costs money. If you learn something on your own and spent a weekend on a project, you probably, you know, ruin some materials or your time is worth money. And so there's, there's a cost to all of it. And as a bit smart business owner, it's like Chris was saying, it's like how a great business owner just invests their time and money in the right places to get the roi. And so I, I'm thinking of everything that Chris just said and what you just said. And to me it boils down to, if you want, if you're curious about the AI thing and you want to get better at it, use it as a coach. Just try and solve problems and then ask it questions like, how could I have solved this problem better? Or like what is possible? Do some Internet, like it can go on the Internet for you. Do some contemporary Internet research and tell me what people are doing right now to solve this problem and what are some things I'm not thinking of. And give me a couple reasons why this idea is bad. So using it as a one on one coach is always really helpful to get you just up to the next level. Up to the next level. And then I also just think like anything, um, you need to identify where you are as an AI user and not go to the North Star of everything that is possible that you've heard your friends talk about who work at IBM. You know, this is what AI can do. Like don't try and skip all those steps to go do that thing. Yeah, just try and find what is the next high leverage thing that I can do and, and stick with that and try that first. And then once you get comfortable in this arena, you're gonna ask your through the same process. Here's where I'm at. I'm using AI as a coach. How do I get to the next stage in the arena and the next one and the next One so simple.

Speaker B: And that's beautiful. Well said. That's literally the path that Cody took. Because it wasn't like. I didn't say, cody, I'm going to hire you to become our AI expert. I was like, hey, Cody, I have. I could use some help with operations. And it was, that was the high, uh, like overarching problem he was solving. But then he had space and some resources and some AI subscriptions. And then he started playing and then it was like the first couple things he made, I was like, I mean, cool. Does it work? No. Okay, well, how it's. Does it. It works sometimes, you know. Okay, how do we make it work more often? Okay, how do we. So it's just like that process.

Speaker A: The.

Speaker B: On top of that, uh, the other thing I would add is that there's this even. And you're probably in this boat right now, Molly, you can tell me if you agree with this or not. For someone that's using skills, there is just this hu. It's like change is hard. It sucks. It's just like I just want to get good at something and then like feel competent and capable and like, I know how to do my job. And I feel like what, like what is required is to basically. Okay, yeah. Give yourself a pat on the back. Congratulations, Molly. You use skills.

Speaker A: Yep.

Speaker B: Like tomorrow you need to download cloud code in your cli because that would be like the next step for you. And, and have it give you step by step instruction on how to do it. And like just try for 30 minutes to solve, like send an email through Claude in like that, for example. And I'm, I'm again, whether or not this is right for you, I don't know. But the point being that's like the resistance that you or anyone else is going to feel is that fogginess of just like, I don't even know how to speak Spanish. I don't even know where to begin. And you're asking me to just like go to Mexico and just like start interacting with people at a marketplace, you know? And it's like that's basically what you have to do. Like at uh, just over and over again. And then you get excited because you have this breakthrough. But it's like just know that you can feel good about that, but there's another layer on top of that. And so you have to be like, humble enough to be like, okay, I'm going in. I have no idea how to do this. I feel completely unequipped. You might be nervous or scared. Like, what if I Break something. You can tell. Say all that to chat GPT. Like, I'm really worried that if I get in here I'm gonna break, break something. How do I make sure I don't do that? Or how do I. Can you give me a sandbox? Or how can I play without. You know, like, if you, the more that you can articulate why am I not taking action, it's going to be able to guide you to the next step.

Speaker A: That's, it's a good thing to hear because I would say to you, Chris, what's a cl? Cli. Like you, you would say that to me, download Claude code for into your CLI and be like, well, it stops there. I don't even know what that is, so I'm not even gonna try. But like, that's, but the reality, those are questions that we need to say, no, it's okay that we don't know that. Let's, let's ask, let's ask, let's, let's ask the question what the CLI is and how do I even interpret it? Um, but we hear these, we hear these terms, these technical terms or, you know, mcp. I've seen an mcp. What does it do? But that's, we feel overwhelmed and that's where we stop. And that's where I think a lot of us are inhibited in our growth and being able to take that next level. Um, and Cody, Chris tells me you've built out a framework for how to kind of diagnose where, where individually we are in our kind of AI skills. And yeah, I think you should share your screen.

Speaker B: Cody, I know this is a podcast, but this would be for those of you that can go to look, uh, at on YouTube or whatever. I. Cody, this is like brilliant work and it really puts some context into everything that we're talking about.

Speaker C: Yeah. And this, a lot of thought has gone into this because I, I was stuck trying even to articulate what it is to be, you know, a good AI user.

Speaker A: Yep.

Speaker C: Um, because what people want is, ah, a linear scale of progress that just moves in one direction, one direction on two axes and that's it. Um, like anyone who is training for a race or a marathon, they're like, okay, it is February, I am running a marathon in September. Um, if I go out and run, this is my time right now at this distance, by this time I want, you know, this time or distance and whatever. And so then you can just train and get better and faster and see that progress. And it's very simple. Um, and I tried to do the same thing with AI. Like, okay, I am here with AI and I want to be better at AI. And so what are the things to do to just like incrementally climb up this ladder on two axes? And the finding, for me, and this is my opinion, someone might have a different opinion, is that it's not that simple because there are so many factors that go into your ability to use AI well. Um, and so I kind of have it. I have this whole formula. Um, but the two main components that I feel like people have the most control over, and they're like individual work. Are you as the operator, how are you going in to an AI anything? And I want this to be agnostic as well. Not Claude, not ChatGPT, not Gemini, not Claude code. Just in general, if you want to use AI, how do you go about your business of solving your problems? Um, and then that's component one, you as the operator number two. And this is, I think if you're a business owner or someone who is like, planning for success on how to level up yourself or your organization with AI, it's the second thing that's even more important than the operators. It is the data integrity and the data structure. And so you could really level up as an operator but have poor data and then not be as effective with AI as you could be. Um, on the other hand, you could have really, really excellent data and a very basic understanding of how AI works and get really similar results. And so I have two ladders, operator ladder and data ladder, that I'm going to go through. And Molly, I would love to use you as an example. Like, please, let's, let's use these things. And then stop me at, ah, any point where it doesn't make sense or if you need further clarification on like, what a term means. And then like, I want you to rank yourself. Where do you think you exist in each of these different categories for each of these ladders? Okay, the first one, again, for people who can't see, this is your operator ladder. And level one of an operator, in my opinion, is just someone who talks to their LLM or their AI. They have a problem, they want a solution to solve this problem. And so they go in and they like, send a couple messages back and forth. It'll ask some context. Um, Chris's wife, Natalie is training for a triathlon.

Speaker A: Ah.

Speaker C: And so she is, um, really big on what am I eating, how am I feeding myself? Like, how do I, you know, make sure I have enough fuel to do

Speaker A: this correctly, my husband just completed a half ironman this past weekend.

Speaker C: Really?

Speaker B: And did he use Claude for any?

Speaker A: Sure. Did he? Yeah, we. It was like the couple days before we were gonna make something for dinner, and he was like, no, I can't eat that tonight. I have to eat that tomorrow based upon what Claude is telling me.

Speaker C: So you get that?

Speaker A: Okay, I get it completely. M. I've seen it used for. In multiple cases for training programs.

Speaker C: Yeah. So let me add a couple layers of complexity to Natalie, Chris's wife's position. So, yes, she is training for a half triathlon, a half ironman. I apologize. And has all these things. What should I eat? And when. That's factor one. Factor two is Chris is completely celiac, can't eat any flour. Um, one of his daughters is allergic to everything. Just eggs.

Speaker B: I mean, Yeah. A bunch of animals and other things. Yeah, there's so many primary thing.

Speaker C: Okay, so like, and then also on Tuesdays, they like to like, have people over and host. And so they need to, like, you know, buy more supplies to all this, you know, meal planning, prepping factors, uh, and just every complexity that you could have of running a huge household, like, with very specific diet issues, like, that is your problem. Okay, so now if you want AI to solve that problem for you, phase one is just go into the chatbot and ask it, what should I have for dinner tonight? Or what should my meal plan be this week? And then the response you're going to get from your AI is, uh, well, I have. I have no idea what. What do you like to eat? And you know, it just. It doesn't know. So then it's just, you could go as long and long and long as you want, like giving it as much information as you can inside that window, and you'll get a really good collaborator and someone who knows things about stuff.

Speaker B: And at the end you do get. It does get. I mean, if you keep dialoging, it will get better and better, you know, and we've all. That's like, anyone that's used, AI can relate to that.

Speaker A: Yes.

Speaker B: The next step is like, yeah, go take it a step further. And what are the, uh, I'm going to interrupt Cody, because I feel like it's important to think through the way I. My brain works. That will layer on top of this is that you can take a pause out of like how I use it as an operator and my brain immediately goes to data. So, uh, simultaneously is how you're dialoguing and what way you're using AI the way my brain works is let's just brainstorm all of the possible data sources that could be useful for solving this problem. And I do want to talk about

Speaker C: data later because that is the next. The next layer.

Speaker B: Right? But I'm. But that's a. Like, keep in mind that there's two things that we're solving. So one is like, I want to have this conversation, and then there's the other side of just, like, in the back of my head. I can't help but immediately start thinking through, you know, and I'll rattle them off in a moment. But all the different things that might impact meal planning or nutrition.

Speaker C: Right? And so you can either a, uh, type all those things into the agent, like, little chatbot window, or B, you can send your chatbot to, like, go and read something. Maybe you have a file saved that it's like a Google Doc or an Excel sheet that has tracked what you've eaten over the past. And you leave some notes in there about how well it went, how many mouths you have to feed. It could be some data source. It could be a file with multiple things inside of it. And this can be as, like, simple or robust as possible. But still, a level 2 person would just go into a chat window and say, go look at my data, which could be great or fine, and then give me an answer. And that's a little bit easier than having to type in or copy and paste in every piece of information that you think your agent or the AI thinker, um, could use or want. Okay, so that's level one. Just talk about it directly in the chat. Level two, go and find the information and give me an answer. Level three, I call the manager because this person has realized. And this was my big unlock, like when my life changed was when I realized all of the things that an AI agent can go and do for me chatbot, is give me the answer for how I should do this thing. An agent is just simply no matter where your agent lives, you know, there's a million different, like, softwares that, like, you know, we build agents and you can design your own agent. What is an agent? It is an AI bot that goes and executes for you. And so as a manager, you can say, okay, agents, go and do this thing for me, which is just build me the meal plan. Or, like, go online to Fred Meyer, where they deliver my groceries and, and build out my shopping cart and here's my credit card. Literally, you can do this. You can give an agent your credit card with spend limits and here's your budget, here is my like, needs, and go do the thing crazy. Um, and again, this is not the data part. This is just the operator. Like I'm going to send agents to go and do that thing for me. And so how much information has to go in the chat like window is determinant on like your data. Level four is a director. So it's instead of AI, go do this thing. It is, I have designed like specific experts at different things that are going to work together to accomplish the same goal. So Natalie, instead of saying AI, go do this thing for me, it is okay. I have my nutritionist agent who knows what I need for my half marathon training. I have my personal assistant agent who knows what my calendar is and you know, how much food I need to make for, you know, each day of the week. And then I have my chef who actually knows what I like to eat. Um, and so these are now different, like textures of AI that you can like roll your bot or agent through. Because I call like normal GPT or claw just like vanilla ice cream. And then imagine going to like Cold Stone Creamery and you're just like, wow, I can have all these different ice creams and I can put gummy bears on them. I never even considered putting gummy bears on my ice cream. But you can take your vanilla ice cream and roll it through whatever flavors you want to give it that specificity. And the same thing is true of agents. So whereas a level three is just using AI, general vanilla ice cream. Level four is there are different flavors and toppings that I can give AI, so it executes to like, the specific, um, like expertise that I am looking for in this situation. Um, and then the, the top level of operator is the executive, and that is you. Just one simple command. Like Molly, you created skills and your skill like, does things. But if you get to the point where your data is so good, you can just type a simple skill that launches multiple agents, not just in quantity, but in type. Like multiple nutritionists, multiple like, uh, chefs all doing different things, logistics. One of them is the shopping expert. So it knows like, what deals and coupons like that that is possible. That is level five. Um, and it is like a cascade of things that all you have to do is at the very end check the thing. Or this could be running autonomously. And so the groceries are literally showing up at your doorstep every Monday and Wednesday because your agents have the authority to like, go and do these things. And that's completely overwhelming for anyone who like, is still using the chatbot thing, which is why that should not be your goal. Yeah, um, but like, that sounds wonderful. It's amazing and it's so possible and it's the kind of stuff that we're doing for organizations. And so now, like, take it back.

Speaker B: We're not buying groceries, though. So to apply this to, uh, we could. We actually have had doordash delivered to clients during meetings or whatever. That's cool. Um, but the. Yeah, go ahead, Cody. Are you going to jump into or. Actually, I would be curious.

Speaker C: You are a person who just, like, gives all context via what you type into your chat window. Your next step is not build an orchestrator agent that, you know, controls all these other agents to these very specific things. Your next step is, what if you downloaded a voice, uh, to chat software so that you can get your thoughts down faster? Like, that would be really cool. Or what if you downloaded like, an AI note taker, like granola that you can just open up on your phone and then instead of having to type the thing, you can just upload your conversation that you had with your wife about the stuff. And so for our clients and for people in Nolan, it's not your nutritionist and your chef. It's like, um, your sales expert or your marketing expert or whatever that can do these types of tasks for you. So that's ladder one. Like, are you typing into a chatbot window on your phone or are you orchestrating multiple types of agents to do things autonomously?

Speaker B: Where do you feel like you are? Molly?

Speaker C: Love that.

Speaker A: Um, I'm probably. I'm definitely at the consolidator level. I mean, I definitely utilize files and sources and, you know, have it analyzed and do things from there. Um, I don't think. I think I want to be at the manager level. I'm not like, I'll have IT create things for me, but I'm not. Um, like, for example, I'm redoing our website and I'm doing. I've done all the content and I've provided all of the data I've pulled in. I feed it so much, and then I'm working through what the pages are going to look like with Claude. And so we're having, we're having a dialogue, but I'm providing a lot of the information and then having it create for me. So does that make me a manager?

Speaker C: I would 100%.

Speaker A: Or is that more consolidator level?

Speaker C: No, I would say 100%, yes. If you're having an, um, an. An AI agent do things on your Behalf.

Speaker A: Yes.

Speaker C: Even if it's as simple as make me a PowerPoint presentation.

Speaker A: Okay.

Speaker C: It went out. It did the thing. It's not how should I format. Here's a screenshot of my PowerPoint. Like, where do you think I should move things around? There's a lot of that.

Speaker B: Uh, I think another. You're bringing up an interesting point with how you phrase that. Molly is like there. It's like you have the capacity clearly and you have a, uh, you have like some receipts of being a manager, but we all fall back to just there. It's not like you should never use a chat window ever again. Right, right. It's just that, yeah, there's times for me where I'm like on the go where I use, like I don't use Google Search anymore. I'll just ask Claude for a quick answer to some concept or question. Right, right. And then I'm in that moment. I'm acting just as like a chatter. Right. But it also would be silly to like design some orchestrated system for a question about like, uh, yesterday, did you know that trees actually get stronger because of the wind and the uh, Just like the external elements. This is like a real thing that my wife was telling me I didn't believe and like that. Like we had like a one minute cloud conversation of me just chatting. I'm not going to go build something for that use case.

Speaker A: Yeah, right.

Speaker C: Okay.

Speaker A: Yeah, I think, I think I probably. I probably lean more on the consolidator side. I have moments where I get to manager, but I'm probably in between. In between those. Those two levels right now.

Speaker C: Yeah. And that's all great. And honestly, being the consolidator is in my opinion the most important thing that you can do. Like you need to really spend some time in the consolidator space to set yourself up for success for the levels that are above that, which is the segue into the next ladder. So this is ladder one. How do you use AI? Uh, do you talk to it or do you send it off to do things? You know, and at what level?

Speaker B: And Cody, let me interrupt to give the. I'll. I'll say just my. You. You organize my, my data thoughts. Okay.

Speaker C: Okay.

Speaker B: So back to this meal plan analogy.

Speaker C: All right.

Speaker B: The way my brain works is. Okay. If I'm going to solve this problem really well, I could have. I'm just going to list things and if you can think of any other ones I forget, tell me. But like there's fridge and pantry contents, things that are in there. There's health data potentially from like my whoop tracker. I could have a fitness goal that has a date and like a distance that I'm going to run or a type of triathlon I'm going to do or something. There could be allergies, there could be food preferences, There could be Google Calendar access to like people. Maybe we're going out to dinner or we're going to an event so we don't need to cook at home. So it's like just. I'm trying to illustrate there is anyone. For any problem I would highly recommend thinking through what are the possible data sources, whether or not they're organized or like I can even access them. Some data sources might just be in my head, right? So there's. There is a layer that Cody's going to go through here of like first just knowing sometimes people don't even there. And this is I would say a big place that people get stuck as well. They're thinking about it from that operator level. And the unlock in this mapping out of these two ladders, in my opinion is it's just making you aware that there's another thing that increases the effectiveness of how you're operating AI and that is data. And so it's like again, make a list of all them. Then the next step would be what can be systemized or digitized. And then after that it's digitized. Now what can I give it access to? So there's tiers of that and then Cody can go into a little more detail. But that using that meal plan analogy, just keep in the back of your head. There's all these ways I can chat or, or level up how I'm integrating. And then there's the making a list of the data, organizing it, structuring it and then keeping it fresh is all another thing.

Speaker C: So yeah, so as far as levels go, anyone who has data, wherever it exists in your life again could be on a piece of paper, a notepad in your head. Google sheet calendar Level 1 of data usage and data organization and structure is just copying and pasting. Like I went to the thing, highlighted it all, copied and went into my whatever. Because you could even be a level five who copy and paste. In theory, you're still copying and pasting, but telling whatever orchestration of your agents to go and do stuff with that material.

Speaker A: Same as like copy and pasting or uploading a file. What would you put those in the same category?

Speaker C: The same thing, right. So like you're just finding the source of material and like giving it Directly to, um, your AI or agents.

Speaker B: And again sometimes that source could be your brain, could be your brain. Brain dumping verbal, verbal like, uh, conversation. That's possible as well.

Speaker C: Yep. And then level two, after copied is routed. So instead of giving directly like manually all the information that your agent needs to go and do the thing, you could say all the information you need to do the thing is over there. So go look in there and then do the thing. Uh, so this is really like tied to the consolidator. Like if you don't have information that's consolidated, you can't get to this step. So routed is number two. And then the next level up is when you actually connect it. Like if you've heard of MCPs and APIs or if you just go to your, if you use Claude or ChatGPT, you can just authorize a hundred different apps. Um, your Notetaker app, your Google Drive, your Outlook account, and now your AI is connected to these things. And so you can go into your agent or your chatbot and say, hey, go read my calendar, go and read my um, Google Drive and find these PowerPoints and then do something with them for me or answer this question for me. Um, and this is really useful if you have. I just did this for my wife actually. She, um, has these quarterly check ins with her clients and for years and years as an expert, all the information in her head, she has decks and decks and decks and decks of ones of like how she likes to run these types of meetings. And so I can create with that an automation since I have access to all those things. I didn't go and save and download and upload into some file. I just said, agent, go and find every single PowerPoint presentation that has this title or like serve that purpose. And then my wife told me these are the like the builders who are actually good. And so now we have a list of like 25 PowerPoint presentations. And now we can build a template and a skill to be able to, you know, go and make something with a press of a button instead of the hours it spends to make these all from scratch. And so that is possible because we are connected to it. And then after connected, you would think that's, that's great, uh, it can see everything now. What else could I possibly need? And the answer in our experience was instead of having your agent go out and find all that information and have to comb through every line item of every object of every software and tool that you're using, you have all of your tools instead pipe into your Own database. And you can control this database and organize it any way that you want. And the way that you should be organizing your personal database is in a way that agents are able to read it, which is mostly MD files, markdown files, which is very, very simple, plain text. Um, a term I did not know about until six months ago. What's an md? Never heard of it, never would use it, because they're not built for people. But we don't need people to go through our files anymore. We need agents to go through our files to get the answer that we need. And this is really helpful because as you organize and design your agents, the reason I have a nutritionist and the reason I have a, like a, uh, chef is because those agents know to go look here and there versus, you know, here and there. And so that saves on their context. And as far as, like AI works, they have perfect memory. As long as you don't overload its context, it can remember everything perfectly within a million tokens. But as soon as like whatever you need it to do goes outside of that window, now it's going to start making mistakes. And so the more specific, like, data location that you can give your agent, the better. And the best way to do that is to structure a database that is aligned perfectly with your entire system. Um, so that's level four, centralized. And then the only thing that is better than a centralized data source is a living, fresh data source. Which means what are the headlines? What are the true stories about our data as of today, this morning? And is our database reflecting that, like, um, that truth? Because, like, a great historian, like, writes a great book that you love to read. It told a story, there are chapters, it's organized in a way that's great for a human. Um, a great historian does not go read 200 newspaper articles and then list them in his book chronologically from the first one to the last one. Nobody wants to read that. So you need something that is going into your 100 or 200 newspaper articles, whatever it is, from all of your sources, your CRM, your Google Drive, your, your finance app, uh, and then taking all that information, turning every page, reading it, and then organizing the data in a way that makes it really accessible to your agents. And it does this autonomously. So that is the goal, is it

Speaker A: at this level, when you hear people saying they've purchased a computer just to have something running all the time, like, is that, is that when they're, they're, they're at that level four, level five,

Speaker C: you actually don't Need a, uh, separate computer to do something like this. Um, um. And the complexity grows, the bigger your team is. You, um, could have something as simple as a database that lives on your computer just in your normal files, just like right next to your pictures of your kids. Um, that as long as you design your agent so that they know where to look for this stuff, they can go and find it.

Speaker B: And then, and then there's also things like simple tools. On a Mac, for example, there's programs called like Amphetamine that just keeps your Mac on. It doesn't let it go to sleep unless you tell it to go to sleep. So like you can, you can leave your desk and then still be chatting with your AI agent on your computer. And then when you come back to your computer, it's right where you left off and you keep, keep going. So that's like features like with Claude, it's called remote control. So you can use AI, like you can continue the chat that you were having on the computer that has access to everything because it's on, on your computer. But I can control it from my phone while I'm out for a run. And it'll ping you and say, hey, I have a question on how to proceed with this step. And rather than waiting till the next m morning to say yes, and then it starts thinking, you could get that text at five o' clock, or not the text, but the notification, you go in, push it and then it can think all night so the next morning it's actually ready for you, you know.

Speaker C: And so, um, when it comes to data and these two ladders, like who you are as an operator, how you use AI and how your data is structured in order to utilize your AI better. When it comes to someone who is starting out and like, where do I begin? We really press hard that the most important thing that you can do for yourself and your organization is to get really clean data. And clean data is. It's not a gimme, it's hard. It's like, how do I get all of these things to talk to the same place so that all my information lives in one spot so that my agents can act on it? And the first step is you got to use the right tools.

Speaker B: Um, we, yeah, there are two. It's, it's like we'll have conversations with business owners and they're using softwares that literally just don't have an API at all. An MCP is even more advanced, doesn't even have an API to go in and get information. So the only way I can get anything out of it is by using a browser or the app on my computer to go read the files or export a CSV or something because it doesn't have the ability for a computer to go automate that task. So when you're selecting tools to be part of your stack, that's a big, big thing that you have to be considering is. And that's. I feel like anytime I am in a software demo and they're showing all these features, I've gotten to a point now where the first thing I'll ask is like, hey, can we just start with your API and MCP connectors that you guys have? Because I don't care how cool the interface is. Because my thought or plan is that I'm probably hopefully not going to have a human using this as often as I'm going to have an agent. Yeah, a human might use it, but not. I don't want that to be. The primary reason I sign up is because it's. I could. This human can do a really cool thing when they use the app and

Speaker C: just take a step back to what is API? What is mcp? If you just think of Gmail, if you go in and click a button like I want to click who I'm sending this to, I want to click into the CC and I type some stuff in and I fill in those categories.

Speaker A: Mhm.

Speaker C: That's user interface. Ui. API is the ability of a software to have a key or a doorway to those same actions and perform that thing for you. And so there's different scopes for API. You can read only you can write also. So if you want to only allow something to go and read all those objects that a human could like type and enter and press send, that's a

Speaker B: not have the ability to accidentally delete something, right? Exactly.

Speaker C: So it's like however comfortable you are with AI, doing things for you is up to you. But when Chris is talking about what is your API API ability? If I have a software that has UI ability, I can click this thing and add those people and send that stuff. But it doesn't have API capability. That means that I couldn't have an agent go grab it or go do it. And so in Google Drive or Gmail, what doesn't have API is you can't go change the logo in the top left of your Google. Like it's still going to be that mail thing. So you can't do it. API can't do it. Maybe I'm wrong about that. Maybe you can't change your logo. But you get the point. There's some things that you can't go in and change. But if there's anything a human can do in a software, it is critical that that software is designed so that it has API capability to do the same thing. And so that's a really, really good first question to ask when you're out shopping for tools.

Speaker A: Chris, it's funny, you and I, it may have been our last podcast. Um, maybe we. No, no, no, it's not our last podcast. It was maybe two, two back, but we did a whole podcast on the importance of clean data. And this was back in November of 2023, before all of this crazy AI stuff came about. But you were saying, like before you even think about automation and like these node code tools, you've got to get your clean data.

Speaker B: That's the principles. The principles are like, they stay true. And yeah, if you think about AI is just more advanced automation. It's just more of the same thing we've been doing. Like computers were the start of automation 50 years ago, or whenever computers started happening, you know, like the initial punch cards and stuff, those were, that is an automation. And it's just been this progressive automation all the way through. And we've just decided to call it different things. But at the end of the day, AI is just the most advanced automation that we've ever come across and it can do all sorts of crazy stuff

Speaker A: from a, um, scorecard perspective. Um, code that was awesome and that you showed us. And so I want to encourage listeners to. You can check out kind of the YouTube video of this where Cody actually showed his screen and walked us through the different levels from an operator standpoint and a data standpoint. Um, you guys do have a scorecard available. And so I want to make sure that people know how to access that because I think it's, it's, it's powerful when you realize where you are and uh, what, what kind of the next step is. I mean, I'm. When I think about myself from the operator standpoint, I'm in that consolidated, sometimes manager level. But when I look at the data side, I hesitate to tell, to like, give access to my, to Claude, to be able to go into my Google Drive and do all of these things. So I'm, I'm inhibited in my growth on the operator side because I've, I've kind of stalled out on the, on the data ladder. Uh, and that I feel like that's probably a normal, a normal place that a lot of people of my level find themselves.

Speaker B: Yeah, I mean, I would say. I would 100 say there is, um, what is the word? Fear around that, which we don't understand. You know, if, if AI is just like a black box and it's like, I put stuff in and then stuff happens and then you have all these just like the concerns and people talking about like, everything bad that can happen. Right. And it's just like, yes, you need to know about that. But I would go back to like, AI is not going to lie to you specifically when you're asking for things that are verifiable where it can, like, look up API documentation. When people say, like, AI lies to you, it's because you're asking it to come up with, um, like I said the bio thing, like, write a bio for me. There's nowhere for it to ve. If you said, go and look at my website or my blog and write based on anything you can find that's documented and give me a bio, that's going to be a different bio than if you just said make an impressive product bio for me.

Speaker A: Right.

Speaker B: But when you're saying, like, I want to do this thing safely. And so like, on a scale from 1 to 10, with like, security concern, I'm at like a 9. So just keep that in mind, Claude. Give me all of the steps that I would need to follow to put protection and barriers and, uh, approvals and ask for permission, all that stuff. And that's when it's like, it'll tell you exactly what you ask it. It's not the one thing that was. I don't know, I feel like it's just important to understand is that we are. AI is not like some nefarious, like, um, hacker that's like, trying to get in your good graces and then just go rogue.

Speaker A: Okay.

Speaker B: Right. It's like reassuring you if you build it from the, from the beginning, you have control over this whole thing. And so you can say, like, you are never allowed to send emails. Right. So, and, and I might even say I'm not even going to allow you the API access to send emails because, because of that. But I could also give it access. Say you are never allowed to send emails without, like, my express permission to.

Speaker C: Yeah. And in my experience, I have been in spots. It was frustrating. I was like, hey, I just do it. And they're like, uh, are you sure? Like, I don't. I'm about to do this. And I'm like, yes, just go. Like, that's what I asked. Just do it. So it's, it's really cautious. Yeah, but again, nothing wrong can happen. But.

Speaker A: Yeah, uh, yeah, but the.

Speaker B: But the key, I would say is the best analogy that I've heard is that you need to treat AI like this brilliant, super smart employee that you just hired. And they come into your office on day one and you're like, I'm not giving you anything, uh, but I want you to help me. And so they're like, okay, I'll just stand over your shoulder and wait for you to ask me a question. Then I'll tell you what I would do. But it's like how much more value you're going to get out of that employee if you're like, you know what, Bob, you just started. Here's access to the CRM. I just made you a login. And you can also, like, when you're giving it permissions and you're giving it API keys and passwords and things like that, you can give it its own accounts and its own passwords so that when something happens, I can see who edited it. Like, I can trace it back. It's not just like, I gave it Chris's login. So now they think Chris did it when Chris didn't actually do it. I can say this and name the thing, like, this is Claude Codes access token. So when Claude is editing something, I can trace it back and be like, hey, you did this on this time. Why did that happen? And you know, go, go figure that out. But the, again, the key analogy is like, you have this unbelievably smart and gifted thing that just walked into your office and if you don't give it access to do anything, like what do you expect it to do, you know? Or let's say you give it access to your CRM, but you're not going to tell it your sales or marketing goals and you're just like, well, just like, do what you can. You, you know, and it's like, it'll do stuff, but if it has, the more context it has. Like, hey, this is our initiative for this quarter. This is our annual goal. Make sure that we're, we're tracking with that. Let me know when we start drifting from it or whatever. Like, that gets much more useful. You know what I mean?

Speaker A: No, it makes a lot of sense. And I'm glad that you mentioned this because it brings up a question that you and I had talked about when we were chatting a week ago is kind of the cost of all of this. And there's, I think part of the AI Fear and the, the employee fear is that I'm going to be replaced, that this is so much cheaper for AI to do my job, um, than what my m. The salary I'm being paid. And you know, I was asking you, you know, what level of um, different tools do you pay for? I mean I pay for the $20 Claude a month for me. We have different ones for our team. Um, but then you, you brought up that AI is heavily subsidized and we need to keep this in mind from a cost perspective. Can you just kind of bring us into that world a little bit? Because this was, this was new to me.

Speaker B: Yeah. So we're right now in um, in the AI world. It's being coined as the subsidy era. So we're in the subsidy era of AI. What that means is there's been roughly a trillion dollars in the last three years, three to four years, roughly a trillion dollars that has been invested into infrastructure to make AI work. So that's another thing to just like keep in mind. It's like you can get little like local models on your computer, but the MOP. 99% of people in the world that are using AI are using these models that are powered by these massive data centers that have huge footprints and consume a ton of power and energy and just resources to make them. So we've invested as like these major companies. Google, Amazon, uh, Nvidia, Anthropic, all these companies have invested a trillion dollars just in AI. And we're getting to use it right now and have been able to for. I mean what the. I don't even know what the cheap plan is anymore because I, we upgraded so like months, year, like a year ago.

Speaker A: I think it's 20. I think it's what I'm still getting.

Speaker B: It's like basically what I've heard is that the, what we're at right now is the equivalent of the government just giving everyone in America brand new iPhones and brand new AirPods, like fifteen hundred dollar things. And they're saying twenty bucks a month you can use it and then we'll give you a new one every so often. So you get the latest model and it's like it's not sustainable. The only way that this is sustainable is if the uh, amount that the users are paying for these tools, it has to increase. There's like, there's a, it either there's this bubble and everything blows up and you know, the world's over, or there it actually starts balancing out where the cost that they're capturing starts to catch up so that all this investment is like not only does it pay off but like they can continue to invest because there's only. And um, it's not an infinite amount that they can invest. Right. So that's the first thing is like, so bottom line is right now, and there's some studies, I was trying to get some hard numbers on this. People have looked and this is as of like April of 26, a $200 a month cloud plan is roughly the same as $5,000 of compute if you paid through Claude's API for their usage. So they're giving you these tools if you use it inside of Claude directly in their native apps, apps, they give it to you at a cheaper rate because they're trying to gain market share. And then codex for chat GPT, same thing. 200 chat GPT pro plan is about $14,000 of compute that they're currently giving you. So when you use it and you're paying 200 bucks a month, what that's saying is if you like literally were leveraging this thing to the max and every, you were just using it non stop for the month, right? Which most people aren't. But let's say you were just leveraging this $200 plan to the maximum. It would be the equivalent of you spending $14,000 per month to get something that they're selling for 200 bucks right now. So the, the takeaway we're like, okay, these are big numbers, Chris. The takeaway is like, I don't know, like first of all what's going to happen, no one actually knows. But what we can predict or theorize is that first of all the AI models will get more efficient. So maybe the compute cost, they're like say for the same intelligence output, they're able to do that more cheaply. So that's certainly going to happen, but not, not it's not going to cover that entire gap. And but what is likely going to happen is the cost of doing something today that we're paying 200amonth for is certainly going to go up when we don't know, but it's probably within the next couple of years at the very most people, that's what they're guessing. And what that means for the people using it is that you need to make sure that you're building software and scripts and efficient ways of accomplishing tasks. So if you're, if you're using AI and building all these skills and the skills like the whole like software, that that's actually another thing Is that when people talk about like what is a skill or an agent, it's really just a uh, software that doesn't have a pretty user interface. It's just a chat window. But you're actually, you are making apps like software right now, Molly, by the skill that you made. That's a software. It's just a very small like laser focused niche tool. Right.

Speaker A: So we've never known.

Speaker B: But what you need to make sure is that instead of when you run your software program, if it's just going and using a um, language model to solve the problem for you, you're, you're paying API rates or you will be paying API rates soon versus if you try to convert that into a Python script or JavaScript or some sort of app that runs for like pennies to do the same exact thing. So you, you're trying to use what they call deterministic models, which is software like really good at math. So one plus one equals two. Language models are not that great at math. So like, or going and scraping large date large amounts of information. That would be a great thing to have a tool that goes and like researches your database for keywords and then returns that back. That could just be a script that goes and does that doesn't need to be a language model, but then the language model has to do the logic and the thinking part. So making sure that you're building apps in an efficient manner so that the cost of using these things doesn't go, you know, balloon on you and then all these things that you were planning on using are no longer, you know, it doesn't. Pencil.

Speaker C: Yeah. And uh, there's a couple of.

Speaker B: Or go ahead.

Speaker A: How would I, how would I. And this could be a whole other conversation. So just tell me if that's the case. How do you take, see the skills that I've, that I've built and I'm using. How do I make that into the uh, how do I take that out of, out of Claude and start using that in an external space?

Speaker B: So I'll let you try to guess what the right answer would be.

Speaker A: Do I ask Claude? How do I do that?

Speaker B: There you go. So you just moved up the ladder. There's another like micro step upwards.

Speaker C: Yep. And I gotta jump into really quick with this, what Chris is describing, like go and like build the tool or get someone to build you the tool that like does these things that you're like using all these API tokens for. Going back to what we said at the very beginning of this and what I Said is the most important thing for a business owner. Tools are worthless unless they are built on top of really, really good data that stays fresh, that is refreshed with tools that you made. Like, instead of saying, hey, AI agent, go and find this information that you're connected to and bring it back and put it in the database. Molly, you should ask Claude, how do I set up something so this happens automatically? And then the only thing I'm paying for is keeping it fresh once a day, once a week, once a month. Because that takes a brain to go in and, like, read all the things and then, like, give you your headlines. Because Chris and I have really delved into build the tools. We make internal tools. We make tools for our clients, and we watch a lot of people talk about how easy it is to make tools. We went to this conference in San Francisco, just saster conference. It's just AI tool, AI tool, AI tool. We'll build you this thing, that thing. And we were like, uh, one of our big takeaways was this thing that you're advertising that you can make or you're showing me. It's like, yes, anyone can take a piece of paper and draw a house floating on a cloud and say, look how amazing this looks, and look at all the things it could do. You know, it's got a water slide and a, you know, portal that teleports you to whatever. It's like, okay, that's a cool drawing. But does it actually work or do the thing? Because right now it's really, really easy to make apps that look like really cool drawings. But for it to actually integrate with all of your data across all of your systems is the thing that's really, really hard and where people should be spending their time right now. Because it will never be cheaper to on your own, go and try and tie all these things together and then build the tools on top of that data.

Speaker B: Yeah, and I have two, uh, well, three quick examples to just, like, try to put what I'm saying into more tangible, like, use cases. So this is an app that we vibe coded. We were able to vibe code this app in about three days for one of our clients. The reason we were able to do this so quickly is because the data underneath it was already built. So this in particular happens to be running on top of airtable. But the. The problem that this business owner had was I have like four or five different screens up on two different monitors when I'm trying to schedule my projects for. For the week. And he's trying to Optimize. He's in Atlanta, Georgia. He's trying to optimize drive time for his PM so he can intelligently like plan projects to happen that are in the same geography at the same time, rather than just doing them in the order that they were sold. So in this case, like this, this tool is like incredibly powerful because it's pulling in Google Maps real time here. So I can like go in and like zoom in and see, okay, here's a project I can jump to the week of June 8th. It's showing me, okay, this project right here is on this day I can open up that project and say, okay, um, let's go into the, um, right here. So this little window I can go and say I want to send a notification to the user and I want to send a tentative confirmation. So it's pulling the data from the system and it's going to send a text message and an email if I wanted to do that. Or I could modify or I can send my confirmation that this is when your project's happening and I hit send and it sends a text and email directly from here. So these are all things that Seth had to do in different apps and different tools. But he had a very clear vision of what he wanted. And this is where I'm talking about. We used AI to make the app. But for Seth to continue using this when it, uh, when AI prices go up, this is going to cost next to nothing to run. Because this is a fully functioning app that's just integrating through airtable's API to his data in airtable. And so anytime he does something in here, it just writes back to airtable. But it's like a fancy widget or window into his data that's pulling all this information that lives in different spots inside of his airtable into one interface. So that's one example. Another one here is like, um, ah, we'll go to this one. This is a dashboard. Again, we Vibe coded this, um, but it's like the master dashboard for a business owner to see everything from their sales and marketing and their CRM, their production inside of their smart sheets or Airtable or whatever tool they use there, and their marketing spend data from, uh, QuickBooks. So it pulls everything together and compares it against the plan that that business owner had for the year. If, uh, we want to do 5 million or 7 million or 10 million in revenue. And then it works backwards and it's flagging in colors here, all the different places where you're slightly missing or you need to, you know, you're. We're behind our lead target. So the, the owner can now come in and click in and say, okay, what happened with these? Um, the set rate here? We had a 65.3% set rate. So I could go in and look at all the marketing qualified leads that we had and I can actually open that lead up in HubSpot and go back to it and verify that it is what I thought it was, if that makes sense. So there's like this again, is not running on AI. It was built by AI. And now to continue using it, it's just using some API calls to the source data and pulling it all in here.

Speaker C: And that's the most important part.

Speaker B: Yeah, yeah, that's the uh, thing I'm trying to illustrate is that when you're leveraging AI right now, make sure that you're, first of all, you have clean data so that you're able to get the most out of it. But if you have clean data, use AI to build these apps. The apps are like almost. It's kind of weird because it's a very cool tool. For Seth, it's saving, saving him time, but the cost to make software is going to zero. It's getting easier and easier to make these little mini applications. And Vibe coding is like a term that's becoming very prominent. But the, the whole point is as you have to make sure that these Vibe coded apps that you're building are not like dependent on a ton of uh, API compute, or sorry, not API, uh, AI compute to do the result that you're hoping for and instead build it into a system that is like, uh, that's leveraging actual like code and software to run instead of a language

Speaker C: and that it is actually built into the system. Because these things are useless. They are just like the houses floating on a cloud, unless they are integrated to all of your other things. Because what this app isn't is open it up and enter all of the data so you can see it in front of you. All of this data was entered by someone else at their different tool. Yeah. In the, in a different tool at their stage of the job, life cycle. And for like our business, it could have been the coordinator, it could have been the sales estimator, it could have been, you know, project manager. But then all that information that they entered at different spots in their workflow in their software is connected to the things that we are building and that other people are able to build on top of that. And so then all the information is like visible and actionable. In the place that you want it.

Speaker A: Two questions and then we'll wrap this up so we don't overwhelm our listeners. Because I know we could spend hours kind of talking about all of this and maybe that makes for a follow up conversation in the fall here. Um, but I just want to make sure that when we talk about, we've hit, you know, clean data. And again, Chris, we did a podcast again back in November 2023. We, when we say clean data, I want to make sure that people really understand what that means. And if I interpret that correctly, I look at as. Look at it as. I mean, you just mentioned Cody. It's someone entering data like a coordinator, but it's been entered once. And whether it's the formatting, whatever it is, it is able to be transferred through API, through computers, talking to one another. It's able to hit. It's able to travel from one box to the next without, without breaking because of how it's been entered that m. It's been accepted. It's being accepted the same way in all the different platforms.

Speaker C: And when you change it over there, does it change over there too?

Speaker A: Or break? Yeah. Uh, is that the right way to describe it?

Speaker B: On a very basic level, I like that I will add a few things to it. And that is clean data is that it's structured well. First of all, like, is it, um, do I have all of the data points in their own fields so that they're searchable, sortable, filterable, et cetera? And that means like, you could, someone could be like, yeah, I have all the data in a Google spreadsheet and you go look at it and it's like they just put a whole bunch of stuff in one cell of the row. But it's there but like in a human. And even AI could look at that. But that's an example of like, you might get a lot out of AI by looking at your really complex, poorly structured Google sheet. And it's working fine. But it's only working because AI is looking at this mess of information and processing it and then giving you an answer based on what it just read. Like a human would do it. Structured clean data would be like, I know that my project types are interior, exterior, uh, carpentry and cabinets. And those four property or project types exist in my CRM by the same name. They're in my project management by the same name, they're in my QuickBooks by the same name. And you go through all the systems that you have and you're, you're setting that up. So that's the first component is just like you've thought through what data points are needed to generate the reports that you need in your business and are they consistent across all tools? Because when you send data from one to the next, they need to. And then the only other thing that you didn't mention is, is it actually timely and accurate or accessible? So it's like Cody saying, is it up to date, clean? You could have really well structured data, but it starts to drift because it's not getting updated. So like, maybe you get a customer's information in your CRM and then you find out when you go there for the estimate that it was actually the wrong email address and you got to update that. Like, is your CRM getting updated with that corrected email or. So it's things like that words. Okay, it might be correct in one system and not the other. And do you have like, processes built in place to keep it clean so that you're still getting value out of that with. With all the tools that you're putting on to it?

Speaker A: Thank you for clarifying. And then I, I think it's important that we just make sure if that's where the most important place is to start. We want to make sure people understand what that means. Second question I have, and hopefully it's not opening up a can of worms. It's, you know, we talk about, okay, these apps that are now running outside of the AI, you know, the AI platform, like the ones you just showed us, Chris. But what happens when these things break? Like, where do we go to. Where do we go to fix them if we've taken them outside of the AI platform?

Speaker B: Um, interesting. So it depends on how you've built it. Okay, so I actually showed you two different tools. One was built in Claude code. And so, uh, when there's an issue or when something needs to be updated, that's an example of where you are using AI to fix the breakage.

Speaker A: Okay.

Speaker B: But you're not having to use AI through the, like, a record, like a typical, like, workflow that you're doing. That's where. And again, we want to use AI where it's necessary to use AI. Right. That's the key. You are going to use AI when you're modifying and editing code. What you're trying to avoid is building AI as like the fundamental backbone of a process so that in order for it to run like a, a workflow, it's using 80 AI. It's like, can you get it to use 10% to like do some refinement and most of it's done by a software. So the breaking. Yes. All things, uh, all tech needs to be maintained and you maintain it in the same way that you built it, which is through like vibe coding. When you're doing that, you are just saying like, hey, I'm getting this error message or I want to add this feature. Can you, you know, update that? And then it will go and update that.

Speaker A: Got it. Thank you. I'm glad it wasn't. I didn't open up a huge kind of next, next level tangent there. Um, thank you both. There's obviously we, we kind of hit. We got a lot out of this. I feel, I feel like, like I said, we could go on for hours. There's just so much beyond the surface. And clearly when you talk to two passionate people like yourself, it could go on and on and on and we could, you know, continue to learn so much. So we'll definitely have to have you both, both back to continue this conversation. Um, I want people to be able to find and access the scorecard. Obviously they can watch the YouTube version of this video and kind of see it laid out. But how does somebody, somebody measure where they are at on, on the ladder of operator and data?

Speaker B: Yep. So we put it on a super simple, easy to remember, hopefully in type domain. So Paint OS app/AI Paint OS app and there's two Ps and app app/AI. Um, and that'll take you straight to the like the scorecard landing page. It's just like a self help. You answer series of questions and then it gives you the thing that I think is most valuable about this is like Cody is saying the number of people that they'd be like, what do I need to, what do I need to know? How, how should I leverage this? And it's not useful to just jump in and start giving feedback until it's like if you're going to do math class, right, take an assessment to know what level of math you understand and then we'll start from there. It's the exact same thing. So I'm actually super excited about this because it does that quick assessment in a matter of minutes. Like answer these questions and then, you know, tell us where you're at and then it will say you're here on the scale. These are the things that you should try to do next. Okay. Like very actionable. Like go give Claude this prompt. Go try this thing. And then it's up to you. Obviously you got to go study the math and and take, like take the, the, the um, swings at it and then you come back and take it again and then say, okay, what, what else should I be doing? And if you do it well, you might like after doing it the first time. As you move up the ladder, assessments like this become less and less valuable because you can get Claude to essentially assess you. You know, so if we, if we do our job right and you, you take what you're supposed to from this, you should be able to continue to self level up. But until you get to that tier, which is probably like a three or a four, like come back and, and see what, what would be the next best place to spend some resources and time and energy to get better.

Speaker A: Awesome. We'll be sure to put um, that, that URL in the show notes here as well with the description. Um, guys, thank you so much. I appreciate it. It's such a good conversation. I learned a lot. I know that. I'm sure our listeners did as well. So I'll just say until next time.

Speaker C: Thanks, Molly.

Speaker B: Thanks, Molly.

Speaker A: Thank you. Thanks for listening to this episode of out of the Hourglass. This podcast is recorded and produced by the team at Nolan Consulting Group, a nationwide coaching firm built specifically for leaders in the trades. If today's conversation sparked some ideas and you're curious about what coaching could look like for your business, we would love to connect. Visit nolancg.com to learn more. Have a question, comment or idea for a future podcast episode? I want to hear it. Subscribe wherever you listen and we'll see you next time.