Scale Differently in the Age of AI with Robert Duffy, CTO at HealthEdge

(Intro Narrator at 00:00:00) Today, we're talking to Robert Duffy, CTO at HealthEdge, about how he thinks about scaling in the age of AI and about HealthEdge's latest news. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:17) Big news has happened, though. I saw a little something in the news cycle there. Tell me about that.

(Robert Duffy at 00:00:22) So, yeah, we transacted from Blackstone to Bain Capital, and Bain Capital has an incredible thesis about bringing us and another company called Healthproof together. So we recently closed on the transaction with Healthproof, and we're now one merged company, HealthEdge and Healthproof, and that's going to be amazing. HealthEdge runs the software that does the claims back office, and Healthproof operates a business process as a service, which is basically running the whole claims operations for healthcare payers. So bringing these two teams together, we can really focus on driving outcomes at a reasonable cost structure for health insurance companies. Great merger. It's going to be amazing.

(Joel Beasley at 00:01:07) Oh, that sounds exciting. I'm a big fan of you. After our last conversation, I've told like 10 people about what you were doing with the AI stuff and how you tested it across your teams and all that.

(Robert Duffy at 00:01:17) That's amazing. It is truly amazing.

(Joel Beasley at 00:01:19) Yeah.

(Robert Duffy at 00:01:20) And, like, you know, I think—

(Joel Beasley at 00:01:21) You're leading the way. You're in your niche over there doing your thing, but a lot of people are asking these questions and are curious about how you're doing it.

(Robert Duffy at 00:01:28) Yeah. And I've come to the realization that actually what you have to do is just drive usage of tools. Like, you have to really get people to use tools aggressively. And it's not just about making them available. It's actually about thinking every day about how you can get more and more people using the tool. And I think, for me, the best way to do that is to use tools yourself. Right? So if you're using tools and you know how to do things, when someone's showing you a problem or giving you a status update on a project or explaining how they're going to approach something, you can say, "Well, hey, look. Let's just try this in this tool together. Let's try and pull up Claude Code or Amazon Q, now Kiro from AWS, and let's try it. Let's do a quick proof of concept." And you have to be—I think you have to be able to do that. I'm coming to the realization that to lead through this change, you have to figure out how to drive these tools yourselves. And they're different, right? They're just different and difficult. And I don't think enough people in leadership positions are actually learning how to use tools and writing code for side projects or just creating things or sitting in meetings and trying out different things, and I think you need to.

(Joel Beasley at 00:02:46) Yeah. Create that environment, and then that's the first part is getting people to use it. The second part is a meaningful way to figure out if the way that they're using it is good enough to be consolidated into its own service or, at a minimum, append a post to other people in the department on how they're improving their job. Right?

(Robert Duffy at 00:03:06) Yeah. Exactly. And then the other thing that I think is kind of interesting about AI—a lot of people are going through this traditional product cycle and project intake process where they've got an AI council and, you know, you present the AI council with, "Here's my project, here's what I anticipate the ROI to be, here is the investment, here's how we're going to measure success," and then the AI Council sort of says, "Okay, this is an approved project, this is not an approved project." I don't think we know enough to really estimate the ROI accurately. And I think you have to open the cork on the AI project bottle and let a few projects happen first, and then worry about how you're going to measure and which ones are successful. Because we've had things where we're like, "This is the perfect use case for AI. Watch. I'm going to put this into this LLM. It's going to crunch it, and then it's going to do this in two minutes," and the output has been garbage. Like, we're like, "This is never going to work." And then we've had other things where it's like, "There's no way an LLM is ever going to do this. This is so complex and requires so much human reasoning, but let's just try it anyway." And then we try it and we're like, "Wow. This is amazing." So I think we're just—I think we don't have enough pattern recognition yet as leaders and as managers of projects to understand what's going to work and what's not. And I think the only antidote to that is to just get enough buy-in from the business that you're going to try lots of things—not forever, but for a period of time, you're going to try lots of things—and then you're going to get some pattern recognition about what's working and what's not.

(Joel Beasley at 00:04:57) Is that a conversation you had with your peers?

(Robert Duffy at 00:05:00) Yeah. Like, we did the pilot that we talked about last time. Right? Like, that was basically, like, we're going to time box it, and we're going to go all in on AI across the software development cycle. We have enough instinctual data and enough anecdotes around there and enough industry papers and research. We believe that we can get some value here, but let's time box it in that short period of time. Restrict the focus to 50 people, give them the right tools, and then just see what happens if they could go all in and then extrapolate that out across the whole organization as a follow-up.

(Joel Beasley at 00:05:34) That's so cool. After our interview last time, I called my best friend, Derek, and I was like, "Dude, you won't believe what this guy is doing." And I was like, "Do you want to make a SaaS app that helps other companies do that?" He's like, "No." And I said, "Okay. Great."

(Robert Duffy at 00:05:51) Is he working on his own projects?

(Joel Beasley at 00:05:52) The tickle machine?

(Joel Beasley at 00:05:55) I don't know. He's like, "I'm busy right now. I'm playing disc golf and enjoying my life." And I said, "Okay. Alright." So, it was a fun conversation. But that's generally, that's what I like to see. Like, when I see stuff happen out in the marketplace, like, really novel ideas, I'm like, "Oh, how can you make this easier for other people to do?" But I do want to talk with you about this really exciting topic. So if we're going to talk about how companies can scale differently in the age of AI, I want to start at the beginning. Before AI, over the past 10 years, the—you know, let's say, ChatGPT and the LLMs got real popular to the global landscape and the board members and all of that about two to three years ago. So say the 10 years leading up to that, how did we think about scaling?

(Robert Duffy at 00:06:42) I think you think about scale pre-LLMs and pre-AI as like this constant refactoring of the business. Right? So a good example is, a lot of organizations in my career when I've joined, they've had lots of teams doing things—the same thing. They've had, you know, four or five teams just as the company grows and as the company scales, end up replicating some capabilities, some piece of functionality. So you end up having an investment in these same pieces of technology doing the same job in four areas, five areas. Let's just say you're a company, you've acquired five different things. They will have, you know, five different content management systems for their websites. They're effectively doing the same thing. Right? And what you end up doing is refactoring that from an organizational perspective and from a team perspective into one team because it's very costly to maintain five separate areas of capability that are doing the same thing fundamentally. And in order to do that, what you have to do is find this kind of common ground amongst those teams, those five teams, and then you create this central thing. Maybe what you do is you actually take one that already exists, and then you move everyone on the other four teams onto that one thing. And then you refactor the organization and you refactor the technology. And what happens when you do that is it's not that you pick the solution that's going to work for each of those five teams, but everyone has to compromise a little bit. Right? Like, there's always a tiny compromise. But the compromise on things like functionality or the unique process that the other team has is always worth it because organizationally, at a macro level, you end up getting a huge amount of efficiencies from that. Right? Like, your efficiency is the ROI. You now don't have five teams. You have one. You can get some headcount, redeployment of resources there. You have one thing to keep up to date with security, patching the bug, fixing bugs. And you have one place to add any functionality if someone wants to have some functionality. And that efficiency gain is worth it as an ROI of the compromises that any one of those five teams has to make in order to get onto the central platform. And that's how we think about scaling. Right? Like, you think about that from a process perspective. You think about that from a people perspective. And you think about constantly refactoring the organization. So as you grow, you're getting more and more efficient. And your cost structure is not scaling linearly with the growth in your top line. Right? And you use this as a springboard to get more money to reinvest in innovation and go-to-market and—that's how you scale the business up. And that's how we used to think about it. I think that—

(Joel Beasley at 00:09:33) And so what's the—yeah. Yeah. Where does that change?

(Robert Duffy at 00:09:36) I think that changes a little bit with LLMs, and I think it's just starting to happen. I think it's going to accelerate. But imagine that example that I gave. Now the reason that you get some efficiency and you get some ROI is that the total cost of ownership for that software platform is high. It requires developers, software developers who are vigorously expensive, and it requires the management of those people. And in order to make sure that they're deploying their time and energy correctly, you have to have product management organizations, and you have to have designers, and you have to have all of these people around that piece of software. And the reason that the efficiency gain is large is because the investment is large in five different teams, especially software developers. If you think about the cost of producing that website as something that is continually getting smaller and smaller and smaller because we're using AI tools and AI agents, we end up in a situation where maybe that ROI isn't as large. Right? If that website was free—this is a thought experiment. Imagine that building that website uniquely for the unique demands of any one of those five teams was free. Right? Then you would never consider consolidating into one tool because there's no cost associated with it. You're not going to get the ROI of creating a central tool if it's free. So as we use more and more AI agentic tools, the cost of software development, the cost of building unique things starts to get closer to zero. We're not going to be at zero, of course. But it starts to get smaller and smaller. And at some point, it's going to get small enough that the ROI of building and consolidating and refactoring the organization into one, and one tool and one system is going to get smaller and smaller and smaller. And I think that's a change in dynamics. Right?

(Joel Beasley at 00:11:45) And so as it gets smaller, what is that going to—how does that affect the strategy of how you get gains?

(Robert Duffy at 00:11:54) I think how it's going to affect the strategy is that companies that think like that and start using AI tooling and refactoring their approach to scaling are going to outperform massively companies that are stuck in an old way of thinking about scaling and thinking about it as a constant efficiency play without thinking about how can I actually keep the uniqueness of these things, but drive the cost of building unique things down to zero?

(Joel Beasley at 00:12:27) Have you seen any examples of this inside of your company?

(Robert Duffy at 00:12:30) So let me take a great example, which is building a dashboard. Right? We used to think about building a dashboard as a thing that takes up time and resources. Right? Like a business analyst gets involved, captures the requirements from the business, and then ends up creating this dashboard, showing it back to the user, getting some feedback, and then putting it into production. And then when things change, when the data model changes, those dashboards have to get updated, and they have to move around. Right? So it's very costly to put together dashboards. So what we think about is we think about how can we build fewer dashboards that solve more use cases so that we don't have to spend a lot of time maintaining them and provide things to users. And that's how you kind of reduce the cost of that dashboard production. Now you have AI tools that can, in a hyper-personalized way, be interacted with by natural language. You can just ask an LLM, "Can you show me this data? I'd like to chart on X. I'd like to see a chart on Y." You can send me that prompt. I can send you it. You can build the dashboard. And if you think about the dashboard as a sort of mini application, we're now having LLM tools build these mini applications that are allowing people to explore data and get insights out of the databases and the systems that we have without the need for software development. And now we think about, okay, previously, we would think about scaling the business analysts and the business insights, how you scale that. You know, you reduce the number of dashboards and you try and get down to more commonality and standard widgets and you maybe create a dashboard platform team that owns and operates the dashboard infrastructure. Now we're just thinking about how can we create as many dashboards as we can as quickly as possible, but put that into the user's control so that they can build dashboards. And instead of building the team, we invest in AI tools and AI agents that help people do that. And that's where the scale is going to come from.

(Joel Beasley at 00:14:43) That's interesting. Do you guys build a lot of dashboards over there?

(Robert Duffy at 00:14:51) You know, I think we do. We have a lot of dashboards, but we have a lot of people. You know, we have a lot of customers who have very, very complicated businesses. And because of that, the dashboards are sort of your view into how the business is running. And our data platform is how you can extract information and insights on what's going on inside the business. Because I think about a business as like a nuclear reactor. You know? It just looks like a glowing bit of thing on the floor. So you need all these gauges and dials and systems to kind of visualize what's happening inside the reactor. I think as businesses get more and more complex, they get more and more like nuclear reactors, and you need to construct dashboards and information displays to really understand what's going on in the system.

(Joel Beasley at 00:15:40) Absolutely. Has your job changed at all now with this new acquisition merger situation going on?

(Robert Duffy at 00:15:48) Yeah. You know, we are very focused now on delivering business outcomes to our health insurance payers. Right? So previously, we were focused on building software products. And, you know, one of the consumers of the software products was, like, the UST Health Proof, as they were our customer, our largest customer.

(Robert Duffy at 00:16:08) And now, as together as a team, we're much more focused on delivering value and outcomes to our payer customers, leveraging our software. And we do that through automation, and, you know, we're building AI tools to help do that, and we're building a lot of systems and processes around that. So the job is kind of, I don't think the job has changed because the job of engineering leadership doesn't really change over time. But I think, you know, what we're focused on and how we're delivering value to customers has changed.

(Joel Beasley at 00:16:42) Mhmm. Did you get new people on your team, or are both companies still so fresh that they're both kind of business as usual?

(Robert Duffy at 00:16:50) We're figuring that out. Right? Like, it's day one. We're one organization. There's definitely some overlap between the organizations, but it's, you know, we are working to bring things closer together.

(Joel Beasley at 00:17:04) If you had any advice for another technical leader that's going through a merger, what would it be?

(Robert Duffy at 00:17:12) I would say that, you know, change management during mergers is probably one of the most important things. You know? When you merge, there's different players. Like, you're used to your peer group. You're used to working with people. You have a culture. There's another culture that's coming in as a team. And you really need to spend a lot of time building trust across the organizations. Right? Like, you need to think about, you know, how am I showing up as a leader?

(Robert Duffy at 00:17:45) How is my team showing up to, you know, the new members of the team? How are these two cultures coming together? If you have a new boss, like, what is the new boss's API? Like, how do I figure that out? And, like, I'm used to, you know, working within one way, one routine, and one regimen, and that's comfortable, and then you have to move into a new place.

(Robert Duffy at 00:18:05) And, you know, you have to figure out which things are gonna stay and which things are gonna change and how you need to change and, like, how your style needs to change. And I think some people I've seen come in from mergers, like, this is the way I do things. And my job is to educate everyone around me on my way of doing things. And I think that, you know, when you're bringing a large team together, you have to spend time kinda learning about the different approaches and different characters and, you know, like, who worked well together and who doesn't and, like, what style problems are gonna be there. And I think many people who, I see, like, a CTO position, you know, they think the company that they came from, like, because, you know, you have two companies merged over a single CTO. They think that the house that they've built is the best house. And therefore, the job is convincing everyone else to move into that house because that's the best one. And, really, the job is looking across organizations and understanding, like, where are their pockets of greatness? Where are their pockets of duplicated work? And it's, like, you know, you have to sort of reset.

(Robert Duffy at 00:19:18) You know, it's like, everyone is on a level playing field. We're gonna work this out together. And, you know, both companies probably have some level of technical debt and some level of process debt and some level of organizational debt. And, you know, we have to just figure it out together.

(Joel Beasley at 00:19:34) It's like you're the first two houses in a new construction neighborhood.

(Robert Duffy at 00:19:38) Yeah. Like, you're two very old—

(Joel Beasley at 00:19:39) houses together. Or two very old houses. Yes. At a new construction neighborhood. That is fun. That's fun. You just ask the new players for their documentation, for their README.

(Robert Duffy at 00:19:51) Yeah. Exactly. Just give me—

(Joel Beasley at 00:19:52) the API docs so I know.

(Robert Duffy at 00:19:54) Up-to-date documentation.

(Joel Beasley at 00:20:00) That's fun. There was a big trend maybe five years ago with this manager README thing. I don't know if you ever saw that go around in the, uh, in the engineering sphere.

(Robert Duffy at 00:20:10) Yeah. I did. Like, people have managed to go to their personal README file. Yeah.

(Joel Beasley at 00:20:14) I thought it was very—

(Robert Duffy at 00:20:15) I thought it was kinda cool. I actually thought, like, the most interesting thing about that was not that anyone ever probably read my manager README, and I did one. I created one once. It was more the level of introspection I had to do to figure out my own management style was crazy. Like, you would ask questions like, what annoys you? And I'm like, oh, I have to figure out what annoys me. And then, like, I'd spend, like, you know, an hour soul searching on, like, when I get annoyed, what is the real root of that? You know? I'm like, how did that manifest itself?

(Robert Duffy at 00:20:46) Right? Because it wasn't a poorly written document. Right? Like, what was poorly written about it? Like, how was it polished? I maybe overthought this, Joel.

(Joel Beasley at 00:20:54) No. You did a good job. You know? I either deeply overthink or I'm flipping. I would have been like, what annoys you? I'd be like, questions like this. Next. That or I would have went on a soul journey for, like, six months and came back with a book. It's one of the two.

(Robert Duffy at 00:21:10) Did Joel README? This is what annoys me. I hate this.

(Joel Beasley at 00:21:13) Yes. This task. No. I wrote one. I did one. And at first, I thought it sounded like, my first iteration of it was horrible. It sounded like, uh, the instructions you'd leave a pet sitter, like, if you're going away. Likes good food, pet them.

(Robert Duffy at 00:21:31) Yeah. And it always had, like, really silly things in it, like, you know, what do you find hilarious or something? Like, funny things.

(Joel Beasley at 00:21:37) Yes. Funny things. Yeah.

(Robert Duffy at 00:21:40) Yeah. It was a trend, but I thought it was useful. I think maybe it turned a little bit too kitschy and then, like, became, you know, unuseful. But the intent is good. Right? Like, just think about, like, what, how do you interact with people and what is your API and, you know, what is your style? And, like, I think the intent was to have an artifact that rapidly bootstrapped trust. Right? And, like, when you're a new leader or you're a leader who's inheriting a team or whatever, I always think that, like, the goal, your number one goal above anything else should be to rapidly bootstrap trust and do everything you possibly can to make sure that the new organization trusts you. And do as little as you possibly can to erode that trust.

(Robert Duffy at 00:22:26) And, you know, my model for thinking on this is, like, when you come into a new organization, you're like an organ transplant. And, you know, you're the organ, and the new organization is the body. And the body has an immune system. And, you know, what you don't wanna do is for that immune system to sort of reject the new body part or the new organ. And you, as a leader, you have to really think about, like, what is the organization's immune system doing right now? Is it accepting? Or is it, or are there, like, early signals that it's not accepting? And how do I, like, shut those down and build more trust in those areas?

(Joel Beasley at 00:23:03) Well, let's keep with this medical analogy. So did they check? Did the doctor check before they put the organs in that they were compatible? The blood types. Was the blood types compatible? Yeah. That's why they get rejected. That's one of the reasons why they get rejected, you know?

(Robert Duffy at 00:23:18) Yeah. Totally. I mean, that's what you're describing is the hiring process. Right? Like, what you're talking about checking the two blood types is really the hiring process if you're a new CTO. Like, that is, you know, the point of all of these different people interviewing you. But it should also be the point of you being interviewed. You know? So from you personally, you should be thinking about that during the interview process. You should be like, am I gonna get rejected by this organization culturally? Do I align with them? Mhmm. Like, is that, are there some early warning signs that within my style and my approach and, you know, it doesn't align with that culture or that culture, you know, might cause me not to perform in the best way. Right? And so I think that that's during the hiring process.

(Robert Duffy at 00:24:03) And then when you do, like, these acquisitions and you move from a private equity company to, you know, another private equity company or you're in a merger that's backed by private equity. The PE deal team are doing, you know, just so much research on the companies. And, you know, they do a lot of fact finding. They do, like, background checks on all the, you know, people that do reference checks. So, like, the amount of work that goes into you and the amount of diligence and the amount of smart people that look at these deals is amazing. Like, it's just, it is a machine.

(Robert Duffy at 00:24:39) And then, you know, when they, so they, all that's all to kinda check blood type as it were to make sure that they—

(Joel Beasley at 00:24:44) I've canceled mergers slash acquisitions. Everything looked like it made sense. The market looked like it made sense. The financials looked like it made sense. And then when the transactions are long, as you know, and then in those transactions, you interact with both sides quite a bit. And there's been some times where it's like, look. Everything else makes sense, but I just don't like spending time with this other group. And so I'm not gonna sign. I'm not gonna merge them together because, like, I don't want the next five years of my life to be like this. You know?

(Robert Duffy at 00:25:14) I think you should have one pick in five.

(Joel Beasley at 00:25:16) A personal story for me.

(Robert Duffy at 00:25:17) Yeah. Totally. But I'd like, I think you should do that when you're picking, as a, like, when you're picking technology. I think you should do that too. Like, a lot of people focus on the cost. Right? And, like, you know, like, a lot of people focus on, you know, the roadmap and the delivery and the features and all this sort of stuff. But, you know, we are big fans of Anthropic and Claude. And, you know, we often have this conversation where, like, we just, we like them. Like, they're morally aligned with how we think an AI technology company should operate and work and communicate. They seem like the good guys. We wanna be aligned with the good guys. You know? Like, it's, and same with Amazon. It's like, okay. These, they showed up as, like, super customer obsessed and super customer focused, you know, when we were doing a big cloud migration proposal. And, like, it should be part of the equation. Right? Because you're almost bringing those people into your house. It's like, you know, the adopted people. Right?

(Joel Beasley at 00:26:15) Well, you're creating dependencies on them. Yeah. And so your ability to pick the dependencies is a direct reflection of, you know, your ability to execute. If you put in the wrong partner, uptime's bad, that makes you look bad because you're built on top of them. Right?

(Robert Duffy at 00:26:29) Yeah. And also, like, to point to the, you know, you're gonna have to rely on them sometimes for implementations too. Right? Like, if you pick a partner and, like, you know, they're gonna, some portion of work is gonna be outsourced to them. Your team has to interact with them. So, like, when you're picking that partner and you're thinking about it, like, you know, is your team gonna get on? Is it gonna be a giant, you know, finger pointing exercise when things go wrong? Like, you know, when the project's running behind, is it gonna be this, like, oh, well, they didn't give me this or they didn't do that? Or are you more, you know, aligned to them culturally and, you know, can you get on? And I think you have to spend time with partners like that too. Right? Like but not just, like, you, not just the sales team because that's sometimes the trap that people fall into.

(Joel Beasley at 00:27:13) The sales team—

(Robert Duffy at 00:27:13) is a lot of—

(Joel Beasley at 00:27:14) Yeah. Exactly. Everything is good. Optimistic. It'll be done next week. We're good. Just sign. Just sign. Sign the contract. It's gonna be great. We'll go out for a beer. It'll be fun. Yeah.

(Robert Duffy at 00:27:25) Exactly. You know what? You know?

(Joel Beasley at 00:27:27) Some people are fun to hang out with, though. You gotta know. Some people—

(Robert Duffy at 00:27:29) are fun to hang out with, and, you know, they're great. But their job is, their job is to sell you on something. Right? So I think you have to spend time when you're vetting partners, you have to also spend time with the people that are gonna be doing the actual work and, you know, meet those people and kinda understand what they're doing.

(Joel Beasley at 00:27:42) Yeah. I remember the first time, on the other side of this, you look at this from client selection, but also keeping clients and, like, choosing who you do business with. There was a time earlier in my entrepreneurial career where I could never imagine, like, the idea or the concept of firing a client. I was just doing everything I could to, like, get business. Like, just get business, get business. And then a couple years into it, we had to fire our first client because managing them was the organizational equivalent of managing 10 clients. And, like, we could, and they wanted 10 times more for 10 times less. And it was just, it stood out like such a sore thumb from all of our other clients. We had enough clients to know, like, what the standard was then for, you know, the experience. And so I had to have a phone call, and I had to gently let them know that, hey, we're just going in a different direction as a company.

(Joel Beasley at 00:28:37) We're not, you know, serving this specific use case anymore, and they were not happy about it either.

(Robert Duffy at 00:28:43) Well, how did, well, how did they react? Like, what did they do?

(Joel Beasley at 00:28:47) They kept trying to get me to make an exception and to keep them. And to be honest with you, it just, it was, like, it was the truth. And they were, they were an anomaly of a customer. Like, usually, our customers were Fortune 500 companies. This was like an individual with, like, one employee. And just the, I don't wanna go, like, too deep into it.

(Joel Beasley at 00:29:21) But it just wasn't a good fit. And so their initial reaction was like, we'll pay you more money, and we'll do this. And I was like, if we determined inside that you're 10 times more work than other customers paying your equal rate, the only way it would make organizational sense for us is if you paid 10 times more, and they wouldn't pay 10 times more. And so at that point, it was just like, alright. Like, we just, we delivered the rest of their existing contract is what we did.

(Joel Beasley at 00:29:49) So we had, like, a 12 month contract with them. We had that conversation at month nine. We finished out the service with them, and then we just didn't renew the contract. So we held up our end of the thing. We came through on everything we said we were going to do. And they were very happy, by the way, with our service. And they were very happy with the product that we were producing. They were very happy.

(Robert Duffy at 00:30:10) Super, like, dedicated resources for because that cost you 10 times.

(Joel Beasley at 00:30:16) It was just a lot of like, this customer was a lot of work. And so it's like, why am I, you know, you're a CTO, you're at a big company. If you start hearing about a customer too frequently, you're going to dig deep in and figure out what's going on. Like, why is it this customer that I'm always hearing? Or is this our issue?

(Joel Beasley at 00:30:33) Like, is this something, are we dropping the ball? Is it a point of contact? Like, is it the individual at our company who's interfacing with their company? Is it the team around them? Like, where is the break in this chain? Right?

(Robert Duffy at 00:30:45) Yeah. But that's also, like, yeah, think about that in the context of scaling. Right? Like, and you have this sort of, like, that's the ROI of firing that customer was so high because they were so costly. But if you'd have had, you know, AI agents responding to all their requests and the trend of the cost of supporting that customer was actually going down, then it might be a different equation. Right? Like, you might actually...

(Joel Beasley at 00:31:13) A hundred percent.

(Robert Duffy at 00:31:14) Yeah. And like, I think that's what we're talking about, you know, how things are scaling differently, and that's a prime example. Right? Like, what you were trying to do is eliminate inefficiency and change the cost structure of your business so that it didn't scale, or that it scaled better with the top line growth. Right? And now you get this technology that has in the very near future, the possibility to just almost be costless. It's not costless, obviously, but in equivalent to people, it's almost costless. And, you know, you could have serviced that client without even thinking about it. And maybe you would have gone after 10 other undesirable clients. Right? Because you'd have been like, no one else can, the cost of supporting these customers I know is 10x for all of my competitors because they're this archetype. But now I've sort of changed the cost structure of supporting them using AI and LLM agents. Like, maybe now that's my differentiator. I can now go after customers that I know are really, really impossible to support. And all my competitors actually are struggling to support and offer them the best in class support that they could ever need. They can ask a million questions a day if they want, and they'll get the answers. You know?

(Joel Beasley at 00:32:30) Yeah. No. A hundred percent. And this was like seven years ago. So that LLM stuff, if it were around, that might have helped with it quite a bit. If it were as mature as it is today, that would have helped quite a bit. But, yeah, it is important, but it causes you to do a lot of introspection too as an organization. Because, you know, one question I've always played with, Rob, is like, why do some people wake up and create Nissan, and other people wake up and create Ferrari? Right? It's like they're both out there. They're both making cars. They're both killing it, making a bunch of money, but one person woke up and decided on a different level of quality, which then forces you into a different type of customer. Right? And so it's kind of interesting. And I've always been the person that wakes up and I want to build the Ferrari. Like, I want to build the nicest thing that can be built with the available technology today. I'm not brilliant enough to push the technology to create it from the ground up, you know, from the most detailed low level layers, but I can take what's already out there and make the best possible product with those ingredients.

(Robert Duffy at 00:33:40) I think that's an incredibly powerful skill. I think it's difficult as well when you're creating products because every product is a compromise of something. Right? Like, you have to, there's some level of compromise. I think choosing the thing that you want to compromise on is the hardest thing in entrepreneurialism. Because you have to make, you have to figure out how to compromise without compromising your core values. And to do that, you have to figure out what your core values are. And your core values are, I want to put the best possible product. Right? Like, I want to build the Ferrari. But there's compromise you have to make then. Right? Because then the cost goes up. And, like, maybe you can't make 10 Ferraris. You can only make one. Right? There's different costs and different considerations you have to have when you're making those compromises. You know?

(Joel Beasley at 00:34:30) I got to change that analogy. It's definitely a Tesla now. I would wake up and build a Tesla. I do, have you gotten one of these yet?

(Robert Duffy at 00:34:36) No. I haven't. I live in New York City, man. I moved over, I moved to New York from Salt Lake City. Yeah. It's like we had a car. We used to live in Seattle, had two cars, moved to New York City, and then I kept, they kept one garage and a car for a year. I'm sorry. One car in a garage for a year. And...

(Joel Beasley at 00:34:53) Cost more than the car to do.

(Robert Duffy at 00:34:54) Right? No. I mean, it's like, yeah, ridiculous. And I think I drove it a thousand miles in one year.

(Joel Beasley at 00:35:03) You just get around with Uber and walk?

(Robert Duffy at 00:35:05) Yeah. Yeah. I mean, now, if we want to go somewhere longer, we'll just rent a car, and it's cheaper to do that for the number of times that we need to. It's cheaper to do that than paying for parking in New York City.

(Joel Beasley at 00:35:15) Where do you live in New York?

(Robert Duffy at 00:35:17) Brooklyn. Williamsburg, Brooklyn.

(Joel Beasley at 00:35:20) Oh, okay. Cool. My, we have a bunch of family in Westchester.

(Robert Duffy at 00:35:23) Oh, nice.

(Joel Beasley at 00:35:24) So we get out there and then we take the train in. We see the city. We're country people. It reminds us of why we don't live in the city. And then we go, there's some amazing points about living in the city. But at this stage of our life, with the little kids, we're just like, we're going to live in the suburbs.

(Robert Duffy at 00:35:42) Yeah. We have, I mean, we have, I have a seven and a five year old, so like, they love it. They're just, they're entertaining all the time. Yeah. Yeah. And like, Williamsburg's like a community too. Right? Like, it's a much more kind of family oriented place than Manhattan.

(Joel Beasley at 00:35:57) Really?

(Robert Duffy at 00:35:58) Yeah. It's like super, it's super cool. Like, my kid, my kid will go to the pizza joint, and like, he'll be like, oh, hey, Joe. I'm like, who's Joe? Like, and then he's walking down the street, he's like, hey, hey, Tom. Like, I'm like, how does my kid, this sort of like, well networked socialite at age seven? You know?

(Joel Beasley at 00:36:19) They're doing that mayor strategy networking. Right? That is awesome. Okay. So I want to, as we start to wrap up on this topic, rapidly bootstrapping trust, that's very catchy. That's clippable. I like it. I want the one most important thing to rapidly bootstrap trust in an organization.

(Robert Duffy at 00:36:39) The one most important thing, I think, is listening. Just hearing people. And I will, I will, if I can get a 1.5 on that as well. I think listening and doing it, but doing it with genuine curiosity and the genuine belief that your opinions might be changed by what this person is about to say. Like, that is the single most important thing. I'm walking into conversations and just saying, like, you know, show me the systems that you own and operate and show me the tools that you have. And if you can do that in a way that is, you know, deeply genuine about learning about that and you don't have any preconceived ideas about what good or bad is or, you know, whether things are better or not better, and spending time to understand people and their motivations and how they built things and what decisions went into it. And it's an incredible builder of trust with those teams if you can do that. And the sort of opposite of that is like, coming in with, you know, a very judgmental approach and saying, no. This is crazy. This is, you know, you shouldn't have done it this way. You should have done it that way. Like, there's always a better way to do it. I think that's important in any organization with a legacy. You know? Like, they've got, they've got maybe five or 10 years of building software or whatever. Because all of those organizations are going to have some piece of that software that just looks crazy to the outside. Right? Like, why would you ever do it that way? And the answer to that is because at some point, the constraints were different. An engineer had to do something and make a decision quickly, and that was the best available thing that they could have done at the time. And they did it, and they made that choice. And now, like, it's a piece of technical debt that, of course, everyone wants to eradicate. But you can't judge the teams for creating that. You know? Like, you have to be like, okay. Like, everyone has skeletons in their closet, their architectural closet. Closets of closets.

(Joel Beasley at 00:38:52) That's true.

(Robert Duffy at 00:38:53) And, you know, no, none of these software systems that we build are perfect. You know?

(Joel Beasley at 00:39:02) Yeah. A hundred percent. You speak as a man of experience. Right? Because I could see the earlier version of me in my career versus now. And what drives me is just to achieve an understanding today. Today, what drives me is I want to achieve an understanding because there's almost always a great story or reason about why something is the way it is. The earlier career version of me had very specific ideas of what to do, how things should be done, you know, single responsibility principle. Like, every level I had my favorite frameworks and how things need to operate. And then I'd be fairly critical if we didn't put them exactly into those spots. But then I realized after working on enough projects, that's okay. You have to be able to have strong beliefs about how things should be done based off of experience, but they have to be loosely held to where people come in with new information, like saying, hey. It's not so bad to raise the kids in the city. I'm like, well, maybe it's not. We'll take a look at that. Because in my head, I have just me going to the hotel, you know, the Marriott Marquis or whatever it is, and just coming out and being at Times Square and just like, I wouldn't raise a kid here.

(Robert Duffy at 00:40:10) Yeah.

(Joel Beasley at 00:40:10) But then you're like, there's communities and neighborhoods. I was like, oh, maybe. You know?

(Robert Duffy at 00:40:13) Yeah. I think that's, you know, that's a little, it's exactly what you have to do as an engineering leader when you're kind of trying to get trust with the team. You have to be like, okay. Where do you live? Show me.

(Joel Beasley at 00:40:24) Oh, this is cool.

(Robert Duffy at 00:40:26) I remember I had this in a house in Seattle, and they had, like, for some reason, the drain from the toilet went under the shower and then out. And I was like, yeah, I'm like, what? Like, and then you could look at that in one mind, and you could be like, this is so crazy. The person who did this was a complete moron. Like, why would you ever do that? Or you could look at that and be like, there must be something under the house, like, maybe a giant boulder or maybe there was a tree root. Like, there must be some reason why this exists. And at that time, that seemed like the most logical thing to do. But that information has been lost because the house has been sold three times over. And now everyone just looks at it and it's like, who would ever in their right mind do this? You know? I think the same is true with teams and software systems and processes and everything that people have in their organization.

(Joel Beasley at 00:41:28) No. I think you're exactly right. We're going to have some fun here. Okay? Just a last question. What's one piece of leadership advice, can't be listening. What's one piece of leadership advice that you received, put into place in your life, and it's stuck with you and it stays with you to this day?

(Robert Duffy at 00:41:47) Listening? No. I think that the best piece of leadership advice I was given is like, whenever something is happening, you know, to you. Right? Like, you get an email, you get a comment, or you're in a meeting and someone behaves in some way or someone says something. Right? Like, you can have an immediate learned reaction to that. And the immediate learned reaction can be like, this person is trying to X. Right? Like, this person is trying to undermine me. This person is trying to hand me this project. This person is trying to, you know, take some control away from me. This person is trying to steal my org. Right? Like, there's some of these ingrained reactions that you can have. And the best piece of advice I ever got was, like, just assume positive intent and ask yourself the question, like, how would you react if this thing had positive intent? And what would your reaction be? Right? And try and reset that immediate knee jerk reaction from, like, complete, you know, anger or frustration or whatever it is that you feel in that moment and pause and reflect. And then respond as if that person had positive intent, regardless of whether or not you, after reflection, have, you know, determined that they actually have a nefarious intent. I think that is something...

(Joel Beasley at 00:43:18) Be surprised by nefarious intent.

(Robert Duffy at 00:43:20) Yeah. Like, but you do. Like, you can. I mean, you can find, and I think the reason why that's so important is because it's sort of like a habitual self reinforcing thing to have knee jerk reactions to the thing and think, you know, negatively about everyone's intentions and motives. Right? And the more you do it, the worse it gets, I think. And if you can redirect your brain, you know, to have more positive intent, I think what you end up doing is having more positive intent, and that's a much better habit to get into. It is. And then it works to families too. Right? Like, especially when you're dealing with young kids, you know, I have young kids. I have to constantly remind myself that my seven year old does not have the cognitive capacity to construct a situation that is deliberately and intentionally designed to annoy me. My seven year old doesn't have the cognitive capacity to plan a situation like that with the intent of making me just annoyed or angry. So he has no idea about that. So, therefore, this is just a happy accident and a happy coincidence that he's done the one thing that really annoys me.

(Joel Beasley at 00:44:32) Or he might. He might be a genius, baby. Yeah.

(Robert Duffy at 00:44:34) He's very well networked in Brooklyn, so maybe he's playing the mob or something.

(Joel Beasley at 00:44:40) He's been networking with some psychologists, and he's got you profiled. So...

(Robert Duffy at 00:44:45) Yeah. The older kids are like, I heard your dad hates it when you spend twenty minutes choosing shoes in the morning.

(Joel Beasley at 00:44:54) Yeah. Hey. Shoe choice, very important thing. But here's this pro tip. Just throw all those shoes out except for one pair.

(Robert Duffy at 00:45:02) Yeah. That's a good idea. Except the problem with shoes is that their feet grow quickly. So you have to either buy shoes that are too big or you have to, you know, buy a lot of shoes. And then I hate it when you buy something nice.

(Robert Duffy at 00:45:15) I bought him a pair of Air Jordans. They were super cool. I would have worn them. And he grew out of them within two months. I'm like—

(Joel Beasley at 00:45:29) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected].

(Joel Beasley at 00:45:47) Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.