← Using AI at Work: AI in the Workplace & Generative AI for Business Leaders
Using AI at Work: AI in the Workplace & Generative AI for Business Leaders
92: Using AI for Smarter Marketing: Synthetic Audiences, OpenClaw & AI Agents with Justin Brooke
00:51:58
Loading summary
Transcript70 lines
- [00:00]
A
I was able to double my money every month for 11 months in a row, literally turning $60 into six figures.
- [00:06]
B
Let's start with maybe the genesis of like where you came up with the idea and kind of what the tech stack has looked like to help you get these incredible results.
- [00:14]
A
Generally Marketing 101, you know, as you come up with this customer, deploy against it, make it better, deploy against it, make it better. And then finally you have this like really dialed in customer Persona and your marketing, your message to market match is now winning.
- [00:28]
B
How do they develop that Persona, that icp, that avatar?
- [00:31]
A
The one thing to careful of is in order for it to be accurate, you can't just give it a one sentence, pretend to be this person and then it's going to magically know all their fears and desires and life choices and empathy and all those things. Right? And so when we're creating a Persona, it's 1200 words I believe. 2027, 2028. This is step one of marketing.
- [00:55]
B
Justin Brooke is a veteran digital advertiser and the founder of adskills.com and Agentskills AI. And he teaches marketers how to use AI systems to test, score and sharpen ad comp so businesses can generate more leads from PPC with less guesswork. Welcome to Using AI at Work. I'm your host Chris Staigle. Each week we'll be learning how today's business owners, entrepreneurs and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started. Right now every business leader is asking the same question. What are we going to do about AI? If this is you, chiefaiofficer.com has the answer. We give you a simple path forward where we provide executive and team training so your people know exactly how to safely use generative AI in their day to day. We also manage the deployment and implementation to make sure tools actually get adopted and deliver results. And we'll also guide company wide transformation so AI becomes part of your operating system, not just another shiny object. The companies that act now will increase productivity, cut costs and grow faster than their competitors. Those that wait will get left behind. So if you want to make AI work in your business, visit chiefai officer.com and see how we're helping companies of all sizes finally get results from AI. Hey everybody. Welcome to another fantastic episode of Using AI at Work. My name is Chris Daigle. I'm the host of Using AI at Work and our guest today is Mr. Justin Brook. And Justin and I have known each other from I guess the Internet space for probably going on a couple of decades now. But we recently reconnected when both of us were invited to join 30 AI startup founders for a weekend with podcast hosts. And AI entrepreneur extraordinaire Greg Eisenberg and his team had the chance to just really sit around in a huge mansion off of Las Olas and Fort Lauderdale and ask each other, like, what are you doing in the space? Where are things going? What's cool? What's got your attention? And they didn't have many outside speakers, and Justin was one of those speakers that they brought up and we're going to cover what he talked about here. But Justin, in the spirit of Greg's podcast, why don't we start by letting me ask you, what do you want the listeners to walk away from this episode with today?
- [03:14]
A
I would love for them to walk away from this episode with the idea that virtual focus groups, synthetic audiences are real, are accurate and maybe even with the idea that it is going to be a first step in your marketing in the future.
- [03:32]
B
Love it. I had a chance to see this demonstrated real time in Fort Lauderdale, and it's pretty impressive stuff. So, Justin, before we jump in, I think for those of you who aren't familiar with our guest today, let's give Justin a minute to kind of share some of his bona fides and the successes he's experienced in his career.
- [03:48]
A
Sure. Yeah. Yeah. So classically a marketer, advertiser. Always wanted to be that as a kid, grew as I grew up, got older. I was working in construction and fast food and restaurants and stuff like that and didn't love that, that world, God bless them all, we all got to eat. I did not love that world. And so I had this dream of having a website that paid my bills. That was the dream. And so I started getting involved in there's a local organization called SCORE that helps business owners and stuff. And then I started seeing some Internet marketer type guys online, Russell Brun. And I got involved in that. I became his. His intern. And my job as an intern was to study this giant walk in closet of all these courses and seminar footage and cassettes and books. And my job was to review this stuff and literally to write reviews so he could make referral commissions back on these things. So I got the education of a lifetime studying the greatest marketers of all time for all the things I could have learned in that closet. I mean, Joe Polish is in there. Chet Holmes, Dan Kennedy, like the greatest of the greats are in there. Of all the things there was A little ad. There was a little course in there about Google Ads. And it changed my whole paradigm of what I thought advertising was. You know, growing up, always wanted to be an advertiser, but it was magazines, it was radios, it was tv, right? All of a sudden I realized, wait a minute, I can be, I can reach the world in my bedroom for $5 a day and I don't have to ask permission, I don't have to dress up in suits, I don't have to go get a degree or anything like that. And so that kind of captured my attention, you know, pathetically. And also famously in my presentations all over the world, I've talked about how I started with a really pathetic $2 a day budget. I have no money, right? I worked in fast food, you know, so I scraped together 60 bucks from paying only half my electric bill, and then I take that $60, put it on a $2 a day budget. Because of that world class education I got interning with Russell, I was able to double my money every month for 11 months in a row, literally turning $60 into six figures. And advertising captured me after that. Right. It was the way that I paid all my bills. I had a seven figure agency, traveled the world, spoke all over the world, podcast. And so I'm an advertiser by trade. But algorithms, machine learning have come into this space, which took me into AI. And then I kind of realized, wait, hold on a second, I see how AI can replace these ad agencies. And so that's kind of what's brought me into the AI world, is kind of realizing like how we could be using these things in marketing. Like I'm not a coder, but I'm realizing how I can use Claude and Manus and openclaw and all these things for Google Ads, for SEO, for social media, you know, so that's, that's who I am.
- [06:49]
B
So fascinating story. I didn't realize that that sort of the best paradigm in the world is to earn while you learn. Right? And it sounds like that was the situation that you found yourself in with certainly a master. Now one of the things, I mean, you consider yourself like an advertiser marketer and that sort of thing, but based on what I saw in Fort Lauderdale, like you're an AI thought leader, whether or not you want to admit it or accept that title.
- [07:16]
A
Thank you.
- [07:17]
B
I appreciate it. The stuff, the content that you're putting out, I primarily follow your, your posts on X. It's way more than marketing tips or hacks or marketing theory. It is a real Exploration of generative AI. And it's. It's like transformation and involvement in what it means to be like an entrepreneurial knowledge worker. Right. So I'm excited about this conversation now. So let's start with the original takeaway that you wanted, which was this synthetic audience concept. And my first exposure to this, and I shared the link with you, I'm in Fort Lauderdale, was with Colgate Palmolive, had worked with an AI lab in Estonia to see if it was possible to synthesize accurate audience reaction to a new product launch. And what they came out with was some big statistical report with all kinds of charts and math and that sort of thing. And their. Their opinion was, yeah, but as far as, like, understanding it and applying it, it wasn't delivered in a format. To me, though, as exciting as it was, that I could do anything with it. It wasn't until I saw not just that you were doing something with it, but how you did something with it. So let's, let's start with maybe the genesis of, like, where you came up with the idea, first of all, what it means. Yeah, where you came up with the idea and kind of what the tech stack has looked like to help you get these incredible results.
- [08:46]
A
Right? Yeah. So as a marketer, advertiser, we always think in Personas, I think the new cool word is ICP. But these Personas, these avatars, these ICPs, you know, whatever it is, you know, we have this, like, I don't want to say it's imaginary, but we have this target that we create, you know, this target customer. And we try to describe that target customer as best we can. And then we try to deploy against that idea of our customer, and we hope that our marketing works. Right? I mean, that's generally marketing 101, you know, as you come up with this customer, deploy against it, make it better. Deploy against it, make it better. And then finally you have this, like, really dialed in customer Persona. And your marketing, your message to market match is now winning. And when I started playing around with AI and kind of one of the classic prompts is, you know, pretend you're a copywriter or pretend you're a senior developer or whatever, I said, you know, I said, hold on a second. That sounds a little bit like a Persona to me. Like if this thing can pretend. And as I learned that, how the LLMs were trained and fine tuning and all these things, I learned that that's kind of what they do at the nucleus of what an LLM is, is it's kind of A giant pretender. It's pretending to think. It's pretending to do a role. So I said, okay, well, then it can pretend to be my customer, you know. And so I strung together a simple little automation. I don't. I wouldn't even call it an agent because I didn't know about RAG yet. I didn't know about all these, you know, recursive thinking or anything. So I just, you know, I was like, all right, step one, I give you a link or some ad copy, some marketing copy. Step two, pretend to be my Persona. Give me feedback about that. Step three, be a copywriter. Take the feedback. Rewrite me a new version based on the feedback. That's better. So I did that. That was the very first version. It was like, little four. I think you saw the one. I've got, like, 12 different things in there now. I'm constantly improving this thing. But that was the very beginning. Like, I just wanted it to, like, here's my copy. Give me feedback. Write me variations based on the feedback. And that was wild. That was great. I didn't always use it. I used it on some ads, and it immediately gave me back better stuff. And being trained in advertising for 20 years, I could tell. I was like, oh, this is better. Like, this is definitely better than what I gave it. So then I started deploying against it, and I got better results. And then people are like, oh, well, that's just you. And you've, you know, you've been doing this for 20 years, and, you know, it's not going to work for me and all that stuff. And so I was like, all right, well, I got to get some case studies and some testimonials and stuff. Use this for other people in the meantime of trying. Because, like, when I'm showing this, there's not yet Harvard studies and Colgate stuff, and I don't even know what I'm creating or what I'm stepping into. Like, I'm just a marketer trying to create a, you know, a pretend customer. And so it's like I'm trying to show people magic for the first time. And it was like a very hard sell. It's like, listen, I've got millions of dollars riding on this campaign. I can't just play with your new little toy, right? So anyways, I start playing with it with my social media. I said, guys, look, it's not only improving my ads. I'm getting better engagement on my LinkedIn posts. I'm getting better engagement on my Twitter posts because it's just Copy, right? That's all. A LinkedIn or a Twitter, an ad. It's all just copy. So finally, I get a couple of people involved because I do, like, a whole demonstration. I did, like, a presentation to my customers. I get a few people involved. In that time, I meet another guy. He was like an Internet lawyer. He was way smarter. He brought a bunch of math. He said, justin, you know what? I'm using the same thing that you're talking about, but I'm using it with a bunch of pretend. I hate to use the word pretend audiences. You know, I'm speaking historical. These are not pretend anymore. Harvard, Stanford, you know. So he says, hey, look, I'm passing it through. Instead of just one Persona, I'm passing it through dozens to hundreds of Personas, getting way more accurate feedback. And it's much more. It exemplifies the market. And in that meeting that we were at, I showed them. I said, look, guys, we're. Look, there's men and women here. There's young and old here. There's, you know, pharmacists here. There's coaches here. There's, you know, there's a wide variety of people in our audiences. So he showed me the math, and he showed me the way. And so I started adopting that and I started calling it virtual focus groups, because now we got a group of people. So essentially, let me kind of catch everybody back up. Now we're taking our ad copy, our content, whatever it is that you use. Instead of passing it through one Persona, we're passing it through a group of people. I called it a virtual focus group, getting way better feedback, passing it on to the copywriter to get better. And then I added a. I was like, well, I don't know which of these variations to use, you know, and testing is expensive. And so I said, well, why can't I just add another step that's a data analyst that can go and give me a score, you know, just help me score these three variations. So we do that. And now I'm like, holy crap. This thing is actually. It's. It's picking winners. You know, I can't say every time there's legal implications. You know, I'm not. I'm not predicting lottery numbers or anything right here, you know, but I add this data analyst step, who is now scoring the variations that my copywriter wrote based on the feedback back of the original. Right? So. And then I'm realizing as I'm posting it on Twitter, LinkedIn, my ads and things like, that this data analyst is picking Winners quite often, you know, eight out of ten times or better, to the point where I'm just like, I just trust it more. I was like, I'm just gonna go with it. It's right more often than it's wrong. So then I get a couple customers to use it. One guy, he gets crazy runaway results with it. And maybe his early stuff was just garbage and now he was finally getting some good stuff, you know, in there. But he had like crazy 16x ROAS, 19x ROAS, crazy numbers, you know, way outliers. I'm not saying that's a usual result. We test more people, we start seeing Forex 6X. Like it's coming back often that people are running this, running their ads through these virtual focus groups, getting variations, scoring them, just picking the one that had the highest score, run that one. And now all of a sudden, their ads are becoming way more profitable than anything they were. And so the way that I've been like, selling this because people don't know it's so new, I just say, hey, give me an ad. I'll run it through my, you know, my magic. Yeah, my fairies, or whatever you think it is. I'll run it through my magic, try it out. Try it out for free. Right? And I know that it's going to work better because it's going through processes
- [15:47]
B
that I've codified to do on, like, in reality.
- [15:50]
A
Yeah. I mean, I've got 20 years of experience, like you do, you know, and so like, they're going through processes and, and feedback and copywriters and data analyst scoring and stuff. So of course it's going to work better than their stuff. And so normally what happens is they don't. They're skeptical. I let them try one for free. They run it. They're like, holy crap, that was great. Can I do more? And it's like, well, yeah, you can pay me for this thing and we can run it. And so then they run it through their whole account. And so that's kind of what I stumbled into. Started talking about online, trying to get people like, hey, man, this thing works. Then all of a sudden, Harvard does a study and they're finding out that it's working. Colgate does a study, they're finding out it's working. Stanford does a study. Lavazza, New York Times. I'm like, hey, we got something here. You know, like, we. We've really got something. I'm not crazy. And so the world is calling it synthetic audiences now. And so I'm just going with the Name. I think virtual focus groups is better. But hey, whatever they want to use,
- [16:48]
B
you know, use your own language, honestly, because what you're doing, what you're doing is different than what those other folks are doing. So for everybody listening, I want to just give you some perspective. So prior to this, if Justin wanted to get that level of certainty on an ad before they invested hundreds of thousands or millions of dollars behind was either, well, my gut's telling me this from 20 years of experience or I want to run the ad. But before I do that, let me find the people, let me get their feedback. And the likelihood of that happening, the mechanics involved, the juggling of the calendars, that was just maybe with one or two ads, but not with every single advertisement that you're going to run.
- [17:27]
A
It was a long process or it was an expensive process and then unscalable in the nor.
- [17:33]
B
Yeah. So now. And what you just. The aha for me was, you're absolutely right. And we were there in Fort Lauderdale when I saw this. There was a guy who specialized in whole funnels, right. Like sales funnels. And immediately I was like, oh my gosh, that Justin's talking about an ad. But we could use this on an entire sales funnel, on the copy, on our, our thought leadership content, on, you know, what, whatever that thing was. And I, it kind of really opened the door for me because my exposure had been, oh, they're testing it on a new product at Colgate. Well, that's got nothing to do with me. It's interesting, but you've now made this accessible to any small, medium large business that plans on investing their capital into advertising. Really on any platform at all.
- [18:18]
A
Yeah, yeah. I mean, I, like I said, I, my argument is, I believe 2027, 2028, this is step one of marketing. Like you will, yeah. Write something or you come up with a new logo or product packaging, whatever it is before you go deploy it against your, your in house list, your social media or ad spend where you put in real dollars, you're going to run it through your little virtual focus group, synthetic audience, whatever, see if it. Even if you even get a good feedback. Right. And so, and now, because it's like, I mean it cost me like 19 cents now to run this stuff with all the different models and whatever. And so now you can run everything, every logo, every social post, every sales page, every, everything can get feedback first. That way, you know, you're not mailing out duds essentially.
- [19:07]
B
And my mind is spinning. I'm thinking about like we need to do one of these. So one of the questions that I always have when there's like this obvious breakthrough that that AI is contributing to what used to be a very laborious and human involved process is how are the marketing experts reacting to this? Because you think about Mad Men sitting around in the conference room smoking cigarettes and just ideating from the brilliance of the creative mind. But now not so much. What's been the reaction from those who were ideating on the ad creative now that they're seeing that this thing is actually, yeah, that's a good idea. But here's how it gets better.
- [19:51]
A
Well, it's been like, I don't know, three different responses, main responses, and, you know, one response is, you got. Some of the Mad Men are just good sales pitches, you know, and it's like sometimes you watch a commercial and like, wait, how did that commercial get approved? Like, so somebody in that room was like, yeah, that's a good idea. Let's ship that one. So you have some of those groups who, they don't want what I have, because the last thing in the world is they want actual market feedback. Like, they want their sales pitch to win. And so there's those people, you know, but then there's, you know, probably the other almost half of the room. This is such a new and wild idea. And they don't understand how these LLMs work. And so they do think it's pretend. They think it's a toy. They don't, you know, and so they don't know. Right. So it's either they don't want feedback because they're hiding something. Right. Or they don't know that the feedback is real. And then you have a very small percentage of people who are like, yeah, I'm game. I think I get it. Let's try that out. And then they try it all one. And once they get a taste for it, they're like, yeah, this is the future. We have to do this everywhere.
- [20:58]
B
You know, this kind of maps to that, that marketing maxim. It's kind of a joke, but it might have been Ogilvy that said it. But I know that half of my workings, half my advertising is working. I just don't know which half. Yeah, right. Because that's your first.
- [21:10]
A
Yeah. So just like we use tracking. You wouldn't even think of running marketing without tracking these days. That's how I think this is going to be. But on the front end, tracking is like, after you're done, you attach the tracking to make sure you know what it's doing. Well beforehand you want to know that it's good and effective before you mail it and then still track. Right. So I think it's just the beginning of the marketing. Just like tracking. Tracking is kind of the end of the marketing.
- [21:35]
B
So this, I would imagine that this will probably start being baked into certain ad platforms like on Metta or like on Google. Right?
- [21:43]
A
Yeah, I'm hoping I could, I could sell a few before they just eat it up and they just, you know, that's just part of everything that we do, you know.
- [21:51]
B
Hey, yeah, all them today on that. So how do you, did you get started? I mean again you're, you're decades as a direct response. So a marketer who's driven by data. Right. You don't, you don't go by gut very often. The rest, you know, it's like, sounds like a good idea, but let's test it. I'm sure that's your protocol. How do you, how is somebody else that's not have this background but is hearing this and is like, damn, that's pretty interesting. How do they develop that Persona, that icp, that avatar?
- [22:21]
A
Yeah, well, the one thing to be careful of in this space because it is popping off and there's a lot like as you can imagine, you know, the Colgate's, the Lavazza, the New York Times, you know, like big agencies are getting involved, big numbers, you know, five digit, six digit numbers are getting involved in this stuff. The one thing to be careful of is in order for it to be accurate, just the way that you would fine tune any AI, you have to give it a substantial amount of data. You can't just give it a one sentence, pretend to be this person and then it's going to magically know all their fears and desires and life choices and empathy and all those things. Right. And so when we're creating a Persona, it's a thousand words, twelve hundred words. When Stanford and Harvard did this, I think they were doing two hour long interviews with customers and then taking the transcripts of those interviews and putting it in there. Other people have taken substantial survey data. Other people have taken seven years of all your historical Google Ad data. You know, so you need a, a pretty good amount of data in order for it to be accurate. You know, and so sometimes the, the sales pitch is we have 5,000 audience members in our virtual audience.
- [23:42]
B
Yeah.
- [23:42]
A
But you find out that they're actually just like, you know, 5,001 sentence things. That's not going to get you the accurate thing that you need outside of it. So I'd much rather have, I'd rather have somebody have three 1200 word or 4000 word Personas that it's trained on than 500 one sentence word Personas that it's trained on. You're just not going to get the accuracy and you're going to think I'm a liar in all this if you try it that way.
- [24:10]
B
So for somebody who's never really done much thinking about who is our avatar, who does buy our product, what are some of the characteristics that would make up a good ICP definition for.
- [24:22]
A
Yeah, so first you want to start with basic demographics. You know, what is their age, gender, you know, did they go to college or not go to college, what kind of earnings, right? And then you start thinking about their psychographics and this is what do they like maybe what books do they read, what stores do they shop at, what products do they buy? Then you want to start getting into some of the things like what are their challenges? You know, frustrations, challenges, desires, things like that, right? So demographics, psychographics. So like who are they physically, who are they mentally, emotionally, who are they? And then lifestyle, what is their life like? Their challenges, their frustrations, their dreams, their wishes, you know, all that stuff. And then I like to do what's called an empathy map. And this is just a long. I didn't invent this. It's been around for many years. I think colleges have taught it an empathy map is what do they see, what do they hear, what do they think, what do they say, right? And actually get into that, like do real Internet research to find out what are those things and then also what is their pain and what is their gain. And so that's an empathy map exercise. And you can Google them and find lots of different true pages and whiteboard exercises to do this stuff. But because the, because these LLMs have ingested the entire Internet, right? I mean, that's how they're trained. They've already read all of Reddit, all of the YouTube comments, all the Amazon reviews, right? Because they've already read all those things. It's really, really good at telling you demographics, psychographics, lifestyle, empathy map, you know, those type of things. And so those are the elements that we go through to try and get to about 1200-2000 words in our Personas.
- [26:10]
B
So I would imagine that an individual who's never done this before, just like you said, just go to the models and say, hey, I need to create a. I need help developing the demographic of my audience now. I need the help of the psychographic. Now I need the help of or hell, give them the show notes from this and say, you know, or the transcript say, do, do what Justin said and the models can help you get there. And then the end result of that effort. So somebody listening to this might be like, like, do I really need to do that? But the result is 8 out of 10 winners are coming out of your magic.
- [26:41]
A
Yeah, I want to use those numbers loosely for sure, you know, agency's sake. But yeah, if you want it to be accurate, and this is, you know, what Harvard has said and what Stanford has says, if you want these to actually be accurate, if you want to be able to trust the results coming out of it, you have to have pretty good Personas going into it. Right. Otherwise it doesn't know how to accurately pretend how to be your customer.
- [27:07]
B
So I remember loosely that you had referenced the impact that it had on the New York Times and how they had used that.
- [27:14]
A
Do you remember what that they were seeing? 92% accuracy. So what? The New York Times has a long time, they have had this established practice where they would run their headlines, their content, ideas, article ideas, whatever, through a human group of readers. And I used to do this with some of my course courses. It's like beta testers. You know, software has done this for a long time. And so the beta test group gets the ideas first. You see if the beta test group yay or nay on those things and then you run it. Well, they, since this stuff started coming out, they created what they call their digital twin of the beta test group. And so now instead of having to go wait for these people to wake up their schedules to be available, you know, they just run it to this digital twin that's available 247 anytime they want. And you could run multiple teams can be using it at the same time. They're finding that this digital twin audience is 92% as accurate as their human beta testers were. So faster, cheaper on demand.
- [28:24]
B
Yeah, yeah. You mentioned earlier that you're not like a technician, you're not a coding background or anything like that. What was the learning curve like for you to go from, hey, I think this is a good idea, let me try it to. Wow, Now I've got, you know, eight figure, nine figure companies that want me to build these things for them. What was that learning curve on the technology like?
- [28:49]
A
You know, I've been blessed to meet a lot of really smart people in my 20 years. You know, my dad's a mechanic, so I've got good work ethic, you know, and I'm, I'm fairly smart, you know, I'm just shy of Mensa, you know, I'm not, I did, I'm not Mensa, I'm shy of it, you know, but I'm close, I'm close. And so I'm no dummy, but like work ethic and just providing. And so I've met a lot of really smart people and so I kind of had an idea and I just, I believe in sharing the ideas instead of hoarding the ideas. And so as I shared it, you know, this guy, you know, this really smart guy, a guy was a mathematician, he said, here, you should add this to that. Another guy said, hey, your data analyst thing is good, but you should add like a, an algorithm that the data analyst is scoring these things against. And so as I shared it with people, people gave me this idea, that idea, this idea. And I just started collecting them and working with them and, and then testing to see if they actually work. And so, yeah, that's the way that I've kind of stumbled. I don't know if I stumbled because I was very purposeful in doing this, you know, but there's also a lot of blessings in the network that I've built along the way.
- [29:59]
B
We'll just deviate for a second and get geeky. What does the tech stack or the tool stack look like that's helping you develop these and then go through that, that 19 cent cost of, of getting the result?
- [30:13]
A
Yeah, yeah. So I use a tool called Mind Studio. But you could, you could. I've watched people string this together in a, in a chatgpt. You know, they just upload the 12 PDFs every time they want to run something through it. I think that's an incredibly dinosaur archaic way of doing it. But you can kind of hobble something together. You could make a project in Claude that is just dedicated and so Claude has projects and you can put files in there, so you could put all the Personas as files and then you could give the instructions is what you're doing. And now every time you want to run a piece of copy through there, you just post your, your copy, your sales letter, whatever it is to that project so you could string it together that way. I think the best way is to use something like a Mind Studio, which is a, an alternative to N8N or make. And these are tools that allow you to string together automations, which allows me, instead of a project, I can give it a piece. It's going to run through things. It can run in parallel. I can choose the different models, I can change the temperature. I have finer granular level controls over every piece of it. And then I would say the high end, like if you're a coder, you can probably put this together with Crewai or LangChain and you can build a real agent. Yeah, you know, doing this.
- [31:38]
B
So, you know, as we're talking about this, for those of you who are listening and you're like, well, I'm not in the marketing department, I'm in a different department like this. This reminds me of an example that Jeff woods talks about in his book where he helped a client create like a synthetic board that he had to, the CEO was going to have to present to the board. There was going to be some friction because of different positions that the board members had. And they, instead of building the customer avatar, they built the, the digital twin of board member A, B, C, D and allowed the CEO to, to get beat up by the custom GPT first before they went into the live environment. And I would imagine the same type of thing. If you are, I don't know if you're an HR and you've got to have difficult conversations with, you know, certain roles within the company, you can prepare for those by, hey, here's what our, you know, our shop floor avatar looks like. Here's what our, you know, sales, you know, SDR avatar looks like. I've got a firearm or fire them up, whichever one. You could use this way more than just in the marketing. Although I know coming from direct response where the data matters. This is priceless.
- [32:51]
A
Well, I just thought of an idea as you were talking that I almost. Part of me in my mind is like, should I just, just don't even say it, you know, but here' what you could do. A lot of these VC investors are very vocal online. So you could create a synthetic audience of VC investors and run pitch decks to them until you have a really solid pitch deck and then go out and start pitching VC money. You know, if anybody uses that, like, you know, just shoot me 20 bucks when you make 20 million. All right.
- [33:23]
B
Yeah, yeah. Or certainly share that information. So I think this is great because what we're doing is, is right. Like we're, we're transporting this concept into other domains and applications. So as a listener, I want you to think about where else could I leverage this, this idea?
- [33:38]
A
If you're a salesman, you know, create some, you know, before you do your sales pitch. Right. Run your sales pitch through it. Or if you're a Sales manager, or you're the marketing manager and you got to write a sales pitch for your guys. Write it, run it through the audience before you give it to your guys who are going to say it on the phone, you know, so there's every. I think this could be used in a lot of different departments.
- [34:01]
B
You know, I just realized that that's a big miss. So for us in our sales environment, we. We have AI enrichment and stuff. If somebody lands on the website before that salesperson ever talks to the individual, we should be doing something like this where what's important to that individual based on what we can scrape from LinkedIn or from, you know, online. Wow.
- [34:23]
A
Yeah.
- [34:24]
B
Okay, we're going to pivot just a little bit and we're going to get into the. The hot topic of the day. When you and I met up in Fort Lauderdale with that group, I thought I was going to learn how to use Claude code. Like an expert. Right. Or, you know, I guess about a week before you and I went to Fort Lauderdale, Claude Bot had shown up and now it's gone through different names and that sort of thing. And I know that since then you've kind of dug into this, whatever that paradigm of agents is. And I want to hear what that experience has been like for you. What the learning curve was like, lessons learned and how you're using. Because you are using Open Claw, right?
- [35:10]
A
Yeah, I am, yeah.
- [35:11]
B
Walk me through kind of what that. That journey has been like, because it's only really been since, what, the 12th of January that it came out.
- [35:17]
A
Yeah. And when we heard about it, it was like days, you know, a few days old. I think I'm probably. I read in all of the hubbub and the hype of that launch, there was one article I read from a guy who said, my rule is, you know, when something new comes out, I ignore it for two weeks because in the first two weeks, it's like the worst version of it. You know, it's the hardest version, it's done. Safest version. I wish I would have done that because I jumped in at the hardest unsafest, you know, my. So mine is named Iggy. Iggy died. At one point I had to, you know, tell him, like, hey, you died. He's like, what do you mean, I died? So I installed mine on I. To each their own. There's an argument for installing it locally on a Mac Mini, which is. A lot of people are installing it on Mac Minis, or you can install it up on a server where it can run forever. You Just have to give it a browser because it doesn't have a browser like on a Mac Mini. So I use DigitalOcean, which is very scalable, very cost efficient, is $24 a month. So I could buy a Mac mini for 500 and $600 or for $24 I can run it per month. So I can have like three years for the cost of a Mac Mini. And then running it on a server, especially DigitalOcean, you can run it on something called Docker, which kind of creates a container around it. So there's just a few security things that I liked about running it on DigitalOcean in the Cloud. It's always on. It's, it's not gonna, you know, my kids aren't gonna come, you know, spill something on it or kick the cord out or, you know, power's out here, WI fi is out, you know, it's, it's always running on a good server and I can access it anywhere. And so you can use it the same way. You can still talk with it on WhatsApp or Telegram or iMessage, you know, the same way whether it's on the cloud or whether it's on your local computer. So now it's a much easier. There's a bunch of one click installs. DigitalOcean has a one click install. There's a bunch of different easier ways to install this thing. Now, all said and done, having used it first, I'll say it's cool. It's a pretty cool thing. I think it's maybe the dinosaur version of the thing we're all hoping comes from all this AI, you know, that Jarvis or you know, if you watch the Jetsons, you know, is the robot walking around that does all your dishes for you? I think it's the baby, baby dinosaur version of that thing. And that's kind of cool. Yeah, it's kind of a cool tool. But all the hype aside, its brain is still the same AIs we're using today, right? Its brain is clawed or its brain is OpenAI or Kimmy K2 or whatever you decide to choose as its brain. Its tools are the tools that we're using already. Whether you connect it to your WordPress or you connect it to your Salesforce, right? So it can't do anything new. I think it's easier to automate because you didn't have to go learn N8N you didn't have to go learn make or Mine Studio. You could just text it, hey, do this thing for me. It will go out Read all the documentation, learn how to do those things. So I think that's the plus one that I think this is added to the world is you don't have to go learn how to make automations, it will go learn how to make automations and do that for you. But that also brings a level of danger with it like when you don't know how it's doing what it's doing, it's building something you may not know how to unbuild or what's in there or. So it can be a little dangerous for the, On a local PC it has access to your imessage, to your email, you know, so if you're, if you don't know anything about cyber security and you're installing this like overlord thing on you on there, that's far smarter than you are. Yeah, that's a risky thing, you know. So I would say for most people it's probably not ready for you yet. If you understand a little bit of cyber security, if you put it up on a, on a, on a cloud where you can maintain it in Docker and things like that, then I think it's safe, it's cool, it can figure things out for you. But at the end of the day it can't do anything that the tools and the AI that's already available can't already do if you just learned how to use them better. But the plus one is it makes it so that you don't have to learn how to use them better.
- [39:58]
B
You know, when we were in Fort Lauderdale again, I thought I was already late to the party when it came to claudebot. It was called Claudebot at the time, but I had already built and just for listeners like you're hearing this stuff and I know that the hardest part for a lot of the AI capabilities is cool, but how do I use it? Right? Like identification of the use cases. So for me, I knew that there was an area where I was just, I was failing and that was keeping an eye on everything that was happening in the company where startup got 20 plus people building, talking, doing all kinds of stuff. So I said it'd be great if I had a chief of staff that was paying attention to all the calls I couldn't be on and the email threads and all that type of stuff and was just extracting signal from all the noise. Yeah. So what I did was I built my version of Iggy and it's connected to certain Slack channels where we're having conversations about clients and that sort of thing. Our Fathom video account, which Fathom, for those of you who don't know, it's a. An AI powered call recorder that shows up to zooms and everything. So it looked at all the transcripts that were labeled a certain way. It had access to my Gmail, access to Google Drive, and access to my calendar, and it was able to look. And the calendar of all of our chief AI officers that work for us, fulfilling for clients. So it knew who was having which meetings, who was going to be on which calls, it was reading those calls. And then I gave it some rules and twice a day, beginning of the day, end of the day, it basically gives me a red, yellow, green on things that I need to be paying attention to or whatever. And it, I integrated it to ClickUp and if it was like, hey, are these good? Yes. It would turn those into ClickUp assignments and then assign those to whichever the appropriate staff member was to do that. Now, as a business owner, if I were to pay somebody to do that, I mean, I did. It was very expensive and they still couldn't consume and ingest. Everything that was happening was better than me alone. But it wasn't, you know, holistic. This, it's become holistic. And I followed your advice. I. Mine's on Digital Ocean currently. And there's. I, you know, there's like a whole nother episode I'm gonna have to do about best practice, because the number one thing you talk about and every listener should think about this is that cool AI tool. Awesome. Well, what are the security implications? What's the security posture you need to have? Right. I went bareback. Like, I was just like, wild west, let's do this.
- [42:31]
A
Yeah.
- [42:31]
B
But yesterday in the chief AI officer community, we had a cyber expert come in who is doing a whole thesis on this. Like, how do we create a secure environment for this agentic capability that with something like Openclaw, anybody can access for 24 bucks a month. Right. So, interesting stuff. Where do you see this going specifically for. For your business? Because I know that you don't like to have a big team, you like to run lean and really leverage the technology a lot more than like, you know, needing to show up to the office and have all the buzz going on.
- [43:07]
A
Yeah, yeah. I got up to 13 people and I was like, I don't want to go farther. That's. Yeah, that's too many. There's too many people. So for me, I think it's an amazing reporter, like you said, you know, building new dashboards because it could go log into all the different things, you know, sometimes that's what I hate about this dashboard tool. And that dashboard tool, it doesn't have an integration to this and integration to that. Like it can figure out all the integrations for you and it can have the things custom for you, your own custom dashboard. So I think it replaces a lot of the dashboarding tools for sure. I think it's going to be amazing at customer support because it can do your customer support not only just in your email inbox, but it can go watch your DMS on social media and then it can watch your Slack group and stuff for you. So it can manage all those different things. So I think that there any data entry like hey I have these 3,000 records, I need you to port them over here and then over into there. You know, is the cheapest, fastest data entry that I think we're ever going to see. So data entry reporting support at least first layer support. Like I like to have some automated support because people do want fast answers but people also want deep answers as well. So we provide AI for the fast answers and then we have humans for the deeper connection type answers. So I think it'll be that first layer of support. So I think that that's, that's really what we're finding. I think this year is really about like, like honestly kind of end of 2005 this year is the first time these things were actually good enough for professional level work. You know, they don't have six fingers anymore. You know, text is right on the images. Like we're finally getting something. We can actually turn into the boss or we can turn into the board or the customers and it actually and it pass. So I think we're figuring out what does it look like to have this in our workflows. But I do think the immediate areas, there's a lot of benefit in using this in marketing because it can create a lot. If message to market match is one of the biggest things in marketing, well this thing can create a lot more messages faster to get you to message market match. Then you deploy your human high paid talent on that message to market match but let the machines do all the bulk grunt work to get you there. So the reporting, the support, the marketing, there's a lot of areas where this is going to touch. And I think we're all figuring that out this year as we go forward.
- [45:36]
B
Man. You know, I think back to so when, when Chat GPT 3.5 came out at the end of November in 2022, I was like big deal. I'm not a technician. I'm like, it's going to be a huge opportunity for somebody. I just didn't understand generative. Right. And then pretty early, like January of 23, somebody was using it and I was like, oh, my gosh, this is incredible. And immediately I thought, I'm behind. Right. Even back then with 3.5, which sucks. Right? You know, I mean, compared to what we've got.
- [46:10]
A
Right. Yeah. That was like sticks and rocks back then.
- [46:12]
B
Yeah. Now anybody on their phone can download an app. That's incredible. It's interesting to see because we, we, we were talking about agents and our chief AI officer certification in 2023. But the stuff we were talking about then versus what anybody listening to this can do today was completely different. And if you think about it, Thanksgiving 2025, like just a few months ago, none of this stuff exists. I'm saying it didn't exist. I didn't know about.
- [46:39]
A
Yeah, I literally don't think Claude Bot or Openclaw was out then. Opus 4.5 was just coming out. So, yeah, there's. There's some things that, you know, Thanksgiving and beyond, all of a sudden it went.
- [46:52]
B
And like, we're still early in 2026. And just in the past week, GPT 5.3 for Codex came out 4.6. Opus is now released. Like this acceleration. So you and I, what we're talking about Today, buddy, in 12 months, it's going to look like we're talking about sticks, right?
- [47:11]
A
Yeah.
- [47:12]
B
Right. And it's so crazy that we're like in the middle of this velocity. So for listeners, business people who have no idea what some of this stuff is. Hold on. Ride the bull, as we call it. Because this is how you get ahead this knowing about this stuff and having somebody on the team. Are you saying, you know what, Let me figure this out. Because the reality is no degree is required. No, like burning the midnight oil to study coding. None of that is required. A shift in how you think about stuff. Like, you know, I'm sharing some examples and Justin's like, oh, that just made me think about xyz. Right. When you get to that point and you're able think about generative AI and these tools through that lens of how does this fit for me? The whole world will open up to you on this AI stuff and you will no longer feel overwhelmed. You'll no longer feel like you got to stay on top of all the tools and everything. So an amazing time to be alive. Justin. I want people that are listening to this to Be able to follow you because as stuff is happening, I know that you're like talking about it right
- [48:12]
A
away, at least in the marketing world of it, you know.
- [48:15]
B
Yeah, but what I like is that you're not. You're not like the hype boy, right? Like there's a bunch of hype people out there talking about how great stuff was. But oh yeah, and there's big security breach, you know.
- [48:25]
A
Yeah.
- [48:26]
B
You're not that guy. So I want people to plug in. Where are you sharing most of this research?
- [48:33]
A
Most of the. I'm sharing on Twitter, you know, so if you look up Justin Brooke on Twitter, I've got, you know, like a marketing profile. Then I got an AI profile, you know, I'm the same guy. But like for the algorithm, I have to split it up, you know, and so at Agent Skills is. Is me on there. So yeah, if you look me up on Twitter, that's my main home base. If you look me up on Google, you'll find, you'll find me. I show up first for my name. Luckily, there's not a whole lot of other Justin Brooks out there, you know, so Google me, you'll find me pretty easily. But yeah, I'm @AgentSkills AI is my website.
- [49:07]
B
Awesome. And we're going to put all that in the show notes for the listeners so that I would encourage you. There's not a lot of people that I follow on X that I don't have to run it through a filter of, you know, the BS sniffer. And that's one thing that I appreciate about the stuff that, the perspective that Justin is sharing and that is like, he's mature enough in his business journey that he's not all about like, oh my gosh, this is amazing. He's like, it's amazing but. Or it's amazing and. But like, that's the perspective you want in this AI space. You don't want to. Just the hype people, so. Well, Justin, thanks so much for taking the time today to share this. I was excited about this episode. Is there anything that you want to leave the listeners with before we wrap up this amazing episode?
- [49:56]
A
You know, I would say I hear a lot. How do I keep up? How do I stay up on all these things? It can be impossible to try and keep up. You know, a friend of mine, Matt Wolf, who does this for a living, like it's his job to report the. He has even said, hey guys, I'm only going to cover these things, you know. And so I would say for you, like, for Me, I'm just keeping track of AI for marketing. You know, what is your thing? What is the one? Like, me and my wife, we have the idea of, like, there's a few balls we can't drop, you know, and so, like, what's the balls you can't drop? Just focus there. Hone in a little bit. There's a lot of great stuff. It'll be there. You can get there. But for me, I pick a couple of models. I pick one lane that I'm in, and it's like, all right, now I can filter it. Now I can understand what's coming through there. And then I can bring value through that because I can. I can drink from that water hose,
- [50:55]
B
you know, so that's great advice. I would second that for sure. For anybody that's feeling overwhelmed, pick your lane and just get good there. Don't worry about the rest of the stuff.
- [51:04]
A
Yeah, yeah. Chris, thank you so much for having me on.
- [51:07]
B
Pleasure. Thanks for being on here and everybody, you know, my closing advice all the time is just go use the tools. Go use AI. So we'll see you on next week's episode with again, another amazing guest. And thank you so much for being a loyal listener to Using AI at Work. Thanks for tuning in to Using AI at Work. Don't forget to subscribe for more conversations about how to use AI at work. And a special thank you to our sponsor, Chief AI officer for empowering businesses with AI education and training. This. Visit their website for a free AI readiness assessment and AI strategy guide to help you get started using AI at work. That's www.chiefaiofficer.com. follow us on Twitter at the handle usingaiatwork and visit www.usingaiatwork.com for free resources to help you harness AI in your role.