Podcast: Play in new window | Download
Thank you for reading this post, don't forget to subscribe!
Subscribe: Apple Podcasts | Spotify | Amazon Music | Android | Pandora | iHeartRadio | Podchaser | Email | TuneIn | Deezer | RSS | More
Episode 515 – The Language of Procurement Isn’t the Language of Utilization: Rethinking AI for Your Business
Introduction
Hey, everybody. If you’re a business leader out there and you’re wrestling with our old friend AI, our new old friend AI, the question has probably run full circle in your own mind, your own organization. But today we’re going to dive into a slightly different angle and of the strategy and thinking. My guest is someone who has a lot of experience in this space. His name is Maximos Lih. Maximos, welcome to the show, my friend.
Well, welcome. Glad to be here, Doug. Appreciate it.
Yeah. When we got together, I was intrigued. What? One of the things that really caught my eye. First, let me say, Maximos does a lot of work for privately owned businesses. The entrepreneurial space helping owners scale up beyond that painful startup phase and get out of the Chief Everything officer mode into building a team that can execute for you. But in the ever present challenge that’s out there, the question is, if you’re a business owner, what do I do with AI? What should I be thinking? What do I need to do? So Maximos, let’s start with kind of the I guess, what you might call philosophy or strategy that people should be considering.
Meet the Guest: A Decade Inside Google Ventures
Yeah, I think so. Well, let me preface this just by showing a little bit about how I come to my understanding of the AI ecosystem. So I really got my deep dive into technology companies through almost ten years of working as an operating partner at Google Ventures. So what does that mean? Google ventures is just a part of Google or I guess, alphabet now that writes, checks and invests in startups. We invest purely for returns, not to buy companies for Google, and I’ve seen everything from Blue Bottle go from two cafes to, you know, the acquisition and to Nestle to Stripe. I saw Uber go from 80 to 500 people in six months, from four markets to 20 markets in six months. And so, you know, stripe wasn’t the portfolio. Slack was in the portfolio. So, you know, if you think about when I started at Google Ventures, we had about 80 companies, startups in the portfolio. By the time that I left we had 400 plus active companies. So it was really a great time in my career to kind of get a lot of data points at a very fast pace, especially for companies that are growing from 20 people through 150 and then to 500 and then beyond.
So I got a lot of my, you know, sort of framework and knowledge and analytics about what is the pattern matching of the companies that make it from a small team to an even better team, to the companies that really struggle to do it, the companies who don’t make it at all. And I left TV right around the time of Covid because I realized that people, leadership, HR as an industry was being reinvented. And I was like, I kind of have an itch to be operational. I don’t want to be on the sidelines. I’d also really suffered some trauma from helping companies do layoffs during that time, because they had not built their teams properly, and they had over hired and they had below. And then it was such a damaging thing emotionally. And the way that they were doing it was also quite nonhuman first. Right. You’ve heard the horror stories of, oh, I found out I no longer had a job when I couldn’t log into my email because I missed the one East Coast, you know, zoom call.
As somebody who was living in Hong Kong. It was such a compassionate time, all in the name of efficiency. And then fast forward a couple of years, AI comes into the market, and there’s another wave of extremely non-human force in on compassionate utilization of these tools that as I looked in and study the patterns around, it was like nothing about this actually makes a lot of sense if you actually understand what the tool is trying to do.
Cars, Horses, and Horsepower: The Analogy That Reframes AI
And so with the story that I often tell, people make the analogy all the time of AI is like when the automobile was first created, or AI was like the new industrial age. But if you actually study when the automobile first came to market, something that people often overlook was that in the early days, the automobile factories or Ford would send salespeople out to farms and ranches, and they would say to these ranch owners, hey, I have these things called cars. They’re going to change your life. And if you have cars, you’re not going to need horses anymore.
And the farmers would say, okay, how many horses are we talking about? Which is why the salespeople invented the language of horsepower, an obsolete term that we still use today, because that they can go back to those farmers and say, if you buy this car that equals ten horses, you buy this car equals 100 horses. But if you were a farmer and you used to have 20 horses that took, you know, five wagons to five different markets and you bought one single, 20 horsepower car. You you can’t go all the places I used to go anymore. Right. So the language of what it takes to buy a car, the language of procurement, doesn’t automatically translate to the language of how you use the car properly. The language of routes. Right. So if you are a farmer who used to think in terms of wagons and you bought a car thinking you could replace cars and wagons, you either overpay, right? I bought five cars to replace five wagons and I don’t have horses anymore, but that’s very expensive.
Or you buy one car and you don’t think in terms of routes, and then you have rotted carrots in your car because it couldn’t go all the places at once. Many of us who are engaged in AI are familiar with the very famous or infamous MIT report that said, up to $4 billion have been spent in the last two years on AI pilots. Of these corporate companies, 95% of which generated zero ROI. Not like neutral, but zero. And they’re like, I don’t even know how this could possibly have happened when I use ChatGPT or Gemini for my own stuff. It seems to be very helpful what’s going on in a large scale? And it’s like, yeah, because the language of procurement is not the language of utilization. And if you have wagons and you need to think in terms of routes, but you haven’t taught anybody to think in terms of routes, you end up with a lot of waste or a lot of overpayment. And so I think philosophically we need to understand this is where we are, right? The sale of how you buy these tools and the cost of it is all horsepower.
The Internet Is Now Agents
Language horsepower doesn’t translate automatically to words. So then the second thing that I think is interesting for us to know is that as of this month, I think the reports have shown 57 to 60% of internet traffic as we know it is bots. So if you were one of the people who thought, I’ll just wait and see and I’ll follow somebody who’s figured it out, well, the landscape has changed underneath our feet. The market hasn’t waited for us. The internet is now agents. So if you were used to doing marketing or sales or advertising SEO based off of human based clicks, 65% of those clicks are now bots, my friend. You’re just going to fall into scams. So we can’t not use AI. But how do we actually flip into that language of utilization? And that’s actually where I think a lot of leadership and organizational behavior is lagging. And, you know, I think philosophically we just need to understand that that’s actually where the gate is. It’s actually very easy to buy ChatGPT or Gemini and give everybody a $500 expense account to use it.
The companies that I know who do that either end up with a lot of hallucinations and therefore increase quality control, right. Or they’ve just scaled all their dysfunction. So you can’t just democratically share $500. You know, like spread, like spread and spread. Kind of like idea for the way that you’re using AI, because the language of procurement is not the language of utilization. And utilization is something fundamentally different in a post agent internet.
Why Tech Companies Struggle Most: IP and Judgment
Yeah. What I found interesting I have clients spread across a wide spectrum of fortune 500 companies, and what I find interesting, this is just one man’s opinion. I want to make that clear. And it’s it’s totally anecdotal, but the companies that seem to be struggling most with the proper deployment and return on investment for AI are the technology companies.
That can be. Yeah, yeah, I would say the companies that are doing really well with regards to AI utilization, other companies who are very clear about which humans in their organization are holding IP and which humans hold judgment, which means that if you are a 20 to 50 person company and everybody feels a little bit like family or average tenure is 4 to 7 years.
You actually have a pretty good grasp of how to use AI, because you know how your people systems are working in a healthy way. If you’re 100 to 500 or if you’re changing your business revenue engine, it’s very difficult to identify IP and human judgment. If you are a 100 to 5000 company where you have previously relied on middle management to sort of keep people in line, you have no clue where the judgment sits in your company. And those are the ones that are scaling dysfunction more than anything else. Or those are the people who are just cutting out humans in the name of efficiency. And my conviction is that in the next three years, efficiency will just be commoditized. Everybody will buy the same tools, but you’ll confuse productivity with activity with impact.
Yeah, I think that’s a great way to put it. And that’s the ultimate measure of ROI is, you know, what is the actual productivity? That it’s not activity for the sake of activity, but it’s productivity that adds value to your customer and ultimately to your own bottom line.
Compensation in the Age of AI Agents
Yeah, absolutely. I mean, I think one thing that is really interesting we’re seeing right now in the marketplace is how does AI and AI utilization, especially in a AI, can replace headcount capacity, affect compensation? So we know that there’s sort of these hypotheses. And I think it’s true. You might need fewer engineers to power any company because AI can do a lot of coding for you now. But the few engineers that you have will be very, very expensive. And I think that’s fair. But how do you do the math around that? All right. So if I was a company where, I mean, I have a client now, $33 million are seven employees, right? Every employee runs a team of AI agents with a supervisor. You know, harness and $33 million. Seven employees. Right. Do I pay each one of those employees? Ten people? Five people. Three people. I wouldn’t pay them. Just the top of the market. As if they were an individual contributor.
Because they are not truly an individual contributor. Their roles are significantly above that, do I? Am I able to copyright the things that they put in their suite of agents so that they couldn’t leave my company, start a consultancy, or go to a competitor? Technically, but we saw this in Google during the ads days. You could copyright, you could IP protect, you could personal proprietary information protect a lot of things that are built in the system. But the algorithm thinking that’s within an engineer’s brain, you can’t protect against that. And so Google would say, oh, we’re going to benchmark our comp against 100th percentile of market, because it’s better for us to keep them and throw $100,000 per year sign on bonus than to let them go to Bing or Alibaba because that talent is so valuable. Well, in an Ads ecosystem, you could benchmark against 99 percentile because the data centers that power the ads only sit inside of Google. But if I’m doing it with agents, I no longer have the same protection of the databases that are locked within Google.
So how do I do that properly as especially as a small team? And how do I actually change my compensation framework from there? That’s a really interesting question that I think we’ll see in some changes within the next 3 or 5 years.
Well. And not to mention the challenge that’s going on, even in the legal community, about copyrights and intellectual property protection relative to work product produced by AI. That’s like who really owns that? Who really did the work and what you know. What argument can you make? And yeah, it has if anything, it’s really muddied the space between universal knowledge and true intellectual creation. Intellectual property rights.
Hiring in a World Where Everyone Uses AI
Gosh. Yeah. Well, and it’s fundamentally changing so many things. I mean, compensation is always related to performance, but also in hiring, because we are really the first generation of hiring managers to be interviewing people who are expected to use AI. Right? I think it’s just going to be more and more taken for granted that any piece of paper that you write would have gone through some sort of AI brainstorming writing layer, so you could produce a lot of things that look legit without any deep understanding of it, and therefore have no capability to fix it properly when some or catch the hallucinations as they come up.
So whereas I used to interview people and say, oh, produce something for me. You know, give me an analyst report, give me a business insights report and then do a presentation on it. I have to add two layers deeper of comprehension to that interview because I assumed that they were using they must have referenced, you know, not just Google but Gemini or, you know, tragedy or any of the tools out there to generate that. And how much of that did they really understand and how much of it did they just nod and let through? Right. Because it sounded pretty legit that then I get, you know, sort of tempted into, oh, they must be a specialist or they must be an expert because they use the AI tool as well, but they didn’t understand the content.
Yeah. Well, the whole question of accurate presentation at an interview is, is an age old problem. I mean, even way before anybody was thinking about AI, I had experience in an industry that was packed with acronyms and abbreviations and terminology, and people could study enough to connect all the dots with a stream of those terms to sound like they knew what they were doing, but would in fact have no idea of how they really get there.
And I’ve told this story before. At one of the companies I founded, I finally had to implement a practical test. I gave people a piece of work during the interview and said, you know, go in the conference room, you can work on this and let me know when you’re done. And the way the test was designed, a good expert in that field could accomplish my task in about 15 minutes.
Nice.
15 to 20, something like that. But if they went shorter, they they gave up and said, no, I can’t do this. Or if they went longer, it was proof that they generally had no idea what they were doing, because someone that knew the work could rapidly connect the dots in my test and give me the output I wanted. And in the test case, I had, you know, had some errors in it that needed to be identified properly by the analysts that was doing the work. And if it wasn’t a perfect, pristine example of the work, and that was part of the job is to identify those errors and call them out appropriately and stack everything in the right order.
And it was my hiring efficiency went up, you know, probably tenfold when I implemented that, because it was real easy, real fast, and even on the phone screen. And when I would tell people we were going to do that test, a lot of people would self-select and drop out. Yeah, I’d never, never come for the interview because I knew they knew they didn’t have it.
So I’m so I’m so glad we were talking about this, because I do think that’s one of the things where I’m very, very clear with my clients. Do not let AI conduct your interviews on your behalf. That is the one place where human judgment determines a lot. It’s very expensive to hire a higher morale, you know, productivity, all the things. Don’t outsource that. Do not outsource the human judgment. And yeah, I mean, if you’re if you’re not talking to people who have jobs that are passive candidates, I think that kind of on site is really, really valuable. If you don’t have the luxury of bringing people on for an hour on site to do something like that, then at least throw a wrench in the problem when they’re talking back to you.
Right. And they just feel like I’ve changed the constraints. What will you do if we had to do this on Mars? Right. What would you do if we had to do this in India so that you could see them process, you know, synthesizing information and driving complex solutions and talking through their measure of comprehension on site. But you need something like that right now. It’s just too easy to cheat. And we know that the more people that are using AI, the more people will use AI inappropriately, the more hallucinations. There’s just a quality control piece built into the tool because of how it’s designed.
AI for Hands-On Businesses: Trades, Home Services, and Manufacturing
Yeah. Let me ask you this. For companies that are more oriented to, labor and I mean hands on the implement kind of labor and like, for instance, home service businesses, plumbers, electricians, HVAC, or even certain types of manufacturing. What do you see as the opportunity in those kinds of businesses to apply some AI for productivity, but probably not replace the boots on the ground, you know, going into the home or, standing at the machine on the work floor.
Yeah. Well, I think that one thing that I can see happening very quickly because of AI coding mechanisms, is that small companies that used to have to pay a large amount of money to get Shopify or to get Salesforce CRM type tracking workflows, and then sort of make the larger systems work for their own workflow. You can actually now within three months and for 8 to $10,000, create your own instance of a CRM that works specifically for your company’s workflow, where you own all of the data and does not come with a subscription. So I definitely think that that’s an area where you’re going to see, you know, if I’m doing a Home services, I don’t get that much advantage from knowing how other home service companies are doing it or how other Salesforce customers are doing it. So the data layer that may come back to me as a cohort isn’t as useful as having something that I don’t have to massage to, like, do it the way that I want to do it. And you can do that very easily and you don’t have to give your data away.
You get to own your own data for training. So I think that’s one instance. The second thing that I’m happening I see happening, particularly with respect to chip manufacturing, is, the testing interface that used to be very costly or very specialized at the end of a cycle. You could move a lot of that upfront. So I know, you know, clients that are overseas, for example, in the chip manufacturing space, saying, I would be so much more innovative against my flight to Nvidia if I knew for sure that this thing was not going to fail 15% of the time in fab. So what can you do to be able to simulate that? And they’re like, well, I could literally run these things from a physics, you know, electrical level, you know, as a as an infrastructure layer, as your architecture scoping and designing the architecture. So we could be running these tests in real time. And that actually produces a lot of work. Hey, you’re doing machine technology technicians that used to go to, you know, each one of these things that I failed and say, as I understand it from the client that I was working with, my first solution is what is common and cheap, and that works 75% of the time.
And if that doesn’t work, then there’s seven different things at increasing levels of expense. And like a 3000 page manual that I have to work through and, you know, make a document of, it’s a very, you know, like you just hope that it doesn’t hit the 75%. Right. Like everything hits in that point. But the other percentage is very long. It’s very slow and very costly. Well, what if I could synthesize that information and give you the top three, and then you could try those things all at the same time? Right. So then that shortens any time, any downtime on the factory floor without actually replacing the technician. Because once again, there’s a lot of things around human IP and human judgment where a good mechanic will look at something that’s failed, like for your car and be like 72. Chrysler is probably this. And they have an institutional knowledge that even if you were to put into a textbook, the feel of it is just not the same. So I don’t want to take that human judgment away.
But why default to the 70% cheapest option, when I could probably use AI to derive the top three most likely options? And then the complex things, the human judgment really comes in and they spend most of the time on those escalations.
Knowing Where to Whack It: The Value of Human Judgment
Yeah. Talking about the early days of the automobile, you know, there’s a famous story about one of the machines in the Ford factory had broken in. Ford called a guy in to look at it, and he tinkered around a little bit and then took a hammer and whacked it in one spot and sent Ford a bill for $10,000. And the guy said, you know why so much? You know, you just you gave it one whack. And he. So the guy restated the bill, you know, all right. For the one whack $10.
But knowing what about.
990 knowing where to whack it, you.
Know. Yeah.
That’s right. Absolutely.
So you know, well, well, this is clearly a conundrum for a lot of people. And I know firsthand there are people out there that particularly in the smaller business community who are looking at this and just saying, no, I’m not even going to try.
I, I’m going to just I, I have a model. I, I know what it does for me. I know what I can do and you know, I will soldier on with it. I, I have struggled trying to be as objective as I can. I have struggled to find a compelling argument to tell those guys they ought to consider something. Do you have a thought on that?
The Things That Only You Can Do
I have clients that are in the nonprofit space, and there are certain nonprofit spaces where any use of AI is highly stigmatized, right? You didn’t do your own work, you know. Are you sure? You know, that feels really dangerous or somehow right unproductive to be using those tools. And I feel bad for those companies because at the same time, you are also organizations that don’t have a huge bank account to be hiring a lot of people. So most of your functional leads are departments of 1 or 2 people, and they do everything from board level work to janitorial work. Right? And from my perspective, I would just say, you know, one, the leadership question that I always champion is know the difference between everything you can do and the things that only you can do, right? Yes.
As a CEO, if people saw me cleaning the bathrooms, it would show so much to news like humility and, you know, grit and gumption. But that’s an hour cleaning the bathroom that is taking away from all the things that only you can do, like fundraising or understanding policy or talking to donors. So what are those things that you currently can’t get away from that is actually not moving your job forward, that you wish you had money to hire an assistant or an intern or minion to do? Those things are almost always possible to do with AI to do right? Hey, I know all of the content of what I want to share at the board meeting, but I suck at making presentations. I will do that for you now. And like pretty credible, right? It doesn’t have to write the content for you. Speak it into a voice note and say, I want the PowerPoint to be red and blue. I want it to look professional. I don’t want more than eight, you know, pictures, eight words per slide.
And I want lots of graphics. And it will make something for you that you used to have to hire an intern to do and like, even if you were just doing that and it gave you back two hours of your time from not stressing over a presentation, that’s two hours that you could spend on doing something that only you can do, right. So once again, I feel like, look, the goal is not how is AI going to replace humans? But how does AI as a tool give back time for humans to do the things that only humans can do? And and I think that that is always an interesting place to start, because there are things about your job that do not contribute to the things that only you can do, may not even feel like things that only humans can do. And that that’s kind of where I would start. Just do the cost benefit analysis of not being able to hire an assistant.
Leading AI Like You Lead a Team: Context Over Task
Yeah.
You know, one thing I’ve found interesting, I’ve my whole life I’ve been an early adopter.
So when the doors to AI opened up, I was right at the front of the line. And and I’ve kind of ridden the right wave of all the different rollouts and phases so far, and a core concept that I find still prevails. And I don’t know that it’s going to change. Maybe it will, but it’s not too different from what a good leader can do leading a team. And what I mean by this is if you think totally in the leadership mode for a minute, so I’ve got a team, I’ve got a task, maybe a project that I need to get going. I will call the team together and I will give assignments to the team. You know, Maximos, you’re a good analyst. I want you to handle the analytic part. But if I’m a good leader trying to clearly communicate, I’m not just going to give you the task. I’m going to give you a little context about the the environment, the maybe the tie it to the history of the company or the history of a product we’ve got or some functional aspect of a product product that this project is trying to expand.
I’m going to paint that picture with not just task giving but context giving. And that’s what I’ve learned AI needs the AI model I choose to use. Not only do I need to, when I’m speaking the prompt, I need to either create some institutional contexts, like I need to spend a little bit of time training it to know what my business is about and how I like to communicate, how I like to speak to things. And then at that point I can give it a task, but I’ll specifically tie it to one part of the context. And when I do that, I get good, meaningful output.
Yeah, absolutely. I think that one of the things that is so important at this time is you have to know the you have to be a leader who knows the people that you’re leading and the tools that you’re trying to give them in order to lead it. Well. Right. If we go back to my wagons and roots analogy, used to be really good at working with horses. You’ve never worked with cars before.
You think they’re just going to swap one on one and everything is going to be great? No. Right. You have to do some work yourself to be a good leader of now, a hybrid fleet or a fleet with a totally different animal, which is now a mechanical one. And so I think that, you know, exactly to your point, the companies that are not using AI well think that cars will behave like horses are wagons and they always fail. It’s a very expensive fail. Great leaders think not only in terms of oh, how does a car behave differently from a horse? But maybe I don’t want to be in the farming business anymore. You’re in Texas. Maybe I need to get into building gas stations now. I’m doing something that’s completely different, that’s going to generate so much more revenue, generational wealth, because I saw the trend that’s coming. And instead of saying, well, let me see how other farms are going to use their cars and I’ll just follow their template, which is essentially saying, I don’t use any judgment for myself.
I’m just going to copy somebody else’s template. Right? I start looking at, oh, how what is a leadership moment here that actually allows me to do something that would have been impossible six months ago. And that’s a that’s a different kind of root space thinking. Right. And those are the companies that will really make it to the next generation of wherever we’re going as a market, you know, a generation ahead because they they thought, well, like if there’s going to be cars going all through here, you know what? I’m not buying any cars. I’m going to keep all my horses. I’m going to furiously look for gas in my area. Right. And then I’ll have my writers go to a farm that used to not be able to make it to markets that are days, you know, right. For me, because it was too far away. And tell them you’re going to be able to access different markets because my gas station is here and now, that’s a totally different kind of business.
You haven’t lost any horses, right? But it takes leadership and strength, understanding. Right. And like context assessment to really be able to make some of those bets based off of your conviction.
Let Everyone Else Get the Arrows: A Lesson From the ATM Era
Yeah.
You know. I haven’t quite put a name on where I stand with the whole advent of AI other than, like I said a moment ago. I’ve always been an early adopter, so I’m a fan, I’m curious, I’m willing to learn, etc. but you’re right, from a leadership standpoint, it’s very important to be strategic in your thinking about the new thing. And that reminds me of a chapter in my banking career. I’m old enough. I was there when we did not have ATMs and and everybody, all the banks were rushing. When the new technology came out. All the banks were rushing to implement, and I, I was in the room when our CEO was giving a presentation about what we should do as a bank to implement these ATMs, and it was a classic stuff, you know, and I won’t go into all the details.
Just suffice it to say it was a typical, you know, new product rollout kind of idea. And after the whole presentation was done, the CEO shook his head and he goes, not now. He said, no, we’re not going to do it. And everybody in the room kind of went. You got to be kidding, boss. You know, we’re going to fall so far behind and.
Yeah.
And all that. Well in in two years later. And he actually spoke the words, he said, I want to let everybody else get the arrows in their back.
Sure. Yeah.
And two years later, what was revealed was the fact that the convenience of the ATM needed a network. It wasn’t sufficient for that device to be wired to the home bank.
Yep.
People wanted the flexibility of. Well, I’m in a town that doesn’t have your brand, but I need cash. How do I get it? There needed to be a network of these things. And our CEO stood up and basically designed the pulse network.
That was one of the brands that was used in the early days. And we became we built this public utility of networking across institutions. And there was, you know, there was a toll to be paid for being on that network. And we made millions and millions of dollars being a founder of that network. And it was way different from just having the box in our lobby.
Yeah, yeah.
I mean, I could not agree more. And thanks for sharing that story, because that’s the difference between people buying tools for efficiency. Right. Hey, let me install these things and then lay off 30% of tellers. It’s a 1 to 1 swap. It’s just going to be commoditized like the machines are commoditized today. Right. But what is the roots version of that? What is the network infrastructure version of that. Well, there isn’t an answer. I don’t see it right now. Great. So I’m not going to put money into efficiency and lay off humans until I actually have that, you know, properly in mind and then be really quick to implement when you do them right.
Like that is that’s a leadership right there.
Yeah. Yeah.
Well and that’s what we did. And as they say the rest is history.
I love that.
The IKEA Story: Turning AI Failure Into a New Business Unit
Well Maximos, I think we’re about up on time for today’s episode my friend. Thank you so much for sharing. Tell people the best way to get in touch with you if they’re interested in knowing more.
Yeah, I’m on LinkedIn. I read about these kinds of transitions. I’m always happy to take the discovery calls with folks that are thinking through it. As I mentioned, I just think there’s so much opportunity for leadership teams to actually rally around and build those kinds of systems, right. That really set you up in a way that still feels human first, but it changes. It’s a very transformative to the way that somebody is doing business. I’ll give one more example before we leave, and I wish this is on US company because actually Ikea, Ikea was one of those companies that was such a beautiful example of this kind of pulse based network thinking.
They had rolled out an AI chatbot for customer service. It did very well and started to handle, you know, 50, 60% of customer calls, I think maybe as high as 70%. And because it was working really well, you know, hundreds of customer service agents were now on the chopping block as a call center. But they had a VP of data and a VP of LMD, and I’m sure other leaders as well, who basically started to ask the question, wait, what is the AI failure telling us about our customers? Right. I know that the AI success means that there are people asking basic, easy questions. What is a failure telling us? And without needing too much time on research, they figure it out that those 30% 40% of customers were looking for. How do I make my home comfortable? I have kids, I’m buying children’s equipment for the first time. What’s the best way to decorate my house so that it can be still clean, but still have, you know, a kids play station? They wanted somebody with taste and empathy and compassion.
And so instead of laying off any of the customer service folks, they said, why don’t we reskill them to do remote interior design? We’ve always thought that that would be interesting. We never really had the bandwidth to do it. I had these employees with a lot of IP in them, right? They already know our products. They know the skew. They like talking to people. They understand taste. Let’s just train them to do interior design. Created an entire business unit. The first year it contributed 3% of global revenue targeted to be 10% by the year 2028. Right. And it’s the same thing with the pulse, right. If you’re asking how that makes me efficient, you’re going to end up with a very commoditized answer. Everybody’s got a box. But if you’re asking what is AI going to allow us to do six months from now, that’s impossible today that 10% of global revenue. And I’m sure pulse was more than 10% of global revenue for your bank, because I was a completely impossible thing.
And you guys had had the leadership foresight to be able to think about that. And like, you can’t do that without being human first. so.
Closing Thoughts
Well. Very good. Maximos. Great point. And again, thank you so much for sitting in. Folks, we’re going to have his contact info and everything in our show notes as always. And I do want to remind you, we have a video version of this show on YouTube channel by the same name, leadership powered by common sense. If you need to get Ahold of me, drop over to my website Doug thorpe.com. That’s Thorpe with an E, and I would love to hear from you. Leave me a comment, leave me a note, let me know and don’t send your bots to my website. Come on over on your own, please and do it. But for now, we’re going to sign off, say goodbye. Go out there, make it a great day.
Meet Our Guest – Maximos Lih
Maximos Lih, an executive coach and founder of Emboldened LLC, who helps early-stage leaders scale their teams with intention, without losing their culture, clarity, and edge!
Drawing from his own family’s legacy, his grandfather was an orphan-turned-four-star-general in China, Maximos learned early that greatness isn’t about titles, but about how you make people feel. After graduating from UC Berkeley and rising through the ranks at Google and Google Ventures, Maximos spent nearly a decade helping over 400 startups scale both product and leadership. He’s seen firsthand what it takes to go from scrappy team to billion-dollar exit, and why so many founders sabotage their own success without realizing it.
Maximos now works with small teams, startups, nonprofits, etc, to build what he calls Olympic Teams: mission-driven, rare-talent teams that need to be coached and led differently than traditional organizations. His insights are sharp, practical, and often paradigm-shifting for founders and leaders navigating scale, AI, and generational change.
Connect with Doug Thorpe
- LinkedIn: https://www.linkedin.com/in/dougthorpe/
- Doug Thorpe Website: https://dougthorpe.com/
- Podcasts: https://podcasts.dougthorpe.com
- Twitter: https://twitter.com/dougthorpe_com
- Facebook: https://facebook.com/headwayexec
- Instagram: https://instagram.com/dougthorpe_com
- YouTube: https://youtube.com/@dougthorpe
