AI Business Transformation with Gerard Murtagh

Introduction and AI business transformation overview

00:00Bill Gallagher: Have you ever had a week when you're completely slammed, but somehow nothing actually moved? Is this one of those weeks? That's not really a time problem. It's a busyness habit problem. My new book, Busy Is Broken, Do Less, Scale More, is all about growing by doing less, not more. Read or listen to a sample chapter over at busyisbroken.com. That's busyisbroken.com. It's also on Amazon and other booksellers.

00:32Bill Gallagher: So today, we're talking about an AI business transformation. This is a real story about a real business who did real things that you could do too. It's not some concept highfalutin AI carpetbagger out to tell you a bunch of snake oil that doesn't actually work or doesn't make a difference. It's a real story about a real company that I met last year at Harvard and was really impressed by this guy. Humble, awesome. Well, sort of humble. He's proud of what he's built. He's going to tell you more about that. So hang in there just a second. We'll get right into all of what happened with his business and what it means for yours.

01:10Bill Gallagher: So hey again, everybody. Bill Gallagher, scaling coach and host of the Scaling Up Business podcast. Our show comes to you every week with stories about growing and scaling your business with gurus, authors, experts, CEOs, all that. More than 620 shows now, wherever you're getting this right now, or at scalingcoach.com. And our website scalingcoach.com is where we also have monthly workshops. We just did one yesterday, and we talked a lot through the basics of Scaling Up now, but also we started to add in what are the AI prompts and the things that we're using with our clients? What are the methods we're using to refine strategy to get the planning done faster than ever before with more ease? So all that and more at scalingcoach.com.

01:54Bill Gallagher: Alright. Alright. Alright. So I want to welcome now, he's on the screen already if you're watching, Gerard Murtagh. Gerard, what city are you in?

Discovering $1.4 million in fraud during hyper growth

02:06Gerard Murtagh: I'm on the Gold Coast, so I'm in Australia. So Gold Coast. Morning to me. Good night to you. Good afternoon. Wherever you are in the world right now, thank you for listening.

02:16Bill Gallagher: Or wherever you're listening right now. Right? People may be listening live. They may be listening on a recording or watching later. So whether you're watching us, however, Gerard joins us from Gold Coast, Australia, where he's got a company. He's a CEO of a company called MouldMen. And he told the story of how their business really transformed and being ongoingly transformed with AI in a whole variety of ways. And it kind of started with a surprise, kind of a crisis that he'll share more about. We'll get into all that in just a minute. The surprise, the thing that led to it, what they're doing now. But you guys, this is a cool story of not just one thing about how their marketing got a little bit better with AI, but like, across the board in all areas of the business. So thanks for joining us.

03:07Gerard Murtagh: You're welcome. Thank you for having me.

03:10Bill Gallagher: So you started as an entrepreneur at age 17. Your family had a business. Right? There's a lot of that. Before we go into the crisis and what led you into AI, talk a little bit about kind of that entrepreneurial upbringing.

03:25Gerard Murtagh: Yes. I come from an entrepreneurial family. My father had always owned businesses. My mother had owned businesses. And at a young age, I worked out that I wasn't sort of going to be going down the path of a normal career. I was lucky to have a number of different neurodiversities that made learning very difficult, but they also gave me an opportunity to lead and to use my voice as a way of achieving my goals. And I was very lucky to have some great mentors around me that said, Gerard, you're not going to get a job somewhere, but you can own the company. So if you want to be an architect, start an architect firm. If you want to be a doctor, then own a surgery, but you're probably not going to be doing the actual thing. So that was sort of my upbringing. My father and mother were in their small business. And then coming out of school, I was like, well, I'm just going to be owning businesses. I'm not going to be getting a job, so I won't be able to compete. So I started my first business, then sold that, and then a year after I finished school, my father passed away, and his small business needed someone to run it. And so I was 19 years old and thrown in the deep end, and it was probably a baptism of fire, and that business still runs today, and I run that with my mother. And that's not MouldMen, but I'm very proud of that start. But that gave me a huge opportunity to, you know, I wanted to learn about business. Well, the way to do it is to be in it, and I was from a very early age, and it set me up for where we are today.

05:09Bill Gallagher: So MouldMen is a mold remediation and restoration company?

05:16Gerard Murtagh: Yes. Yeah. So MouldMen came out of a flood that we had. We hadn't had a flood since the seventies. In 2011 was the first flood that we'd had in our city, and nobody really knew what to do. Like, everyone was like, what are we going to do? And obviously, there was some incumbent mold remediation companies, but they were all taken by the insurance companies very quickly. I, at that point, owned a pressure washing company, which was just another entrepreneurial venture that I got into at that point, and we'd been running for two or three years at that point, and we had all the contracts with the city councils around our area, so they were calling us and saying, hey, you need to come and pressure wash this or that, but then they started talking about mold, and I was like, okay. Well, when was the last major flood in the world? And that was in Hurricane Katrina over in New Orleans. And so I jumped straight on a plane, met some people, went to the city council and said, I'll work for free. Can you just teach me how you remediate homes after a flood? After about two weeks, I thought I knew everything, and we came back to Australia. And after about three days of being called the MouldMen, I changed the name to MouldMen, and the rest is history. So we now are a national business here in Australia.

06:31Bill Gallagher: That's a remarkable thing that not everyone would do. You heard about a related event on the other side of the world, and you got on a plane, and you went and rolled up your sleeves and learned firsthand. Right? That's not a normal way that people jump into a business. Right? It's a remarkable thing in and of itself, like just actually digging in and doing something.

06:51Gerard Murtagh: Yeah. It is potentially for most people might seem a little bit odd, but the reality is that for me to really understand the business, I have to be able to do it. And a lot of my businesses at that point were about doing. So the business that I had with my family was exporting different products into Asia. So getting on a plane wasn't really that hard. It was five weeks after my father had passed away, I was in Asia meeting clients for the first time and shaking hands and trying to navigate streets that had no English on them. And so for me to just jump on a plane, by that point in my life, I was about 26 years old. I'd been doing that for a number of years. And for me, business is always about, you know, the E-Myth is another, is a great book that most people have read or should read if they haven't.

07:38Gerard Murtagh: Oh, that's amazing. Well, that shows, like, as business owners develop a business from scratch, they sort of need to understand. And then once they can document it, iterate it, and then be able to train it, they can then move on to the next thing as it grows. And that's, we'll come back to that because that links into what everybody needs to be doing with AI. But that's how I learned. And I thought after two weeks I knew everything about mold, but it was a great experience being in New Orleans. But it was time to come home. And like I said, there was 200,000 homes that had been inundated. The insurance companies were backed up for years, and there was an opportunity to help people that were underinsured or that just houses had just gone mouldy because it was just so wet here. And then from there we realised, okay, there's a real business here, and there's no one servicing these different segments of the market, which are property managers. They're, like, the people that haven't been flooded, but their houses have gone moldy. Their bathrooms are poorly ventilated, subfloors, basements. Like, no one was really servicing these clients. And so as we've gone on for the last fourteen years, we've been able to continually grow the company by making sure that we address our core market segment, which is property managers, and now we have over 6,000 around the country that use us, and they control 2,800,000 rental properties. So for us, our goal was to be the first choice in mold. We are the largest now in the country, and there's no sort of stopping us from growing. So it's just ourselves.

09:12Bill Gallagher: You had a crisis. You had a thing happen, and that led you to your first dabbling with AI. Talk about what happened.

09:20Gerard Murtagh: Yeah. So as businesses go through these different changes, you start a business from scratch, you know, one man band, and you go through these different stages. As a business, we were growing at, like, 50% a year, year on year on year. And then we were in three cities, and we had just, we had a flood in every single one of those cities in one year at the same time. So this big weather system came through, and so we went from a really good local business in our state, doing really well to three states and during COVID as well. So you're in the middle of COVID, you can't really travel, you've got these three floods that are happening, you're the first choice for a lot of companies to use, and we grew that year over 300%. So for anybody that's been in a hyper growth market or a hyper growth industry, you go through these things where we all get told this perfect beautiful line, and it's awesome, which is, you know, what got you here is not gonna get you there, but no one finishes the sentence. They never give you the context, and that is you need to throw everything out and start again. And some of those things you'll put back, but the reality is that a lot of what you throw out that got you to say 1,000,000 to then to two to four to eight to sixteen to 20, those different iterations, you actually need to step back and throw everything out and then rebuild the business for a $20,000,000 business or wherever you're going. And that's where your strategic plan, and that's where Scaling Up comes in and that. But for many years I thought, okay, well, we were here. That worked. So let's just add on another process and add on another process and add on another process.

Using AI to analyze financials and uncover anomalies

11:04Bill Gallagher: Did that come at the same time as the fraud, or was that an earlier event?

11:09Gerard Murtagh: No. So the fraud happened because of the 300% growth, because as we had created all these processes for a business that I could eyeball everybody, everybody knew my name, they knew my family's name, we were a close knit group that had grown this business very well up to that point, and then you start introducing different people, but we had the same processes in place for a business that was three times as small only, like, nine months earlier. And that gave people the opportunity to join our business, and we're talking like more than 80 to a 100 people joining a business in a year. They've got the opportunity. They go, well, how do you, for instance, how do you make a credit card payment? Well, we go into Jared's office, and you get his credit card, and you buy the flights, and you go. Or if you need something, you go and do that. And so for me as a business owner, I was extremely naive. I made a lot of mistakes, and what I did is I transferred that trust from the 40 people that I knew their names to a 120 to a 160 people, and there was too much opportunity for people to do the wrong thing. And unfortunately, good people got probably got too tempted. And by the time, because we were growing so much, the jaws of margin were just creeping up, but the margin wasn't, was not getting any bigger. It was still, but for a business that was making profit of 5 figures a month, and now it's going and making 6 figures a month, you're like, wow. We're doing so well, but we should have been doing way better. And that was where, that was unfortunately, by the time we did work out what was actually going on, it was just rampant. And I was on a plane probably three to four times a day, my accountants once they picked up on it, we, it was a wildfire. But yeah, it was that, I think at the end of the day, there was over 600,000 misappropriated, and there was a lot of tax and other things that weren't lodged and paid for that were documented, but then weren't done. So in the end, it's been about a $1,400,000 hole that we had to dig ourselves out of. And to go from a business that had always been profitable and then you think that the profits are gonna be in those 7 figures to then be going the opposite way. Yeah. It's been a, it was a massive learning experience and one that has shaped me and has now made this story.

13:40Bill Gallagher: So I think it's useful to point out that just like the basic phenomena. So you work on things, and then either your strategy is right or suddenly fits world circumstances, something shifts in the world, and then your business is flooded with business. Right? When you go through growth, when you start to hit a 100% or more a year, it's really overwhelming. Now it's a little bit different if you're, say, a SaaS business or something where the scaling is a little more automatic. You could just turn on more servers. It's different if there's a lot of doing. If there's a lot of people and trucks and things running around, it can be more overwhelming. Even in a SaaS business, you've got new customers to make sure they turn on right and get successful. But in anything, there's a moment in inflection when everything, like, gets overwhelmed. And in that space, you had this lack of control and then things got out of whack. And through direct and indirect actions, like, now you've got a $1,400,000 hole.

14:49Gerard Murtagh: Yeah. Yeah. It's, it's been, like, I call it, like, the best lesson I could ever learn. Like, as a business owner, I'm constantly learning, and I learn from failure. And we try and fail fast and hurdle those stories. But, you know, we had to work out what our plan was.

15:07Gerard Murtagh: What's the problem? You know, what did we learn? How do we make a quick call? And then how do we make sure we don't do it again? But in that, we also have to make sure that you don't just add another layer of process or add another layer of stopgap or employ somebody to be my gatekeeper, you know, at six figures a year to just make sure that people aren't stealing things. Like, there's got to be better ways. And unfortunately, when you look at a business that's been around for, you know, more than ten years, you do have a lot of these legacy processes that you just hold on to, and that's where something I learned through this is that don't add another process, actually throw it out. So we'll start again. And there'll be parts that you'll put back. That's fine. But what you really need to be looking at is there may be some technology out there that you can use that actually stops twenty steps and saves you a lot of money. And for instance, we found a credit card company called Airwallex, which was an Australian based business that we never knew about. That means I can put a credit card on your mobile phone within, like, ten minutes. I can give you a certain amount of funds, and then I can cancel it immediately or freeze the card. And that was just the one instead of having more credit cards or more stopgaps or more problems, we were just—everybody that needed to buy things, you know, were given different processes. So that still meant that we could run fast, but it meant that we had control over the spending because there was—oh, I can run it on my phone. I can turn people's things off. My accountant can turn things off.

Cutting admin from thirty-six people to fifteen

16:32Bill Gallagher: You used some AI to investigate the fraud, right?

16:43Gerard Murtagh: Yeah, we did. So when we—we got our—so we knew that we had a problem. And so what we did is we downloaded all of our everything into a CSV file, and then we uploaded that into pretty much into a ChatGPT wrapper that said, these are all your anomalies. And one of them was $9,000 in Uber Eats in one month. We had nineteen subscriptions for Spotify. We had—there was people taking flights that weren't my flights, but booking them under my name. There was—and so using that was, like, the first step of, like, okay, this stuff's really amazing. It can get me information quicker than someone else can. And being a business owner that had been so focused on process, and for me personally, don't get me wrong, I'm the guy that said put in a layer of process. Like, a lot of this stuff comes from me. Like, there's definitely no, you know, amazing genius here. It's like, let's put a process in so that it doesn't happen again. Well, those processes then, like I said, had to be removed. But because I loved what actually AI is, which is train a model, get the data, train the model, give it the outcome or the prompt, and then be able to then make a decision. I started to realize very early that this was my dot-com boom. This was—somebody that wasn't in business during the dot-com boom. This was the dot-com boom, but it was something that I needed to invest in every single day. And then I'd harp on it a lot to my team and to people, to anybody that'll listen, is that this is the thing that we all need to be, like, leaning into, and it's not just for the big companies. Because if you left it up to the big companies, they'll run this. They'll create businesses that are so much more efficient and profitable, and then the little guys will be left with less than ten percent of the market. And so what you need to do is make sure that you get into that ninety percent, because if not, you'll be just left with the crumbs. But we can talk a little—we can dive into that a little bit more. But that's how—like, going back to your question about that's exactly how we found out that there was a lot of abnormalities, because we were able to compare data from a few years earlier where the accounts were not being abused to there. Now it's like, well, yeah, there might have been some things like, yeah, your sales went up and your costs went up. But we dove into that, and we found all of these different anomalies.

19:13Bill Gallagher: So let me just throw another little teaser out. Your business now across operations, across scheduling, across customer interactions has AI processes running, hiring, other kinds of parts of it. What came next after the fraud analysis and starting to look at the mess that was there?

Automating email routing and job entry

19:35Gerard Murtagh: So we'd grown an admin team to thirty-six people. So that's people answering phones, booking in jobs, and that's a really, really expensive component of a trade business because you need them to do the business to and to be able to run a company. However, they can also drown the business, and that was one of the parts of the problem. When you've got so much process, you can throw people at a problem, but usually you've got the process that you've got to then identify, then you've got, can it be automated, can it be done through AI or automation, and then you've got the human element. But before AI, it was, you've got a problem, either perfect the problem through process or throw people at the problem, and a lot of people, especially in businesses that aren't looking at AI, will throw somebody at the problem. Like, Gerard, we need somebody to make sure that no one's defrauding the credit card. Let's go out and buy that and go get a, you know, a new CFO or a CFO's assistant or a bookkeeper. And you're like, hang on a minute. We're just spending all this money on somebody to do that. We're just throwing people at a problem. But when we were growing so quickly, there were some days that we doubled the size of the company in an eight-week period. You're like, just hire somebody. Just hire somebody. Just hire somebody. So once AI—so there was this inflection point of you've just found out that you've been defrauded or that there's been a misappropriation of funds right through to, hang on, but you've got this tool now that can absolutely revolutionize the way that you're going to run your business. So the next part after that was what is our admin process, and how do I cut down that number of admin that we have in the office? A, because I can't afford them anymore, and we have to perfect the process. Because if we don't perfect the process, then we're going to send ourselves broke. We're going to grow broke again because we would have just continually throw people at the problem. So the first part of the process was that, and that's where we've messed up as well. So at that point, most companies that were previously building a website changed their website around and said they were building AI. And so I was, you know, yeah, we can revolutionize your admin process, Gerard. And I was just doing all these projects, and none of them worked. And it took me a few mistakes to get to the company and to find the team of people now that we're now using, that we actually trust, and that have actually started to build it. So the first part of—to go back to your question, the first part of our process and the first part that anybody needs to be looking at if they're running—if they're looking at investing heavily—

22:04Gerard Murtagh: The first step is: what is the process and what is the problem that you're trying to fix? So identify that. The next thing is you must remove, you must cut it back. You need to cut back. And Elon Musk said if you don't have to put 10% back, then you haven't cut hard enough. And so we live by that. I would prefer for something to break because we didn't—or a process to break because we took a cog out of the wheel to speed it up—than to worry about having too many cogs. So that's the next thing. Then you start looking at AI and automation, because if you start building AI for an old process that is redundant, then you'll end up building things that you don't need, and that's expensive.

22:47Bill Gallagher: With so many companies—and this is a very old thing, this is not a new thing—don't go automate a broken process. Right? If you have some weird roundabout thing that you do in your business that isn't smart and you don't know any better, you want to first examine: is what we're doing, does it make sense? Is it as good as or better than our industry? Right? You don't go build software or routines around something that you shouldn't be doing in the first place.

23:18Gerard Murtagh: Yeah. And so the fourth step—I actually 100% agree. And the reason is, and this is the biggest problem that people are going to face as they're building AI, is the people aspect of it. But I'll just go through this. So the five steps are: what is the process? Cut it back. Can it be automated or can AI be put into it? If it can't, can it be done by somebody offshore or in the Philippines or something else? And then if it can't, how do we optimize it for an onshore person? So what we're finding is that there is that element that we can automate a part of the process, send it to the Philippines, and then have it double-checked either by AI now or by an onshore team member. So for instance, something that would take eight minutes now takes less than fifteen seconds to check. So those are those little one-percenters that we're looking at, which is seven minutes and forty-five seconds every single time. And if you're doing that by 300 action points a day or 400 or 500 action points, that time adds up. But one of the big issues that people are going to face with this is that third step of can it be automated. And we've found in our business—and if people, my staff are watching this, they're going to either nod or they're going to say that was me—but we have had people that have created process because it has validated their role. And what people are going to find, that CEOs and the COOs, the CTOs, any operations people in your businesses right now, your team members are going to fight back against AI and automation because it validates their role. You can tell them till the cows come home, like I do, that I'm not getting rid of you. I'm just getting an eight-minute task down to a fifteen-second task. That's what I'm trying to—I still need you because it's called human-in-the-loop interaction. You need to have a human in the loop. So I'm not looking at getting rid of people. What I am looking at doing is giving them the ability to do a vast array of different tasks that are high-value tasks instead of doing that, but doing the low-level task or the task that can be done via automation. And that's where we're finding the biggest struggle that we're finding in even our business today, is that processes are being designed to validate people's value inside a business, and they don't want to give it up. They're like, I have to check the credit card statement. And I'm like, no you don't. You put it into our system. It checks it, looks for—it's got its gates. We call them gates. Like, how much are we spending on Google? How much are we spending on subscriptions? How much are we spending? And if those gates come back red, then you know you've got a problem. That takes checking your credit card statement right down to a minute task every day instead of it being an hour long or even a full week of someone's task. So if anybody takes anything from this today—like, there's a number of things that you can take away from it—but understand that the people that are inside your business aren't going to be as excited about AI as what you are, because you can do so much more with the team you have if they're engaged in the process. Or what they will do is they'll just go, you know, I'm out. I'm like, you're going to fire me, so I'm going to be the strong one and I'm going to remove myself from this place because I'm not going to have a job because you'll just get rid of me. And it doesn't matter what I say to people. I'm like, you're either going to be using this technology here for the first time, or you're going to be using it in one, two, three years' time when your new CEO implements exactly what we're building here, which might sound a bit arrogant. But the reality is that we're building things that we haven't seen, and that's why we're having to build them.

26:49Bill Gallagher: I think around a lot of different AI applications that we're seeing right now, that human-in-the-loop checking is really important. When I think about planning strategy work in particular, as a coach we're using with clients and that continues to evolve now, but we get really interesting useful drafts of things. But it's not like the final thing. You can't just use a bunch of prompts and then go, okay, that's your strategy. It gives you sort of like a draft, but it gets you so much further down the road so much faster. So we'll come back on that. Why don't you describe what it's like to do—like, what's the customer experience today and then the operations behind the scenes, and where is AI going on within that? Give us a picture of that.

27:35Gerard Murtagh: Yeah. So most of our clients are property managers or real estate customers. So they send us a work order that comes through their job management system. And in the past we would have looked to try to create API links or Zapier links between those. However, every time they've changed their technology or they update it, the APIs don't work and it just doesn't work. So our AI is able to read our emails and identify what is actually going on in that email. And if it says certain things and it can understand—because it's had, it's got over two million emails inside our data lake—so it can understand what's going on. It can create—it goes through, it identifies it, and then it puts it into a new folder. That folder then scans that PDF and puts it straight into our job management system. So what would normally take somebody a couple of minutes to put a job into the system, an email comes through when they manually do it, is all done within about fifteen seconds. So that's all done now via automation and AI. So that's that first part of the process. For a job, interpreting whatever the customer sends over, putting it into your job system—all that is happening by AI.

Route optimization saves time and increases efficiency

28:45Gerard Murtagh: Yeah, and as we train the model more—and we train it every single day—it's now being able to work out, hey, this doesn't have these things, I'm going to go back and ask it questions. So you can ask the client, hey, can you send photos? Can you give us some more information? Have you had a water leak? Do you have an exhaust fan? Do you have these different elements of the process that we need to make a decision? So that when it gets to the agent, the human in the loop, they can look at it and go, everything's there. This is—we can actually quote on this immediately, and we don't need to go out to this site. We have a photo. They don't have a water leak. They've got no exhaust fan. The mold is being caused by a lack of ventilation. We know what to do. We've done tens of thousands, hundreds of thousands of jobs. We know exactly what to do. Here's your price. It gets it prepped, collects additional data if it thinks it needs it, and then the human looks at the thing and says, okay, here's what you got. They do the final check on it.

29:38Gerard Murtagh: And so what we also do is we train the model based on what it did. So well done, you did this well. Like, we talk to it. Its name's Tye, because we kept calling it V-A-I, so it's now called Tye. So Tye checks it, and Tye will look at it, and then it will give—it'll wait. It has an alert. Somebody, a human in the loop, will check it, and then we'd go tell Tye that it did a good job. Now if Tye knows it's done a good job, it'll continue to learn that it did that well. And just like we train a human being, you're training this brain, a model, Tye, on how to make sure that it learns. So as we get further down this path, I won't need to ask somebody. It'll just send the quote. And instead of us having to check every quote, we'll check 95%, then we'll check 90%, and then we just keep iterating constantly and making it get better.

30:29Gerard Murtagh: So once the job's in the system, we then need to book it in. So what we've built is our own scheduling bot. Now, we're still training this, but it gets used every day by a human. So a human in the loop basically goes in: build us a run. Instead of it taking a day to go, where do I want the technician to go, it doesn't have to do that anymore. You put in—you ask for it to build a run. Tye builds the run. It then says this is the most optimized run. Now Tye is 57% more successful for optimization than a human, and it can do it within a minute when it can take a day to get a human to build a run. Like, where do I want to go now? Everybody, I want you to think about your local city and create a circle around it. You want people doing jobs either in that street or within a couple—like, 500 meters of that street. Now if a job's come in right now and Tye knows that our technician is going there tomorrow, but it can fit in that inspection or that job, Tye will be able to then put that in. And instead of the technician driving past that person's house, it gets that job in immediately instead of them having to wait two weeks when we're back in their street.

31:38Gerard Murtagh: So I think that's—because not everybody works in route scheduling and route optimization, whether it's moving things around within a factory or a warehouse or being out in the world and dealing with traffic.

31:50Bill Gallagher: Or your kids, or about Saturday sport when you've got multiple kids. Think of a complex problem and difficult to do and manage. And people have played with software forever for this, but AI is a uniquely useful tool for this kind of route optimization scheduling.

32:09Gerard Murtagh: Yeah. So the way that we've been able to do that is that we've been tracking our vehicles for many, many years. So we're able to use that data from that tracking. We also have got an API from Google, so we know what traffic—like, we don't go near schools at 3:00 in the afternoon. You don't go over that bridge at 5:00 in the afternoon. You stay away from those shopping centers on certain times. Yeah, it knows where to go and what not to do. As we iterate and we get better at this, we'll get that data live from Google. So that means if there's an accident, it can reroute the guys. It can reschedule jobs. This all just takes time, and this is why—like, we're eighteen months to two years into this, and we're still learning, and we're still training, and we're still getting better. It's not just—we didn't just buy HubSpot off the shelf and we can start using it after. We're working on this every single day, but Tye is like having an intern that you're trying to train on how to do a task and it never turns off. It's next to every single person, and then over time, it'll just become much more intelligent and it'll be able to then make decisions based on real-life data and not on people's work hours or availability or understanding that the guys are in that street right now—which is why the person's probably called us because they've seen the van, and that's why they've called because they've got mold, and all they need to do is walk across the road. But they call a call center and no one knows that they're actually across the road unless they ask the question. Hey, how did you get our number? Oh, the tech guys are across the road. Perfect. I'll send them across. Tye doesn't need to know that. It knows where every van is right now, so that gives us that element of immediate action on what's coming in and what data is being used to then make these decisions.

Employees resist AI because it threatens their validation

34:11Gerard Murtagh: From there, you've got customer experience on—how to—what questions you have. So frequently asked questions are now all uniform, so I don't have to worry about anybody going off script. So as a business owner, somebody over-promising or saying we can do something when we can't, or not telling somebody that we can do something. Like, hey, can you guys fix my bathroom? No, we can't. We only do this. Or, yes, actually, we can fix your bathroom with mold. You know, we don't have to worry about those decisions not being made. And then from there, there's a customer experience going past that. So how do you stay in contact with people? How do you make sure that they know that you're still here? If there's a weather event in the city, we should be able to pick up on that very quickly and then be able to identify the clients that are in that area and then contact them and say, hey, do you need us? We're here if you do. So everything that we're doing now is instead of throwing a person at the problem, we look at the process, throw AI at the problem, and then just try and get that smarter and iterate.

35:09Bill Gallagher: So how many people are you today?

35:09Gerard Murtagh: So we're a lot smaller than what we were a couple of years ago. I think we're maybe just under 100 or 100, but we're pretty close to where we are right now. Better. And we got a lot higher than that, but we've got 15 or 16 admins now and not 36, but we're doing more revenue than what we were doing when the fraud happened. So it just shows that when you've got the systems and processes and you've got this in place, you don't need as many people because you're more efficient. Our technicians are doing more jobs, our admins are being able to do more with less. And in the background, the investments now, we might be building a new piece of technology in a new part of Ty, but by adding a little bit more into Ty, it may be maybe a quarter of the price of a full time employee for the year, so it's a very good investment.

36:10Bill Gallagher: The customer service follow-up, invoicing, emails, collection, any of that stuff got AI?

36:18Gerard Murtagh: Yes. Email collection is definitely automated emails. Invoicing, no, not just yet, but it's definitely in the pipeline. That was offshore, so that was one of those processes. What is the process? Can it be, can we remove steps? What is the, could it be AI automated? There's parts of it can. Can it go to the Philippines? 100% it can. So it's in the Philippines and then an Australian based team member double checks and then they're sent off. But our team in the Philippines are amazing, so they're getting 100% scores nearly every day. They're not having that, my team have turned an 8 minute task into a 15 second check per job.

37:04Bill Gallagher: There are so many companies that are doing offshore talent for a whole range of tasks, and you can find some really lovely, competent, and even highly educated people in some of those offshore folks. One of our sponsors, regular sponsor here, is probably not the company you use, but we use a company called DOXA and recommend that to a number of our folks and worth checking out if that's something you're looking for. Tell us about the tools you've used and tried, and what are the mix of tools? And you said you went through a couple of vendors and that you finally settled on somebody that could be a good partner for you. Talk about that, tools and partners.

37:38Gerard Murtagh: Yeah. So the tools, so obviously, we've all got access to ChatGPT and Grok and Llama 3 and all the others that are rolling around now, Claude. These tools are absolutely amazing, and they need to be used. So I always recommend to people, part of my process is that I have like a separate Instagram account that I just follow AI people. And so every day I have time and as you know, get distracted with AI or distracted with technology, but I jump in and I go through everything that's come through that either through X and on Instagram of anybody that is of influence that is talking about AI. So that's one of my tools that I use on a daily basis to stay up with when are the announcements happening, what are they announcing, and how do they work, and then what. And then because it's not on my normal feed, it's certain amount of people, and then I get through that feed that says you've seen everything you need to see, and then I get out of it. And that's one of those that's a very, very important tool because I wouldn't suggest you're doing it in your normal feed because you'll get caught up with your best mate's trip to Tahoe or Japan or whatever, and you'll start going down a rabbit hole or that, and you'll realize you've wasted all this time. So so that's the first tool. The next tool is you need to be signing up to all these tasks, and you just need to be looking and learning how to use ChatGPT or Claude or any of these others, because some are good for some of the tasks and some are better for others. But what we're realizing is that these big LLMs, they're going to become the engine for what we do, and then what we're going to have is these wrappers that are going to go around it. But what you need to be careful of is if you are using ChatGPT, you don't have any of your identifying data. So if we put in a, if I put in my, say, any of my financials into ChatGPT like I did in the early days, which I don't do anymore, I made sure that we didn't have any of our company names or anything in that data. It was more about, we had obviously the Uber Eats and all of that, but we didn't have any of our data. And anybody that's putting their own company data into ChatGPT, you're pretty much giving them access to everything that you're giving them. They own it now, and they can use it to keep training their model. And that's how these models are getting very good very, very quickly. So be very careful how you use them, and if you are, if somebody is building you one, make sure that they're not just building it on ChatGPT or one of these others where you don't have the security of your data. That's changing every day at the moment because people are working out how to do it without giving your data over and just using that LLM as the engine. And so what I like to say is, like, your car might be the, let's say you've got a Mercedes-Benz, and you're going to go put a Bugatti engine in it, and then the Bugatti engine might not work anymore. So you'll go and put a, an Aston Martin in or you'll put in whatever it is. The car is not going to change, and the car is the data. And that's why, we'll talk about what the process is for people to actually go through to start really investing in AI, but the car is not going to change. The people, the contents, the data, but the engine that makes it go fast, that's what's going to change in the future. So being able to identify what works to go fast or slow or whatever you need it to do, that'll become very important. The other thing that we then ended up going down was building our own LLM, our own data lake. So moving on from my last point, if you are really serious about transforming your business, you need to be looking, the first thing you need to do is your process and identifying that. But the next thing is to start looking at building your own data lake. And your data lake is like a brain, and maybe this isn't the right way of putting it, but you think about a brand new baby brain, parents pump information into it. It takes information in from everywhere, but as you raise a child, you are giving it the information that you want to give it, and an LLM is pretty much that. It's a naked brain with nothing in it, and you just give it the information that you want. So if you don't want it to know anything about goats in Africa, then don't give it that information. Just give it the information for molds like we did, give it its emails, we put in our phone system, so it's got all of our phone calls, all of our scripts, every interview I've done, every podcast I've done, every email we've ever sent. And so Ty, which is our data lake, has everything that it has, our voice, so it sounds like us. Our marketing can go into Ty now and say, please write me an email, an EDM for the, we were in an awards night last night, congratulating all the winners for this awards night that we had addressed last night. Bang. It'll then write the email as if it was myself or my marketing team.

Building a proprietary data lake on Azure

42:17Gerard Murtagh: In our voice to do that. So the first step of anybody's transformation to AI is that data lake, because then you will be able to then plug in the things that you need. So for us, it's the scheduling, it's also the email routing and replying, it's also the marketing side. So we've got the marketing component, we've got the phone system that is within eight weeks, you'll be able to call MouldMen and speak to an AI agent, but it will be making decisions based on our data lake. So if you said, 'How did MouldMen start?' it'll tell you how MouldMen started. You can talk to our—you will be able to talk to Ty for hours on end at 2:00 in the morning and learn all about mould. It won't know anything about anything else, but it'll know everything about MouldMen, it'll know everything about mould, and you'll be able to ask questions. So some businesses will have to bring in their different operating systems, but if you've got this core brain, your developers will be able to get the API from your job management system or your operation system or your CRM, and then you'll be able to plug that in. Every email you send, every message that goes out, every phone call, just makes that brain even better. And that's the first step of that. And we've built ours on Azure, but the way that it's been built is that if Azure can't keep up or that engine can't keep up, we'll take that out. And who knows, maybe plug in Llama or we may plug in Grok or we may plug in something else. It'll all depend on that. But your data lake is your secure lake of data that has everything in it. So if you're a CEO that no one knows how to do anything inside your business, that's what you need to do. You need to clone yourself. So if you're Bill, you've got all of your—the way that you train people—your data lake, your legacy is going to be your data lake with everything that you've ever done, these 600 podcasts, everything, the way that you speak, the way you write an email. That's what is the future of this. So people say, 'When we give up business,' and I said, 'Well, I'll probably never give up on business because whoever owns my LLM or my data lake in the future will have—unfortunately, they'll have my brain.' They'll be able to—that might scare the shit out of them, but anyway.

44:38Bill Gallagher: That's a good tee-up for a little thing that I would love to sneak in and offer people. It's not yet public, but we spent a lot of time looking at a variety of different tools and did some tests with a few of them. And we've essentially built the Bill AI. And what I realized is that I have twelve years of blog posts and speeches and workshops and things like that. And almost every last one of them, I recorded and captured. And so I was able to upload more than 8 million words over that period of time to an AI and build a Bill AI that can answer scaling questions. And we have a few clients using it now, and we're using it for some of our company stuff in the back end, writing emails and posts and things like that and getting things ready. But we do—it's not yet publicly available. If you'd like to be one of the folks—if you'd like an invite—send me a request. Say, 'Let me into your AI, Bill,' and send it bill@scalingcoach.com, and I will add you to our early list. We'll give a few people some trial to it and see what you think of it. I think mostly it's really surprising what you can build with that. And we have also a voice thing. Like, it's trained. It talks. It sounds a lot like me when it—you could talk to it and ask it things, and it answers like Bill, which, I don't know, might give somebody nightmares. But yeah.

46:02Gerard Murtagh: Don't get me wrong. I've been for many years—I got through—I've been a Young Entrepreneurs' Organization, EO, for nearly twenty years. Twenty years is my twenty-year anniversary of being in EO. And one of the things I learned early on was about journaling and writing. And I've kept that up for many years, but you've got all this information sitting in a diary. It's pretty gnarly stuff. And I've got them locked away, and I'm like, my kids will get these one day. But I changed it up recently a couple of years ago, and I went to voice notes and notes. So I sit here in front of my computer, and I've got a designated phone, and I just talk to my kids. And I did it for my kids so that if anything ever happened to me, they would have all of this knowledge of these books that I've read and the podcasts that I've been on. Or I just talk to them as if I was having a conversation, and obviously, it's somewhat one-sided at this point. But I'm hoping that in the future that if anything does happen to me, unfortunately—and my dad passed away at 58, and I've got, I think, maybe ten minutes of him talking—where my kids are now going to have hours of my voice, which for them potentially will help them through their life. Because I'd love to call my dad and say, 'Dad, my old man would have loved this time of humanity. He would have just—this would have been his jam.' And so that's what sort of forced me to do it. And so I encourage people to do that regardless, because as you—you can do this for business, but you can also do this for your family, you can do this for your grandkids, and your legacy can sort of live on. Because we learn so much, but there's nothing better than—like, I got to—my grandmother lived into her nineties, and she was so wise. And I loved my time with her because I didn't have my father to sort of help me through some of these challenges in my life or in my adult life. So I know it's a bit off track, Bill, but it's not just about our business acumen and the things that we've done in business, it's also our ability to help our friends and our families and the kids coming. Yeah.

Why waiting for vendors means losing to competitors

48:12Bill Gallagher: The one that I've trained is really all of my business work. It's not all the little things, but we think about that—that our friends and family who want to leave a legacy, and there's that whole personal potential as well. It's just getting better at the whole thing, at creating something that's useful so that you can remember. And that's obviously a very poignant experience for you to lose your dad at such an early age and really feel like you wanted more. You wanted him around for longer.

48:44Bill Gallagher: Yeah. So we've talked for a long time. I think we've given people a lot of things to think about. There are tools, there are partners, there are things. And it's not just getting your marketing right or banging out another blog post. It's a whole range of it. In our world, strategy and planning and a whole lot of other things are getting done today behind the scenes with AI. We're getting more things done faster and up-leveling the quality of it. But in your business, not only the financial analysis—

49:13Bill Gallagher: The recruiting, the whole customer flow is all getting automated more and more at a faster pace. You've got a pretty big business to only have a hundred people working on it, and it's really an inspiring thing. I'm really glad to bring this to people to get them started thinking about maybe we could get it done and put some stuff in a lot faster, right?

49:37Gerard Murtagh: Yeah, look, this is an arms race, and I sort of mentioned it a little bit earlier on. Think about your big ten-ton gorilla competitor that you might be in your industry right now. They're investing millions of dollars into AI because they know that that's going to be their competitive advantage. I did a speech recently, and there were 500 people in the room, and I said, "Who is expecting in the next two years to invest in AI?" And only 20% put their hand up. 80% said they weren't interested. And then the next question I asked was, "Who in the room is expecting their technology providers to build the AI for them?" And they all put their hand up. Everybody put their hand up and said, "I'm expecting, you know, HubSpot or our job management system or email or Microsoft or someone to just come up with AI." And that just makes—there's no point of difference if everybody's got the same technology. Everybody's got it. So what the point of difference is going to be is going to be your data and the way you run your business. And that was the big eye-opener to me, is that people are still sitting back and saying, "It's going to be okay. I'll just wait for them." But what I'm really pushing people to think about is if your biggest competitor is investing so much money into this and you're waiting for someone else, you will end up with the crumbs. You will end up with the worst clients, the worst paying ones, the ones you don't want, the ones that you learn at Scaling Up with EO and all these other different people say you need to go after these high-value clients, the ones that pay you on time, the ones with the big ticket items. They're going to be going with the guys that can book a job within fifteen minutes or fifteen seconds, or that you don't have to speak to a human because you can do it all on your phone. Like, for instance, in the future, you will get onto either MouldMen's website or you'll call us, and before you either put your phone down or stop talking to Ty, you will have got your quote and booked in your job, and we'll probably have your credit card number. And then you'll think, "Wow, I've just booked in the job. I think I need to get two or three more quotes." You'll call the competitor and they'll say, "I'll come out next week and inspect." And you've got—MouldMen's already told you that they're going to be out there tomorrow, and it's going to cost $400, and it's all going to be resolved. What are you going to go with? So that's just a real-life case that MouldMen—we're right there where it's not fully turned on. There's lots of human at the moment, human in the loop. But there is a ten-ton gorilla that I'm worried about in trades building this before I do. So if you're in your industry and you feel like you can get somebody from first contact to giving you their credit card in fifteen minutes, regardless of whatever it is, then that's what you need to be focused on. Because if you're not, someone else will, and then you'll just be getting the scraps.

52:27Bill Gallagher: All right, our show is running long. We've given people a ton of things to think about. Let me just point them to a couple of next steps and good calls to action. So one, if you want to get in touch with Gerard, we'll put up a link for him, gerardmurtagh.com. It's on the screen if you're watching. It's also in the show notes. You'll be able to find it and check him out and get in touch with him if you want to know that. Let me just say, from a lot of companies that I am currently working with and recently working with, there's a lot of nervousness about doing stuff with your data, and you probably have a lot more data than you realize. Some of you have decades of transactional information, shipping information, and there are patterns there. We can create lookalike audiences. We can figure out where to go and who to do and what the best ones are. We can analyze financials. We can analyze transactions. We can look at your email. We can look at your voicemail. There's a whole bunch of stuff, probably more than you realize that you have that could be gleaned, and you could start to build something. But get started. It's not all done yet, but don't be the last one in figuring out how to do it or wait for your software vendor to give you something that everybody has and you have no specific real advantage beyond anything else. So those are some great things. If you'd like to try our little Scaling Up, or my Bill Gallagher Scaling Coach AI, send me an email, bill@scalingcoach.com, and we'll put you in the next batch of users that we let in to play with that thing and do it. Let me give a big shout-out and thanks to my friend and mentor, Verne Harnish. Gerard and I met together with Verne at our Harvard CEO annual summit, which is again late August. If you'd like an invite to that, you're a CEO, you're working with us now or a regular fan of the show, send me an email also, and I will—that will sell out soon. It's not sold out yet, but that's a really great conference. And then, of course, thanks to Verne Harnish for starting this whole thing with the Scaling Up framework and the EO organization that we've been around for so many years. Thanks again to the folks at Story On who get our show produced in and out every week—Seth, Christelle, Cameron, and the team over there. It's a lot of work to do this on a weekly basis for more than 600 episodes, so really appreciate our team, wherever they are in all realms. And thanks again to all of you for listening, for watching. Until next time, keep scaling.

55:11Bill Gallagher: Thanks for listening today. One last thing. If anything in this episode hit home, my book digs into it further. Busy Is Broken. Do Less. Scale More. It's all about how to stop drowning in work and build a business and a team that scales without you. Available right now with content samples at busyisbroken.com. Go grab it. Be less busy.