Brad Johnson: Well, Michael, we’re back.
Michael Hyatt: Back again.
Brad Johnson: I feel like you’re becoming just a regular on the show, which I love.
Michael Hyatt: I love that. Well, I’ve told you before, you’re one of the best interviewers I know, so I’m always happy to do a podcast episode with you.
Brad Johnson: Well, this is a fun occasion. We’re in Nashville, and we’re getting ready to kick off the AI mastermind you do inside of Triad, which is going to be really exciting. In fact, we’ve got…
Michael Hyatt: Very excited about it.
Brad Johnson: …some members getting ready to roll in. But we thought this a cool opportunity to hit a different topic that we’ve never talked about before. And as I thought about why you, when it comes to AI, how we originally got connected was you geeked out on some different technology back in the day, blogging…
Michael Hyatt: Yes
Brad Johnson: …before that was a thing, and that was how we originally got connected. But you accidentally built one of the biggest blogs in the world at the time, which led to your New York Times bestselling book, Platform, which was how we got connected. And you’ve just kind of had this knack, as I’ve gotten to know you better, whatever’s next, you’re a tinkerer, whether it’s the new technology, the new iPhone, the new MacBook, whatever, but also with what I’d say technology trends, early internet, blogging, podcasting, now AI, and you’ve dedicated multiple years not just to like, “Hey, how’s this work?” But literally diving in deep, building your own agents, figuring out what breaks, what works. And I love that about you. I’ve learned so much from you.
Michael Hyatt: Thank you.
Brad Johnson: But I want to kick this off with maybe a controversial question. Why do you think that ignoring AI is one of the biggest mistakes financial advisors could make?
Michael Hyatt: I think this is probably the biggest technological change that we’ve seen in the history of humanity. And I don’t say that lightly, because the internet was a big thing, right? The smartphone was a big thing. It changed all of our lives, and it’s hard to remember what it was like to not have a smartphone.
Brad Johnson: Yeah.
Michael Hyatt: And I think AI is that times 10. And I think there are a lot of people that will benefit from it and a lot of people that won’t. And I kind of consider it, and I’ve used the metaphor with you of a tidal wave. And I found this interesting story about a guy in Southeast Asia when there was a tidal wave coming in, and everybody was running for the hills, trying to get out of the way of the flood. And he thought, “You know, it’d be really cool to surf this thing.” And so, he swam out to it and surfed it in. Now, I think that the people that are aggressive like that, that embrace AI and are willing to get a little wet, to take a little bit of a risk, are going to benefit enormously from AI. But I think this is one of those things that if you stick your head in the sand, and if you don’t engage with it, you’re going to be a casualty.
Brad Johnson: Well, that’s a pretty bold statement, but I can’t say I disagree with it. I mean, I look at, you know, there’s the doom and gloom side of AI at this point, and then there’s the, hey, optimistic approach. I think whichever side of the fence you’re on there, what’s interesting is Pandora’s box is already open.
Michael Hyatt: It is.
Brad Johnson: Like, if the government tries to regulate it in the US, guess what happens? Now, China gets ahead. And so, it’s like this arms race, like back in the atomic bomb days, back in World War II, it’s like neither side’s going to stop working on it. So, now, to your point, this is the future, this is reality. You need to embrace it. So, one of the things that you also shared with me, it doesn’t start with just opening up the laptop, downloading Claude, Chat, Gemini, whatever happens to be the leading agent. You said it actually starts up here.
Michael Hyatt: It does.
Brad Johnson: Mindset. So, why does it start there when everything is basically on your laptop and computer?
Michael Hyatt: Well, because the mindset is what’s going to carry you through when you get frustrated and when you kind of want to stick your head in the sand. And I’ve seen this over and over again. Like, first of all, I’m 70 years old. This is not an age thing. I have a guy in one of my masterminds that is older than I am, and he joined the mastermind, and that’s an AI mastermind as well. But I could see in the very first time we met together, he checked out, hasn’t been back since, and he paid a lot of money to be part of that mastermind. Now, he may be thinking to himself, and I don’t know, I can’t read his mind, but he may be thinking, “Whoa, this is over my head,” or, “This is intimidating,” or whatever, but that’s a mindset issue.
Meanwhile, I was telling you my dad, who’s 92, is using ChatGPT every day. And so, apropos to your audience of financial advisors, my dad used it most recently to work out a financial plan for his later years. He doesn’t have a lot of means, doesn’t have a lot of money, but he thought he’d go to ChatGPT and see if he could figure out a plan. My dad has always been like that. So, back in, I think this might have been around 1979, my dad comes home with a RadioShack TRS-80. Do you even know what that is?
Brad Johnson: No. I mean, I remember RadioShack as a kid. They don’t exist anymore.
Michael Hyatt: They don’t exist anymore.
Brad Johnson: But this was before Apple, before IBM.
Michael Hyatt: Yeah, IBM, PC, before any of that. It was like RadioShack’s super hobbyist-level computer. You know, these are guys that are, you know, before that were building transistors and shortwave radios and that kind of thing. So, he brings one home, and I mean, it’s just like a blinking green cursor. It has no real ability, but we sat there in front of that thing, and I was in, I think I was in high school.
Brad Johnson: What was the language? This was probably before Microsoft DOS was created, right?
Michael Hyatt: I don’t know. It was. It was some kind of operating system. I don’t even remember.
Brad Johnson: Wow, okay.
Michael Hyatt: But then in 1981, when the IBM PC came out, I bought a $5,000 IBM PC and I was…
Brad Johnson: You personally in 1981.
Michael Hyatt: I personally.
Brad Johnson: How old were you at the time?
Michael Hyatt: I was just out of college.
Brad Johnson: Wow, okay.
Michael Hyatt: And so, I paid $5,000. My salary was $25,000. So, this was…
Brad Johnson: 20% of your income went to your PC?
Michael Hyatt: Yeah. I borrowed the money, which was kind of against my principles.
Brad Johnson: How much was a car at that point? That had to be not too far off from a car.
Michael Hyatt: I don’t know. Yeah, maybe $10,000 or $11,000. I don’t know. Believe it or not, my wife was enthusiastic about it. And that gave me a headstart on technology, so that at the company I worked at, at a relatively small, probably 200 people size publishing company, and I started building applications on that IBM PC. And it put me so far ahead of my peers, and it’s an investment that I would absolutely do again. Best investment I ever made in myself. And so, I’ve always wanted to build and wanted to create and wanted to learn, which is kind of the first, I think, mindset attribute that you have to have, which is curiosity.
You know, if you’re not curious, you’re not going to survive what’s coming down the pipe. Because it’s easy to get overwhelmed by it, but that kind of leads to the second discipline, I think, is your ability to persist. Because like anything else, if you can’t build frustration tolerance, you will not ultimately break through to where the benefits of AI are, and the benefits of technology are. And so, if you’re the kind of person that gets frustrated with technology, and my wife is a little bit this way, and it’s just like, “Ah, I just throw my hands, and I give up,” you’ve got to persist. And if you do persist, there’s a lot of benefit on the other side of it. And then I think the third discipline, if I were to give one, is the discipline of resilience.
You’re going to have setbacks. At the time we’re recording this, I have about 25 agents working inside of Claude. And one of them is breaking almost every day. And I get so frustrated. It’s just like, “Can’t you get this right?” Me speaking to the agent. “You know, why am I having to re-prompt you on this?” And it’s not like I haven’t built sophisticated memory so that all my agents have access to the same database of knowledge, the same skills. They do. But it still keeps breaking. But does that mean I’m going to stop? No, I’m going to keep going because I know that AI today is the worst it will ever be. It’s getting better.
Brad Johnson: That reminds me of a story. There was a guy, actually, he was a crypto expert, so this was early in the crypto days. And he was trying to explain it, right, like, “Here’s where it is. I know it’s overwhelming to a lot of people,” back to mindset. And he said, “You realize email, when it was invented, how hard it was to use?”
Michael Hyatt: So hard.
Brad Johnson: He’s like, “Literally, you had to have a computer science degree to send an email.” And he’s like, “Now look how easy it is.” And so, to your point, this is the future. It’s not going backwards. And so, now if you get a little uncomfortable, maybe in the tough parts, think about how far ahead, whether in business, whether in personal utilization, and I just think that’s so key. And guess what?
Michael Hyatt: Me too.
Brad Johnson: Business is hard.
Michael Hyatt: Yeah.
Brad Johnson: If building a business was easy, guess what? Everybody would be an entrepreneur.
Michael Hyatt: That’s right.
Brad Johnson: That’s part of the journey. So, anything else on mindset? Like, just other things like… Because I feel like you’ve done the same iteration during multiple technology cycles. Is there any other things that you’re just like, “Here’s other advice I would give somebody that’s just starting out”?
Michael Hyatt: Well, I’d say you’ve got to get in there and tinker and play. You got to try stuff, and you got to try stuff that makes you uncomfortable. Like, the first time I started building an agent, I didn’t have a clue what I was doing. But I just decided to roll up my sleeves, watched a few YouTube videos. But there’s nothing like actually playing with the technology to learn a technology, and you’ve got to be able to make mistakes, and this is part of resilience, is to reframe failure. You know, failure is really just feedback. I think Tony Robbins says that, but he’s 100% right. You know, it’s just feedback.
And so, when something doesn’t work, it’s like, okay, that’s data, and I’m going to take that, and I’m going to take it into the next iteration. I heard Elon Musk being interviewed this week, where he was talking about the three rockets, the first three rockets that SpaceX launched. Did you see the story?
Brad Johnson: They almost went bankrupt. I’ve read his biography, so I remember the story.
Michael Hyatt: Yeah, I just heard it on a video clip. And he was just saying that the first two rockets didn’t work, and everything was riding on the third one, and if it hadn’t worked, then they would’ve gone bankrupt. And now he’s, like, one of the most extraordinary entrepreneurs of certainly our generation, probably of all time. But it’s because he refuses to quit. And I think that frustration tolerance, and I would think financial advisors would have this in spades, because just getting through the certification process, the education, all of that proves that they have the ability to demonstrate tenacity and resilience, but they’ve got to use that with technology too.
Brad Johnson: Let’s start to apply this to finance. Obviously, that’s most of the audience.
Michael Hyatt: Okay.
Brad Johnson: I think back to mindset. If I was an advisor watching or listening in, I might start to create a little roadblock here saying, “Well, that’s great. You know, AI works for Michael’s business, and maybe he’s got these agents to help with the planners and everything that you do around Full Focus, but I’m dealing with personally identifying information, like people’s finances.”
So, there’s maybe a firewall in their brain. But what I’ve seen you do personally is before it gets to, like, your personal finances, there are lots of automations you’ve already created that don’t even touch what I would say is super sensitive information. But an agent is like, “You’re treating it like an employee.” So, what are maybe some applications, maybe this goes into your AI ascension model of, like, beginner to advanced?
Michael Hyatt: Yeah.
Brad Johnson: But if I’m an advisor out there kind of getting stuck, how should I start to think about it of this is the first step I would take?
Michael Hyatt: Well, you’ve got to make sure you’re operating with the latest information, because a lot of the problems that financial people have with AI are problems that either have been solved or are being solved. So, for example, you can connect Claude today using Plaid, which is a common interface that guarantees SOX compliance and all kinds of privacy and all that. You can connect that to QuickBooks, to your brokerage accounts, and all of that. And how secure is it? I don’t know. That’s what the advisor has to look at that. But I’m telling you, those tools are coming, and they’re going to get nothing but better. People don’t think twice about using QuickBooks, even the online version, even though that it’s going up to the cloud, they hope that it’s safe.
A lot of the issues, even with PII, are being solved as we’re talking now, personally identifiable information. But a lot of that’s being solved. And so, there are local AI models, then nothing has to go to the cloud. That’s one solution for it. Another solution is that you anonymize the information, and there are lots of tools like AI Guardian, like Redactable, and basically, when you submit something that has PII on it, the AI doesn’t know who it belongs to. It’s just a number, right? And even if you’re uncomfortable with that, there is so much that AI can help you with, all the stuff that doesn’t involve client information, all the kinds of what my friend, David Sparks, calls donkey work, the stuff that most of us hate, most of us don’t enjoy, but it’s the kind of repetitive process that can be specified in an SOP and handed to a computer or an agent and let them do it.
Brad Johnson: So, quick disclaimer, because I’m guessing compliance is going to review this before it goes out. I think one of the things that has always been tough when it comes to technology and finance, like if I look at other industries, finance is typically a laggard because it’s such sensitive information. If a hacker can hack a financial institution, that’s the goldmine they’re looking for, right? So, I do think even with the advancements in AI, finance will probably be, even if you want to be on the leading edge, there will be certain regulations that kind of maybe make it a little slower to the dance.
But that doesn’t mean you can’t get super familiar with your own personal application before you even apply it to your business, to where now you start to understand how to utilize the different tools as they evolve and come out. And I kind of feel like we’re in the dot-com era of AI. We know it’s the future. We don’t know who the winners and losers are going to be right now. So, I know one of the things you’ve even talked about is the mindset with the technology build. Like back in the day, it’s like, “Here’s my tech stack,” which is kind of static, like pillars. Now, it’s like the pillars get swapped out every other day, depending on which model is winning. So, how would you start to think about the mindset of how you even build technology with the way it’s just evolving like crazy right now?
Michael Hyatt: Well, it’s true because there are going to be winners and losers, and for example, for the last six months, I’ve been touting the benefits of Claude. I love Claude. But they don’t have access to compute like ChatGPT or OpenAI or Google or some of the other bigger companies that have ginormous data centers, even Grok. So, what that means at a practical level is Claude has, as we’re recording this, gotten increasingly slow.
Brad Johnson: May 27th, 2026, by the way, just for a timestamp.
Michael Hyatt: That’s right. So, like I was having it help me with an email, a response to one of my newsletter subscribers, and at the 10-minute mark, I’m like, “I just asked you to write like a four-paragraph email,” and it couldn’t do that. So, I bounced over to ChatGPT, and I’d pretty much dissed ChatGPT since I left it. But ChatGPT 5.5, which is their current model, again, as we’re recording this, is extraordinary, and it did the same thing. I tested it heads up. It did the same thing in 10 seconds. And so, what does that mean for the person that’s dabbling in AI? It means you’re going to have to build your infrastructure, separate and apart from the AI platform that you use, so that it’s portable.
So, for example, and I don’t want to make people’s eyes glaze over, but there are a lot of people like me that are building all their skills, all their knowledge base, all of that in something like Obsidian with Markdown files, which again might make people’s eyes glaze over, but it’s not as hard as it sounds, so that I can access that single source of truth, whether it’s with ChatGPT or with Claude or with Perplexity or Gemini, and it’s pretty seamless.
Brad Johnson: Okay. Let me see if I can work on analogy, and if I butcher this, correct me.
Michael Hyatt: Okay.
Brad Johnson: So, I feel like data has always been a big thing. Why is Google such a valuable company? It’s not Google Search, it’s the data that they’re harvesting from everybody, right? So, they have this foundational data. I feel like you’ve got to have that for any business, so what is the base of the pyramid, for lack of a better term? So, now what I heard you say right there with Claude and ChatGPT, it’s kind of like here’s almost the processing unit, the horsepower that sits over top of the data to kind of harvest it and make it do the things you want. So, what you’re saying is data layer stays the same, because that’s not going to change, and now I can swap out, it’s almost like you’re putting a new engine in the car, and this one’s got more horsepower. Oh, wait, no, that one’s got more horsepower now. And so, you’re able to swap out what’s on top of it.
But then the other thing, it’s almost like the source files you’re talking about. You’ve written a lot of books. You’ve produced a lot of content. If you want it to speak or write like you, I guess that’s part of the data, but there’s like almost skills you can create on top of that. So, how do the pieces of the puzzle kind of fit together there? How bad did I butcher that, I guess, is what I’m saying.
Michael Hyatt: No, I think you’re fundamentally right. You know, I think that those engines, those frontier lab LLMs, like ChatGPT and like Claude, that’s like a horse race. And they’re going to be passing one another, leapfrogging one another, until probably the end of time.
Brad Johnson: Until eventually there’s one or two winners, probably.
Michael Hyatt: Yeah, exactly.
Brad Johnson: Like Apple and Microsoft.
Michael Hyatt: Exactly.
Brad Johnson: Yeah, okay.
Michael Hyatt: And I think that’s already beginning to shake out. I think it’s likely going to be Anthropic, and it’s likely going to be OpenAI. I mean, we’ll see, you know?
Brad Johnson: And why are you betting that? Like, what are you basing that on?
Michael Hyatt: Because they’re the ones that have the largest installed user bases, and Anthropic has really come on in the last few years. And I think that… Last few years, I meant the last few months, they really jumped.
Brad Johnson: And Anthropic is who created Claude.
Michael Hyatt: That’s right. And OpenAI created ChatGPT. They’re both extraordinarily well-funded, right? And they just, you know, it’s really interesting because OpenAI bet initially on a consumer product. Anthropic went the enterprise route, and that’s where the big dollars are, and that’s why their growth has been so extraordinary. And talk about 10X’ing their business. They’ve just this year in 2026, I don’t know what the percentage is, but they’ve probably doubled the size of their business.
Brad Johnson: Kind of sounds like the old IBM versus Apple playbook, because IBM was heavy enterprise business users back in their heyday, right? And Apple went direct to consumer, which…
Michael Hyatt: And there’s room for both.
Brad Johnson: You could make an argument there, yeah. Absolutely.
Michael Hyatt: I mean, those are all multi-trillion-dollar businesses now, which is astonishing.
Brad Johnson: Okay. We’re jam-packed. We’re getting nerdy here, but I think we’re tying it all together very well.
Michael Hyatt: We got to get into the ascension model.
Brad Johnson: Yeah. So, all right, back to I’m an advisor listening. I was actually a computer information systems major.
Michael Hyatt: I know.
Brad Johnson: So, I’m a recovering nerd. It’s all come full circle, in IT out of college, into finance, now it’s coming back full circle back to tech. Let’s kind of speak to two different viewers. We’ve got kind of the mindset. We’ve got a super successful advisor, maybe they’re a little bit older, not that that matters, but maybe they’re like, “Nah, I didn’t grow up with technology. This is a little bit overwhelming.” And then maybe you’ve got a younger advisor that kind of geeks out on tech and loves it, is kind of that early adopter like yourself. How should both of those two individuals that both happen to be very successful financial advisors start to embrace AI based on where they’re at? Like, do they have different starting lines? What’s your model that you think there?
Michael Hyatt: Well, I do have a model called the AI Ascension Model, and it does explain where this is all going. It’s basically a roadmap for anybody at any level to define where it’s leading, for starters. I think we’ve got to have a vision of what AI is going to do for us. It’s kind of like there’s a lot of people that are into productivity just to be more productive. And those people aren’t actually saving time, because they spend a lot of time building systems, they spend a lot of time geeking out on technology, and their lives are full of that. So, I start at the very end. What’s the goal? And I think this is why most entrepreneurs get into business. They want freedom. And in my book, Free to Focus, I defined four different kinds of freedom that I think are incredibly important.
First of all, the freedom to focus. AI is going to enable us to do the most human things we could ever do, and to focus on those things where we really add value. It’s going to be able to take over, as I shared earlier, the donkey work, the routine, the boring, the recurring stuff, so that we can really add value where only humans can add value. And one of those is to focus on that kind of work. The other kind of freedom, I think, is the freedom to be present. And when I go home from work, when I have my grandson drop by in the middle of the afternoon, I want to be fully present and not feeling like I’m constantly looking at my device and constantly sidetracked or distracted. I want to be in the present, thinking about the people I’m with.
And I want them to be able to say, “You know, the thing about Dad is he was always very present. He looked me right in the eye. He listened to me. He felt what I felt.” And that’s a human thing. No AI, no matter how much they mimic it, is ever going to get to that point, in my estimation. Then the third freedom is the freedom to be spontaneous. And I don’t want to have my life so scripted that I can’t, like, decide in a whim to go golfing with a buddy this week because it wasn’t scheduled. I want to have enough margin in my life so I can be spontaneous. Or like the other night, Gail and I watched two episodes of a show we love, and it was just spontaneous.
Wasn’t part of the plan, wasn’t on the calendar, wasn’t on my task list, but I want AI to free up that margin on my calendar and my task list so I can do what matters most. And then finally, the freedom to do nothing at all. And I learned this when I was in Italy, and they even have a saying for it. It’s basically in Italian, the sweetness of doing nothing. And so, one of the things I noticed in Italy, in the bigger cities like Florence and certainly Rome, people go out after work, sit in an open square, drink wine, and just do nothing. Just socialize, and all that is so life-giving. It’s part of what it means to be a human. We’re not robots. We’re not human doings, we’re human beings.
And so, I think to have that destination clear at the beginning, that we’re not just getting into AI so that we can do more and let it consume more of our lives, but we want it to open up space so that we can do the things that are uniquely human, where we uniquely add value to our clients, to the people that we serve, to our business partners, to the people that work with us. Okay, so that’s the end game, and I call that level six.
Brad Johnson: Well, on that note, before you move on, I love the Italy analogy. I learned in Italy coffee is not made to be a to-go.
Michael Hyatt: Ooh, that’s good.
Brad Johnson: Our very first trip over there, my in-laws were a bit jet-lagged, so they accidentally slept in. And so, we had breakfast at the hotel, and they’re like, “Do you have a to-go cup to grab it and take it on our…” It was our Amalfi Coast tour. No, they only had China because it was meant to be sat there, enjoyed, and had conversation. I’m like, “Wow, that’s such an anti-American thing,” and I love it. And so, those little beautiful moments, I look at how that applies to finance. Finance is a relationship business because it is one of the highest trust businesses in existence on the planet because here’s Michael and Gail, who diligently saved, worked really hard, built some cool stuff in your lifetime, got paid nicely for it, and now you’re entrusting your nest egg to someone.
You better have a really good relationship and a high level of trust, and I do know there are studies on AI and empathy. But I’ll tell you, like, until Elon can build Optimus with skin on it, like, we’re a ways out until I think we’re being replaced. I don’t think…
Michael Hyatt: We have at least six months.
Brad Johnson: Six months until they put like cell something over the skeleton.
Michael Hyatt: No, I’m kidding.
Brad Johnson: Until it’s Blade Runner?
Michael Hyatt: I don’t know if that’s ever going to happen.
Brad Johnson: Well, I just think about that. I’m like, how cool of a business is that if you take the stuff that you procrastinate because you most likely aren’t good, and you don’t like it, and now you start to create automations that are doing it better than you ever could anyway.
Michael Hyatt: Yes.
Brad Johnson: And now it frees up more of this, like real connection, real conversation, real human stuff. I think the advisors that figure that out early, they’re going to have a superpower compared to others.
Michael Hyatt: And you know what? They’re not going to be distracted by all the back office stuff because it’s not been AI-ified. They still have to have humans do it. They still have to do it themselves. And a lot of their bandwidth, a lot of their energy, a lot of their focus goes to that. But if they could be freed up to just be engaged with their clients, to hear their stories, their dreams, and really be the person who facilitates them accomplishing all that, that to me is the value of an advisor.
Brad Johnson: Yeah. Well, back to data. We just touched on that a little bit ago. Financial advisors have the most powerful data in the world, and it’s private conversations in a first meeting, in a second meeting. I heard it said once, going to meet a financial advisor is like getting financially naked.
Michael Hyatt: Yeah, it is.
Brad Johnson: It’s uncomfortable, but it’s required to get where you want to go. And now they’re one of the most widely used AI tools in finance right now, and there are multiple AI note takers. Guess what? They’re going to capture it much better than you ever could because it’s a computer. And so, now if you’re harvesting that data, but now the future AI tools that aren’t even built yet, now you can start to apply, because Michael just mentioned wine and Italy. “Hey, we’re doing a wine event in six months. Which of our clients might be interested in a wine event?” Go scrub the data, which would’ve been a probably two-day project for an assistant that’s looking through spreadsheets, and you’re hoping you’ve got it somewhere in your CRM tagged properly.
Now, you’ve got an AI agent that could literally say, “Identify wine lovers. Oh, here’s your list of 20. Go invite them.” Like, there are just so many applications, I think we haven’t even dreamed of yet.
Michael Hyatt: So many applications.
Brad Johnson: All right. So, you kind of hit six. I didn’t let you go back to one.
Michael Hyatt: Yes.
Brad Johnson: How do you want to go through it? Do you want to unpack the six?
Michael Hyatt: Yeah. Let’s just unpack them. So, level one is what I call the oracle level, where AI is basically an answer engine. It’s a more sophisticated form of Google, but this is how my dad uses it most of the time, and it’s where all of us start, and I still use it. So, it’s not like you grow out of these levels. You’ll continue to use them. But if I have a quick question, I’m going to go to ChatGPT, or I’m going to go to Claude, and I’m going to ask that question because I know I’ll get a thoroughly researched answer. And by the way, AI used to really badly hallucinate, even a year ago. A lot of that’s been solved. It still does, but to a much lesser extent.
Brad Johnson: Can you define what that means for those unfamiliar?
Michael Hyatt: Yeah. So, a hallucination, an AI hallucination, is basically where the AI makes stuff up, and it’s partly because it was so goal-driven to please you, the user, that it would make up stuff. Like, I used to use it for writing, and I’d be doing research, and it would give me citations to papers that didn’t exist. And this was famously done in the legal profession, where attorneys cited sources in court and were called out by the judge or opposing counsel because it was totally made up. And those cases didn’t exist in case law.
Brad Johnson: Well, I will say I had, as a Chiefs fan, which I think we brought…
Michael Hyatt: I’m a total Chiefs fan.
Brad Johnson: We brought you on midway. I asked it how many Super Bowls Patrick Mahomes had won, and it answered it wrong. And I had to correct it. I go, “Actually, he won this year too.” Like, “Oh, you’re right.” But to your point, that was multiple years ago. So, I feel like the models have gotten better, but you also, I think, have a way to kind of use AI to fact-check AI, like a triangulation sort of deal.
Michael Hyatt: Yes, I do. So, like, if I write an article and have a bunch of citations, and I’ve done that in Claude, and Claude did the initial research, then I’ll simply go to Perplexity or go to Gemini, and because Gemini is built into Chrome, I’ll just open a Gemini panel, and I’ll say, “Do the citations in this article, basically prove the points the author’s making?”
Brad Johnson: So, it’ll go fact-check the article. And how frequently, when you do that, do you actually catch it hallucinating these days?
Michael Hyatt: Rare.
Brad Johnson: One out of 100?
Michael Hyatt: Probably.
Brad Johnson: Okay.
Michael Hyatt: Yeah. It will maybe disagree with the way that I cited it, or like, “Well, you might be stretching that point a little bit.” Be more accurate to represent it this way.
Brad Johnson: So, it’s like you’re being a bit persuasive with your point there.
Michael Hyatt: Which…
Brad Johnson: Which is kind of cool, that it’s pointing that out.
Michael Hyatt: Yeah, exactly. So, that’s level one. So, that’s where everybody starts. Nobody needs to be ashamed of that. That’s where I encourage anybody to start. Just start using it as a search engine. And in fact, Google, as we’re recording this, just announced that they’ve revamped Google Search so that the AI is even more built in than ever. So, that’s level one. We go straight?
Brad Johnson: Let’s go straight to two.
Michael Hyatt: Okay. Level two is what I call the consultant level, and this is AI as a domain expert. It knows your domain, it knows your organization because you’ve been feeding it stuff so that it knows that, but each session is relatively fresh. It doesn’t retain context across conversations or across projects, but its value scales with the prompts that you give it, and this is where you have to start learning a little bit about garbage in, garbage out. The quality of your prompting is everything.
And so, you start learning a little bit about that. Like, I not only need to give it an assignment, but I need to give it some context behind it, maybe some rules to follow or some description of the output that I want. It’s not very different from delegating to a human. The reason most human delegations fail is because the supervisor doesn’t give the delegate enough information to complete the assignment well. They’re expecting the delegate to read their mind.
And we do the same thing with AI. We don’t give it enough context, we expect it to read our mind, and then we get bad output, and we blame the AI when it’s really us. So, beginning to learn a little bit about prompting, I’ve got a tool that I created called Promptlee, with L-E-E. I’m sure we’ll put this in the show link. But it has a feature built in called Prompt Genie, which in four steps, if you can describe what you want, it will build a highly structured prompt that is killer, that will 10x the output that you get.
Brad Johnson: So, question on, so number one, Oracle, number two, consultant. You said two things that kind of hit home with me. Number one, a consultant without a long-term memory.
Michael Hyatt: Right.
Brad Johnson: Which I’m sure we’ll get into somewhere up the ladder there. So, it’s almost like you’ve got a consultant, but it kind of forgets between conversations, which would be kind of annoying if that was a real consultant, right? And then the other thing I’ve noticed, and this is kind of back to your point on hallucinations, I feel like the first version of ChatGPT when everybody started kind of messing around with it, you almost had to prompt it. Like, I go back to my coding days, almost like in kind of like blocks, like not quite if-then statements, but it was a little kind of nerdy prompting. I feel like a lot of the AIs I’ve used recently have evolved enough where you can almost like type it notes just like you would talk to a friend, and it kind of figures it out. So, is the prompting as important, like the structured prompting today as it was in the early days? Or what are your thoughts there? Has that evolved?
Michael Hyatt: I think it’s faster to get to the end result.
Brad Johnson: So, it’s just better data that it can just pick up directly that much faster.
Michael Hyatt: And for example, in Prompt Genie, I’ve trained that on all the engineering documentation that OpenAI and Anthropic have published.
Brad Johnson: So, it’s like you’ve got the blueprints behind the scenes.
Michael Hyatt: Exactly.
Brad Johnson: It’s like that scene out of The Matrix where they plug the thing in the back of his head. It’s like, “I know kung fu.” So, like, basically, that’s the main line.
Michael Hyatt: That’s right.
Brad Johnson: To get there quicker. Okay. Otherwise, you might have to hack your way around till it figures it out.
Michael Hyatt: You might, and you’re right, though, AI does a really good job. There’s something called reverse prompting, where people just state the goal basically and let the AI figure it out. And sometimes, if I’m not clear what the output I want is, if I’m clear on the output, I’m going to prompt. But if I’m not clear on the output, I just want to see what it’s going to come up with because sometimes it’s beyond what I would’ve imagined. I created an assessment recently that, in fact, it goes with this exact framework. In fact, we should offer this to your audience. It’s a freebie.
Brad Johnson: I think that would be really cool. I mean, first off, there’s probably a lot of people trying to figure out where am I at on this ascension ladder that I haven’t figured out where it’s going all the way yet, but…
Michael Hyatt: This asks them like 25 questions, and not only does it give them a level, but then it gives them an action plan for how to move to the next level. Very cool. But when I was building that assessment, I thought, “I wonder how it’s going to build it.” So, I told it what I wanted it to accomplish, and I said, “Just surprise me.” It was killer.
Brad Johnson: So, let me get this right. You’re like, “I need to create a model that helps me figure out the different levels of AI,” and you used AI to build the model.
Michael Hyatt: Exactly.
Brad Johnson: Interesting. Well, you’re practicing what you preach. Okay, so consultant, got it. Any closing thoughts on that before we get to number three?
Michael Hyatt: No, you have to go through that. You absolutely have to go through that. So, you’re going to have to learn more structured programming or prompting.
Brad Johnson: So, cannot skip to three. Must pass through all to, yeah.
Michael Hyatt: Well, I was, like, trying to skip from two all the way up to four to the agent level and that doesn’t work because there’s some other things you have to do before you get there.
Brad Johnson: Got to acquire the skills along the way. Okay.
Michael Hyatt: That brings us to level three, which I would call the confidant level, and this is where AI is a trusted insider. It has persistent knowledge of you, your work, your preferences, and your world. And so, it no longer needs to be briefed from scratch because it’s briefing itself before it gets into every conversation with you. So, if level two is kind of like a brilliant advisor that has dementia, this is one that retains that context across conversations and it really involves creating some very foundational documents.
For example, I have a document that’s all my personal preferences, all the quirks about me, my family members, my Enneagram number, all the different assessment results, my biography, my history, the books I’ve written. I just basically pour anything into it, like if I was onboarding, Brad, an executive assistant. I would say, “You got to know this about me.” And I would just take the time to walk through that, because I know every conversation with that person is going to be better from that point forward. So, that if Mindy, one of my daughters, sends an email, my executive assistant then says, “Hey, some random person just sent this to you.” They know.
So, another document I have is on my company, a comprehensive document about my company where it’s got our mission, our values, our vision, our origin story, our product offerings, our pricing, literally everything. Mine’s about a 35-page document, which by the way, I used AI to build. So, I fed it the raw material, and then it built that document.
Another one, and this is critical, is what I would call the ideal customer avatar. I wanted to know intimately the people I’m serving and I wanted to know even, like my avatar’s name is Jordan Mitchell. And that describes the person I serve in the AI Business Lab. And so, any piece of content that arrives from a marketing perspective, it’s appealing to Jordan. And then the final document and perhaps the document that will give you the best results in terms of writing or content creation is a writing style guide that’s been trained on your writing, your speaking, so that it sounds like you.
Brad Johnson: And so, it’s funny, as you were going into that, you were like confidant and you started to explain context of knowing you, your work, your preferences. I was like, that sounds like when I first learned of how you onboarded your executive assistant to Jim today, which is exactly what I used to onboard Brooke. And it was everything from here’s my airlines I prefer to fly on, here’s my frequent flyer numbers. Do I prefer a hotel room by the elevator or as far away from the elevator as possible? How do I like my coffee? If I’m in a crunch during lunch and I’ve got a meeting, like what would I order?
And so, essentially, you applied what you’ve done in real life to automation so that now there’s context. One other thing, and I’m just going to throw this out there because I want to get your take. There’s a high level of trust with your EA. I know you trust Jim with very confidential information.
Michael Hyatt: I do.
Brad Johnson: Same with Brooke on my front. And we had to kind of grow into that.
Michael Hyatt: Sure. We did too.
Brad Johnson: And it’s part of the interview process. Now, that’s outsourced to tech, some people might be out there thinking like, “That’s a lot of information to feed technology that, what if somebody hacks it? What if they’ve got certain information you don’t want out there?” Like, what are your thoughts? How would you rebuttal that to the doubter out there?
Michael Hyatt: Well, I would say, first of all, don’t do anything you’re not comfortable doing. I wrote a book in 2001, believe it or not, called Invasion of Privacy, and I was the champion of privacy, and I talked about how to retain your privacy. Now, I’ve largely given up on that. I think that one of the things we’ve learned over the last 10 years is that any system can be hacked. And so, I try to live my life in a way that if my secrets were exposed, I’d be okay with it. And that doesn’t mean I’m going to fork over my bank information, having said that, but like I do, not exactly therapy sessions, but I have an AI coach. And I tell it some of my deepest stuff.
Now, having said that, I have my privacy settings set so that AI is not going to be trained on this information. Could it be hacked? Yes. There’s nothing going to be in there that’s salacious enough that’s going to really warrant a lot of interest.
Brad Johnson: Yeah. Well, I also feel like in today’s world, 2026, you walk outside of this hotel, you’re probably on 50 different cameras, public cameras that are recording every move you make. If you use your cell phone and Google something, it’s being recorded. That data is captured. And now, the next ad is served based on your search, right?
So, I feel like, I wouldn’t say I’ve given up on it, but it’s kind of like the reality of the world we’re living in. Now, with that being said, double disclaimer here, you’re in finance, if you’re listening to this, most likely, or watching this. How you treat your own information is very different than how you treat your clients’ information. And so, I think what’s cool is you’re guinea pigging on you and whatever you’re comfortable with, and obviously, there’s different levels of security if it was somebody else’s information.
So, I’m saying that only because I’m doubling down like three, four times on this. We are on the leading edge of this, which means there’s going to be experimentation, there’s going to be bugs or flaws in new AI models. So, don’t ever put your clients in a spot. You would risk their information.
Michael Hyatt: That’s right.
Brad Johnson: It’s okay if you risk your own. It’s up to you.
Michael Hyatt: Well, and because of that trust, if you were to say to them, “Hey, I’m just going to upload your brokerage statements here to Claude and do some analysis of them. I know it’s got your name all over them, but trust me,” they will trust you. And that’s a fiduciary responsibility that you have to the client to look out for them. And like you said, there’s a distinction between doing it for yourself and doing it for somebody else. But you think about how many times we fork over information. I mean, every time we hand a credit card to a server and they disappear with it, you have no idea what’s happening to it.
Brad Johnson: All of it. All the credit card companies know exactly what you’re purchasing. That’s part of their data.
Michael Hyatt: Exactly.
Brad Johnson: On the other side. So, one last thought on three, because you talked about, “Hey, all of these documents.” And what’s really cool, this is a joke I’ve used, so I’m going to use it again. I’m like the best-selling book in the world ever? We’ll see if you know this one. It wasn’t actually written by the author. You know this one?
Michael Hyatt: The Bible.
Brad Johnson: The Bible. Jesus just told stories and somebody else wrote them down. And the cool thing is all of these documents you just talked about, it’s not like you have to sit there and hack them out. Just let AI interview you.
Michael Hyatt: Yes.
Brad Johnson: Like I’ve seen you do that so many times, it’s just like, “Here’s my end goal of the thing I’m trying to build. Will you just ask me questions so you get all the context needed until you have enough information to build?”
Michael Hyatt: I was doing it this morning. I was trying to write a story for a webinar, and I said, “Here’s kind of what happened, but ask me any stories or ask me for any information that you need to do the best possible version of this.”
Brad Johnson: And so it just keeps– so basically, you have now allowed AI continue to question me until you have all the context needed to get exactly. So, now you’re basically essentially guaranteeing, like it’s going to have enough and it’s going to keep asking. Does it ever ask for too much context? Like, where you’re like, “That’s enough,” or do you just keep answering?
Michael Hyatt: No, I usually just keep asking because it can do a lot with a little. And usually, when it says it’s got what it needs, usually does.
Brad Johnson: And it probably comes up with things you never even thought of.
Michael Hyatt: Oh, 100%.
Brad Johnson: Yeah. Okay, we’re running short on time.
Michael Hyatt: I know.
Brad Johnson: You said number four is maybe an agent. Did you allude to that?
Michael Hyatt: That’s agent, and this is where AI is a semi-autonomous actor.
Brad Johnson: So, define agent in AI terms, because you might be thinking like FBI agent, secret agent. Like, how do you define agent?
Michael Hyatt: Well, if you’re not operating at the agent level, you are the agent.
Brad Johnson: Got it.
Michael Hyatt: Okay? So, you’re the one that’s interacting with AI. You’re the one that’s prompting it. You’re the one that’s asking for context or asking for help. When you turn over something to a semi-autonomous agent, and I’m going to give you kind of a– there’s some nuance inside of this particular level. But you’re basically, it’s going to run on its own at predetermined times without your involvement. I have 25 agents now, and they run quietly, mostly through the middle of the night. I have a couple that check in once an hour and do very specific tasks, but they’re running semi-autonomously.
Brad Johnson: How do they check in with you?
Michael Hyatt: They’re usually producing output.
Brad Johnson: So, like an email prompt or…
Michael Hyatt: Well, just one I have, and probably financial advisors can’t do this because it would involve client data, but I have it triage my email twice a day. And what that means is it goes in and categorizes it based on 1 of 10 different tags, like newsletters, receipts, shipping information. There’s a ton of stuff you get that you actually don’t need to be looking at right now.
But anything, I have it judge on whether something needs a reply, and then it drafts a reply and puts it in my drafts folder. I don’t let it send before I’ve approved it. But then I go in, and honestly, I’m at the point now where I’ve trained it enough that it’s very rare for me to correct something. I usually send it.
Brad Johnson: So, it’s trained on your voice, so it writes like you. Really key point I want to hit, year-old 10-80-10 model that we’ve used a ton inputs 80% middle work, last 10% quality control. You have not given away the quality control. This is a representation of Michael the human, so you’ve kind of automated that front 90, but that last 10%, probably last 2% quality control. Yep, click send.
Michael Hyatt: Well, the first 10% was creating the agent. And I’ll tell you the place for everybody to start, and again, if you’re not at this level, don’t worry about it, but once you get to this level, is that you create a scheduled task inside of Claude Cowork. Now, there may be something fancy later that we’re not talking about, but a lot of AIs have scheduled tasks. But it requires usually that you’ve distilled a standard operating procedure down into what is called a skill, and Claude made this famous, but now almost every AI has this. But it’s an SOP, a standard protocol for dealing with specific situations.
So, what I do is I iterate with Claude until I’ve got that protocol nailed. So, it’s like, okay, you do this first, second, third, fourth, fifth, so forth. And then I say, “Claude, turn that into a skill,” and so it writes a skill, so the next time I want to invoke that skill, I don’t have to recreate the wheel. Like, here’s a good example. When I get output, I want my documents to be in Google Docs. Now, this could also apply to Microsoft Word.
But I always like it to be formatted in a very specific, predictable, kind of boring way. And so, I’ve created that protocol so that when I say turn it into a document, bam, it’s exactly how I want it. It’s using the fonts that I want. It’s using the headers that I want. It’s using the footers that I want. And it’s just work I don’t have to do.
Brad Johnson: So, you could take– so this skill, here’s a raw text file. Feed it in there. It’s kicking out a Google Doc formatted the way you want it. So, it’s cleaned it. It’s basically put your coat of paint on it that you prefer.
Michael Hyatt: Now, most people, what they do is they’re going to load the skill into the LLM or have the– like, when you ask Claude to create a skill, it’ll create the skill, and when it’s done, it will say, “Do you want me to add this to your skills library?” And so, you add it. And in the agent mode also, you can use connectors to the outside world so that Claude can interact with other applications.
And I’ll give you an example. Yesterday, I was trying to retrieve a text thread that was inside of iMessages. And I don’t know about you, but I have a lot of text messages, and I hate scrolling back through in those trying to find a report or a document that somebody gave me.
Brad Johnson: And the search function’s a little clunky in that.
Michael Hyatt: It is very clunky.
Brad Johnson: Yeah.
Michael Hyatt: And so, amazingly, Claude has a connector. It’s called send and receive iMessages. And one click, you’ve given that capability to Claude, and now I just said to it, I said, “Okay, find this text thread that I had with my friend Brian, and give me the last four days of messages.” And then I said to it, “Okay, find me that document that he sent me, and I just don’t have the time to scroll through it,” and it spit it right out.
Brad Johnson: Created a Google search engine for your text threads.
Michael Hyatt: That’s right. And it makes it infinitely better.
Brad Johnson: So, what advice would you give– I take the advice that you’ve got to ascend through the model, so you can’t shortcut it. But let’s say I just arrived at the agent level, final disclaimer I’ll make on this one, in finance as an advisor, like, this is leading edge stuff, so PII, like pretty much a no-no at this point, like, as we stand as we record this, right? But once again, this is the worst it will ever be. So, start to figure out the chops because it’s coming.
But let’s say I just stepped into agent world. What frustration should I be ready for? And what’s on the other side of overcoming that frustration? You’ve shared a couple really cool examples, but this is like, “Dude, this has changed my life. I’ve got like 25 ‘employees’ now doing things for me.”
Michael Hyatt: Yes. And I think…
Brad Johnson: Is that what you should think of in your head? This is, “I have now manufactured an employee.”
Michael Hyatt: Essentially, yes. It’s a very specific kind of single purpose employee. There’s an advanced level of agents that’s more than that, and that’s where we start talking about OpenClaw and Hermes and other tools like that. But Cowork-scheduled tasks are semi-autonomous agents. And so, I think the first thing I would do is if I was going to build my very first agent, the thing I would do is I would say, “What in my business right now or what in my practice as an advisor is really annoying, but it has to be done and it has to be done routinely?” And we already have identified a protocol, an SOP that we use with our humans that do this work.
Brad Johnson: I think what’s interesting, and I may be venturing into dangerous territory, but I know a big frustration in finance right now, obviously, you’ve got your CRM, so we’d have to make sure this wasn’t impacting the PII, but it’s the actual workflows and building those inside the CRM. So, it’s creating the assembly line for here’s where the process needs to start, here’s the end I have in mind. And I think we have already started to use AI to start to build the assembly lines without touching the data to where now you can start to manufacture them, where in the past it would be a very clunky, manual process that was just made you want to, like, scratch your eyeballs out.
Michael Hyatt: Well, here’s a good example, and this was really the first agent that I built was a morning briefing. And I’ve always admired the fact that the President of the United States gets a morning briefing, and his staff really works on that hard. And one of my friends, one of my clients is Amy Porterfield. She used to work for Tony Robbins, and Tony required that of her, and she would spend hours every day preparing a brief so that he was up to speed on world news, on things related to the business, but it was just like a ritual.
So, I created one, and it’s got about 10 different components, but one of my favorite ones is that it takes my calendar. And it serves this to me, by the way, as an HTML file, so I just open it in my browser. I get a link in Slack, I click on it, I open it in my browser, and it has my daily schedule. But for everybody that I’m meeting with that’s not a team member, it gives me a comprehensive dossier on that person.
Now, imagine if you’re an advisor. You’re going into an intake meeting. And by the way, I met with an estate planner, an attorney about a month ago, and my AI had prepared a dossier on him, and I knew everything about him. I knew his kids’ names. I knew where he went to college. I knew every move in his career because there are giant databases out there that track this, like ZoomInfo, like Apollo, and I used the API connectors to get that information. Actually, Claude has connectors to both of those. He didn’t know jack about me. So, I spent $700 paying him a fee for an hour so he could learn about me. And I thought, “If he had the AI, our conversation would’ve started at a completely different level.”
So, I do that today when somebody applies to be part of the mastermind, then I got the AI going out and doing research on them based on their application, so that before my sales team ever has a conversation with that person, they know everything about that person. In fact, it has conversation starters in the dossier based on the overlap of my world and their world.
Brad Johnson: Well, you think of the value creation there, back to abundance versus scarcity. That attorney is a good example, very similar industry to a financial advisor. Pay for the time and the knowledge, right? And if you have somebody that’s on the leading edge with that, now you can use your time more efficiently because most first meetings 100% are, “Let me understand your situation. Where do you have issues, concerns? Where are the gaps in the plan?” And that’s always going to be, some form of that’s going to be required, but the quicker you can get up to speed on somebody’s situation, now you can use more time back to freedom to actually help you, not get your backstory.
Michael Hyatt: Right. And like this tax attorney, estate planner, he thought I was a genius. Like, “How do you know this stuff about me?” It was a little spooky.
Brad Johnson: You were like the dude off of the Dos Equis commercial, like the most interesting guy in the world. He just knew.
Michael Hyatt: So, I think it, it makes the advisor, it gives them a superpower too. Now, by the way, this is information that they would be accessing that would not be coming from the client, right? That’s PII. But this is stuff that’s publicly available, and you would be astonished at how much stuff is out there.
Brad Johnson: Back to Google, like these companies have been tracking all of our data and a lot of people just don’t know it’s out there.
Michael Hyatt: And they’ve sold that data and people have aggregated it, and it’s amazing what’s out there.
Brad Johnson: Yeah. Okay, I know we’re getting towards the end, so kind of disclaimer here. Agents starting to get into the leading edge where you’ve got to be really, really careful and cautious. And I love the fact, like, you’re your own guinea pig.
Michael Hyatt: I’m my own crash test dummy.
Brad Johnson: You’re like the Tim Ferriss of AI. You’re like your own human guinea pig for AI. So, agents, what’s five and what’s six?
Michael Hyatt: Okay, six is freedom. That’s where we’re headed, right? So, six is what I call the orchestrator level. This is AI as a coordinator of agents. So, this is AI coordinating multiple agents, hiring agents, deploying agents, offloading agents when they’ve done their job, but basically, running your entire organization. I was at AI Business World a month ago. I had a workshop that I conducted there. And there was a speaker by the name of Nici Sweaney. She lives in Australia. She has a PhD in data science. Now, get this, Brad. She has a multimillion-dollar business. She didn’t say exactly how much. She has four kids. She has three humans, 25 agents, and she works 25 hours a week.
Brad Johnson: Running this multimillion-dollar business?
Michael Hyatt: Yes.
Brad Johnson: Did this multimillion-dollar business exist before AI, or she built it solely with AI?
Michael Hyatt: I don’t know the answer to that question. But there are a lot of people that have built those kind of organizations from scratch, for sure. So, at the organizational level, we’re already beginning to see some tools now like Paperclip, Paperclip.ing, that essentially throws up an org chart on your screen of your company full of agents. And you can see them working. There’s a Kanban board where you hire the CEO and you give the CEO a goal, and they spin up the agents. They say, “Okay.”
Brad Johnson: So, the agent is the CEO?
Michael Hyatt: Yes.
Brad Johnson: With your instruction?
Michael Hyatt: With your instruction.
Brad Johnson: So, you’ve basically gone straight to board member.
Michael Hyatt: Yeah, you’re the chairman of the board, and your focus is on vision, goals, and governance. And so, you’re telling the CEO, you’re saying, “Look, here’s my goal for this company.” And it’ll say to you, for example, like, “Well, the first thing I need to do is hire an engineer.” And you could build this in if you want to give permission to hire the engineer, because those are going to burn AI tokens. But you can set budgets for every agent, for every position in your company, essentially. And when it gets to that point, it’ll pause and ping you so that you can say, “No, go ahead and spend the tokens.”
Brad Johnson: So, it’s like a real business.
Michael Hyatt: It’s like a real business.
Brad Johnson: Oh, we can’t hire anybody. We’re out of cash flow.
Michael Hyatt: Right.
Brad Johnson: So, we’re out of tokens.
Michael Hyatt: We’re out of tokens.
Brad Johnson: Which obviously, you have to purchase. Just for context, this will continue to evolve. It should get cheaper as compute gets cheaper, you would think.
Michael Hyatt: Yes, it will. It’s going to get expensive in the near term.
Brad Johnson: And it’s because there’s so much demand, right?
Michael Hyatt: There’s so much demand, and they artificially kept the price low in order to train the AI platforms. But largely, that’s been harvested, so now they’re beginning to inch up the cost because they’re more in harvest mode now. They don’t need your data as much.
Brad Johnson: So, for context, whatever you can share, you said you have about 25 agents. Like, if you were an advisor, what sort of budget should you have in your mind if you were to start to build something like this out? Maybe it’s not 25 agents, but maybe it was like, “Hey, we’ve got a couple five agents, like, running tasks for me.”
Michael Hyatt: Well, first of all, I think you’re going to have to get on the max plans. Like with Anthropic, I pay $200 a month for their max plan. And I very rarely bump up against the borders. ChatGPT has a similar kind of thing. But you’re going to pay hundreds of dollars to maybe a few thousand dollars. But what’s the cost of hiring a human or deploying a human to do those same things?
Like, my morning brief, if I were to ask Jim, my trusty assistant, to go out and compile a dossier on every person I was going to be meeting with the next day, that’s literally a full-time job. Now, you could argue, well, I probably wouldn’t have him do it because it’s so time intensive. But wow, the benefit to me is huge.
Brad Johnson: Yeah. Well, my man, speaking of AI, we haven’t had mastermind to actually be a part of. I was thinking this was going to be shorter, but here we are an hour later.
Michael Hyatt: I’m sorry.
Brad Johnson: No, I love it. I think we could spend– well, we will spend another day and a half talking AI. So, thank you as always. Always thanks very much in our conversations and thanks for carving out the time to do this.
Michael Hyatt: My pleasure. No, I’d just say the sooner you get engaged, the sooner you’re going to start realizing the benefits, and there’s going to be some bumps along the way. But again, I go back to mindset. Discipline number one, be curious. Discipline number two, be persistent. And discipline number three, be resilient when you encounter failures, because you will, especially in these early days.
Brad Johnson: At the Johnson house, we say get uncomfortable and do hard things. That’s where the growth comes from.
Michael Hyatt: There you go.
Brad Johnson: So, relates to AI as well. All right, my man. Appreciate it.
