AI doesn't create new principles, it amplifies them
AI is not introducing new principles like curiosity, humility, and empathy; it is just exacerbating them and making them more important than they have ever been.
Nathan Broslawsky is the chief product and technology officer at ClearOne Advantage, a Baltimore-based debt relief company that helps people climb out from under unsecured debt. The model is the mirror image of a debt collector: where collectors buy up debt for pennies on the dollar and then chase people to pay it, ClearOne works from the other side. Clients build good savings habits by setting money aside each month, and the company, drawing on nearly two decades of creditor relationships, negotiates those balances down to less than what is owed. It all runs on contingency, so ClearOne earns nothing until a settlement actually lands. Broslawsky's own charge is narrower and stranger: take a deliberately lean team and hand it superpowers.
He comes to the work as a technologist first rather than a lifer in the debt industry. His previous company operated in a similar space, and he spent about five years there before leaving in December 2024. Then he did something he had never done in his entire professional career: he stopped. The pause turned into a stretch of consulting and advising, until ClearOne reached out and asked, in effect, whether he could do for them what he had done before. He said yes and started in the spring of 2025. With roughly two decades spent building product and engineering organizations, he treats team design and culture as the real craft, and he stepped in just as agentic coding tools were beginning to rewrite how software gets built.
Broslawsky's core argument is that AI has not invented a new playbook so much as exposed the old one. Curiosity, humility, and empathy have always mattered, and now they are decisive. He hires for adaptability over any fixed skill set, prizing people who deconstruct things rather than accept them at face value, and who can admit when they are wrong. He notes that intense curiosity often travels with anxiety, and that humility is what keeps it in balance. On process, he sees the whole software lifecycle shifting left: as execution gets cheap, the hard work moves up front, into scoping and into asking whether the team is even solving the right problem before a single prompt runs.
What animates him now is a worry that easy execution is making work lonelier, as collapsing functional boundaries let people build alone and let relationships quietly atrophy. His answer is intentionality: pod people together, protect a weekly block just for learning AI, and treat the office as a tool for connection rather than a reflex. He believes AI is forcing a company-wide DevOps moment, where every function has to write down its definition of good, or at least good enough, and hand the low-leverage work to the machine so humans keep the high-leverage work. And he practices what he preaches about disconnecting: when the noise builds, he gets on the motorcycle and rides for a ridiculous amount of time, just to be unreachable.
Read full transcript of interview
In this conversation: Josh Rubin (Host, CTO Studio) and Nathan Broslawsky (Chief Product & Technology Officer, ClearOne Advantage).
I always start with the hard question. Just tell me your name.
Nathan Broslawsky.
And Nathan, what do you do?
I am the Chief Product and Technology Officer at a company called ClearOne Advantage.
Which is?
We help people get out of debt. A lot of times when people hear that, they think consolidation loans and things like that. What we actually do is help people who are really struggling with unsecured debt, like credit cards, by basically negotiating with creditors on their behalf.
Oh, it's like a direct negotiation. It's almost the reverse of, what's the word, debt collectors. Debt collectors buy up debt for pennies on the dollar and then try to collect it. You're operating from the other angle.
We have people put money in a savings account every month, get on those good savings habits, and then we negotiate with the creditors. The company is about 18 years old, so we have a lot of longstanding relationships with these creditors, and we can negotiate away their debt for less than what they owe.
And then you're taking a percentage, I imagine, of the payoff?
Yeah, only when we actually settle the debt, though. We don't do anything upfront.
So you're running on contingency?
Yeah.
That's great. How big is the company?
The company is about 700 people.
That's a lot of people.
Yes, it is. About 200 people are full-time, and then we have a lot who are BPOs and offshore people working in our call center. We have about 80,000 people enrolled in the program at any given time.
A Philippines BPO? Where are your call centers?
Yeah, primarily Philippines.
Gotcha. My girlfriend's visiting her family in Manila right now, so I'm constantly looking at things in and around Manila. Do you travel there?
No, never have. Haven't had a reason yet. I've heard it's beautiful, I've always wanted to travel, but they have a pretty good operation there. They haven't required me on site.
In the BPO space, especially the call-center space, they've been dominant for a long, long time. That said, there are a lot of operators in the AI space trying to move in and take that on. Have you guys been exploring that at all?
Yeah, we're playing with the idea of AI voice agents and AI chat agents. I'll be honest, on the inbound side, people calling us, I think there's a bigger use case right now, and people are a little more comfortable, especially if someone's already enrolled in the program and we already have that trust established and they know who we are. We are experimenting there. But on the outbound side, especially sales and marketing, we're still very human-centric, and the idea of calling someone with a bot, I think it's going to be a while before that becomes commonplace and really effective for our business, especially in an environment where people are challenged with debt. It's a sensitive topic and we want to really be there for people.
Well, I imagine you do cold calling and outbound communication, so that's awkward enough.
Now, we're only calling people who have come through one of our funnels, but still, making that leap from digital to a phone call, that's hard.
Yeah, it's a hard conversation to have with someone you know intimately, let alone a stranger.
Exactly.
So have you spent your career in this particular industry, or have you been in tech in general and this is where you've ended up?
More tech in general. The company I was at previous to this was actually in a similar space, and I was there for about five years before I left, I think it was in December of '24. I took a little bit of a break to do the consulting and advising thing, because I'd never really taken a break in my entire professional career. And then this company reached out, they found me and said, "Hey, can you help do for us what you did for them?" And it's been a really fun ride ever since.
So you joined this new company at a pretty interesting time. These new agentic tools are only coming out within the last year or so. Claude, Codex, ChatGPT, really rolling in, being able to change the SDLC in general. As a CPTO, has that changed your approach to your product builds and your team building?
I would say it changes every week. When I started at this company, I knew it was going to be a really lean team, and it was really exciting to me because all of these advances were starting to happen. But this was April last year, and things like Claude Code, that was very, very early if it had even come out yet, and people were still using Copilot and basically fancy autocomplete. But you saw the rate at which this stuff was developing, and I just thought, I could come here and give this very lean team superpowers. We've made a lot of progress, but it really does feel like every single month it's a new tool, a new strategy, so we're adapting as we go. To your point about the SDLC, I think that is fundamentally changing, because the way we've always developed software is we've banked on "this part is hard and it's going to take a long time," so it absorbs a lot of the open questions. Now, with execution getting that much easier, you basically have to front-load all the problem solving, front-load the conversations, front-load the "is this going to work? Are we solving the right business problem?" And then, by the time execution happens, because execution is getting easier, that's the stuff you start to offload. So everything in the SDLC just seems like it's shifting left.
The promise of AI is that it gives you a lot more time back. That's never actually been proven true, because there's an infinite amount of problems, we're always going to fill it up with something else. Where's the bottleneck moved to for you?
There are a couple of bottlenecks. And this comes from a little bit of uneven adoption of AI throughout the process.
From your teams, or in general?
From the teams, from everyone who contributes to the software development lifecycle, because everyone's learning this stuff together. You've got a whole spectrum of, let's just say, engineers. Some are still like, I'm used to doing things my way, and I'll use the tools but it's not going to take over my job, to the complete opposite end of the spectrum where people might be using it too much and just trusting whatever comes out of it. Where you start to see the bottlenecks emerging is when you have those human touch points that still exist and may always exist. Those are the seams between teams, anytime you have a dependency. When software development was a lot slower, you could absorb these dependencies, because I have something to work on while they work on something else and we'll all come together. But these days, all of that is exacerbated, because everyone can move so fast and get so much done that if you didn't pre-work out that dependency, if you didn't pre-negotiate those conversations, then those bottlenecks still exist. Or it could be on the requirements side: we didn't fully plan out this project and we've already started working on it and we're working really fast, and then we hit a wall because there was that thing we didn't think about, and now we have humans who have to go talk to each other.
Well, the number of times you enter a prompt and halfway through it, oh, shit, stop.
100%.
As a product person, I imagine that changes your approach to scoping a project, to articulating what the outcome is supposed to be, because if you don't know at the beginning, you're probably not going to know by the end.
Yeah. Whether it's an engineering problem or a product problem, you basically have to take a step back and start to ask all those questions in the abstract, solve the problem on paper. And you're relying on your teams to be genuinely curious about, but why are we solving the problem? Is there another way to do this? And more and more, as we start to build that context layer internally of how our products work, and we document all the legacy applications better, we can even start to have conversations with the AIs on, are we solving this the right way? What am I not seeing? So we have to front-load more and more of that.
This is the 12th time today I've heard curiosity. And the language of curiosity, humans are a curious species, but it varies how curious we are. I'm wondering, there's plenty of room for ultra-curious people in development, in software engineering right now. Is there room for people who are not as naturally curious anymore?
I struggle with that, because I almost think, no, you almost have to be curious. It's one of the key things I try to hire for: are you the type of person who will not take something at face value, but try to deconstruct it, try to understand how it works? With all these new models coming out all the time, if someone is just complacent and like, well, I think I know, and by the way, anyone who says they know everything is probably just trying to sell you something. But if you don't have that ingrained curiosity and want to pick things apart and stay on the cutting edge, I think you just get left behind.
Optimizing for anxiety is kind of exhausting.
Yes.
Which, if you look at where society is right now, maybe this is why everyone is just a little bit on edge. Because there is a fine line: a curious person is often an anxious person at the same time. They're constantly game-theorying things out, thinking about the what-ifs, thinking about not just the best case but the worst-case scenario. How do you hire for that?
You're right, a lot of very curious people are in that heightened anxiety state, because they're hyper-optimizers. But there's a side of it that can counterbalance it, and that is, are they also humble? The people who are intensely curious but have problems admitting when they're wrong, or when there's something out there they don't know... having a little bit of that humility, understanding, I'm not going to know everything, I'm not going to be right all the time, these two things can counterbalance each other. I want to hire people who are of the mindset, I can learn from everyone around me, and I'm not the smartest person in the room.
So we're going to have to run Dunning-Kruger as well as curiosity tests on people. It's changed how we hire, I imagine. I ask this question to a lot of people: your company gives you infinite money, you can hire whoever you want. When things are changing every week, for a lot of people, I don't even know what I'm hiring for right now. Do you?
I know the general principles I'm hiring for. I brought up curiosity, humility, empathy. As fast as all these things are moving, it kind of becomes all of our jobs to lead that change management and bring everyone along for the ride. And knowing that you can connect with someone and put yourself in their shoes, I think that's really important. It helps to lower the level of anxiety, because you realize you're all in it together. So all of these three things together mean I'm not hiring for a certain skill set or experience as much anymore, although those are valuable. What I'm hiring for is adaptability. Have they shown that they can learn something new? Have they shown that they're questioning things? Have they shown that they've had strong relationships and have been proven wrong and came out of that? So adaptability is probably where I'm most focused on hiring people.
Empathy, adaptability, the ability to take responsibility around things, and agency, to take control of that. Some of these are what we hope are best practices in humans. You want to hire good people.
100%, and that's the thing I keep coming back to. Yes, AI is transformational, but fundamentally these are things we've always had to hire for. We've just maybe had a little more wiggle room in the past. I think what AI is doing is not introducing these new principles, it's really just exacerbating them, making them more important than they've ever been.
Interesting. On the flip side of the positive thing, what is giving you anxiety right now, whether in team building or sociologically?
Let's touch on both. There was a study that came out recently, I don't remember who did it, but it basically said we're getting lonelier than ever in the workplace because of AI. And you try to deconstruct why that is. It means people have way more autonomy, way more agency, they can get a lot of work done, functional boundaries are collapsing, you have a product manager all packaged together, and that means they can do work by themselves. Whereas what used to happen is you had all these different functions who had to come together to jointly solve a problem, and that's slowly going away, and over time, probably for the better. We used to hire front-end and back-end engineers specifically because of two different stacks, two different skill sets, and then as technology improved you started seeing full-stack engineers, and you could do everything with JavaScript, all packaged in one person. I think we're starting to see some of that now, but it means people are doing all the jobs themselves. For someone who's been in this industry a while and takes a lot of joy and pride in building teams, building cultures, and great working relationships, that part causes me anxiety, because now you've got these people and they're not building those relationships all the time, and the relationships are going to atrophy if you don't actively invest in them. So I think that's where we are right now. The pendulum has shifted toward a lot of individual activity. I do imagine the pendulum is going to swing back to more group activity, but it's going to look different than it ever has before.
Do you pod people up on projects?
Yeah, we actively do. More and more engineers are working together, even if they're not working on the exact same thing, they're working adjacent to each other. We have a block of time every week where people are just supposed to learn AI, because we know learning is not linear, and that's where they set aside time and work together. Even if they're solving different problems, at least they're in the room talking to each other, which is a little hard to do when you're a fully remote company, but we're making an active effort.
I think the loneliness epidemic is less about the ability for everyone to do everything on their own, and more about cognitive load. The inundation of information that comes at us has killed boredom, and from boredom is where you get the desire, like, I have nothing to do, I guess I'll just go outside and play with Ricky. But when you can fill every waking moment of every single day with content of various types, it doesn't occur to you that you can just go out and interact with other people, until you realize you really need it, and once you realize you need it, you don't know how to get it.
I think you're absolutely right. This is one of those things that's been coming for a while, even pre-AI. As soon as social media started taking over our phones and we were all connected, I remember we were having these exact same conversations, and it takes a long time to get there, it's taken probably about 20 years for people to realize, oh wait, I need to disconnect, I need to start having a little more balance in my life. I think we're going to start seeing that with AI more, and maybe we don't wait 20 years for it, ideally, but people do need to realize it's the same thing. You have to disconnect, you have to go outside and touch grass. I would get on the motorcycle and drive for ridiculous amounts of time just to be disconnected.
The other part of having interactive teams is the conversation around remote work versus in-office. You're already operating a mixture of distributed, or at least overseas teams. Do you optimize for in-office work?
We have no office, as of the pandemic. Back in 2020, they basically said, okay, we're fully remote, why pay for real estate? So the center of gravity of our company is still around Baltimore, but we've started to hire all over the country, and we're trying to make a deliberate effort to bring people together periodically. At a previous company I worked at, right before the pandemic, when I would go into our San Mateo office, I had one team, and we also had a Phoenix office, and I had another team there. It really was, I had two teams, because people would always bias toward talking to people just around them. And then six weeks later the pandemic hit, and one of the really interesting things was, I had one team, because everyone could now talk to each other with the exact same level of effort and without the bias of proximity. I really liked that. Over time, as we started going back into offices just one day a week, what I realized was I was going into my close office, San Mateo, and basically holing myself up and having meetings. But when I traveled to Phoenix, I was meeting people, I was trying to build relationships. And I had to take a step back and wonder, why do I treat these two offices the same? To me, an office is a tool. It's where you intentionally set aside time to build relationships, and you can leverage a home office as a tool too: this is my heads-down productive time. So I think there's a lot of value in bringing people together for specific things, but I would hate to do it just reflexively, like, oh, we just have to get people together, because there's a little bit of intentionality to it.
Is your engineering team fully distributed?
Fully distributed. We've got a few in Latin America, that's kind of our nearshore contracting arm of the team, but everyone else is distributed throughout the United States.
What countries?
We've got Colombia, Brazil, and Uruguay.
Uruguay, yeah, we're in those countries, a few others. How have you found the difference between those team members and domestic?
We've really tried to treat them as first-party members of the team, and honestly, I don't see a whole lot of working-relationship difference. The nice thing is, because we're largely east-coast-centric, they're one hour ahead. It's not like working with Eastern Europe teams or India teams, which, fantastic talent, but the time-zone differences can be a little oppressive.
There are also some really interesting cultural nuances when you're all awake at the same time, and you all see the news at the same time. Do you travel down to those teams at all and interact?
No, those have been pretty hands-off. They've been at the company much longer than I have, actually, and have some really well-established relationships. I haven't had to travel there either.
Any other bits of the AI thing and how it's affecting everything that you want to get into or talk about?
We talked a little about the SDLC and how that's changing, but even on the hiring side, a lot of the same practices can start to apply. I mentioned we can do more shifting left, do all the problem solving up front on what are we going to build, why are we going to build it, why do we need it, and then you execute. One of the interesting things we've seen is on the hiring front, and keep in mind we're a 700-person company, we have call-center employees, sales agents, so we have a lot of bulk hiring to do. Part of this is, why don't we take a step back and actually define up front what a good employee looks like, what kind of behaviors we want them to exhibit, and how would we know based on reading a resume? You can actually start to train and prompt these AIs to ferret out some of that information when you're reading resumes, and you're able to shortlist a lot of candidates. This has actually caused us to challenge a lot of our maybe hiring biases from the past. If we can dig into the details a little, we can shortlist, and then really streamline our interview process, because we did that problem solving up front.
Yeah, it would also help you avoid class-action lawsuits like Workday is dealing with right now. Which is always important.
Totally.
I think the interesting thing about AI is it's going to force people to optimize for trust, in-person and high-touch, more than any other time in history. So how can the AI tools be used to actually facilitate that outcome?
I think that's absolutely true. But the other thing I think it's really going to force people to do is define what they want. This idea of actually writing down, what is the definition of good? I think AI is forcing us to have our DevOps moment. DevOps really forced us to blur the line between what an operations team did. You have application engineers building something over here, but oh, I need to deploy it, I need to run it, hey ops team, do that for me, and then you have dependencies. DevOps came along and said, actually, we'll give you the tools, you be accountable for what you're trying to do, and those teams can run and be more autonomous. I think we're having that DevOps moment now, only at a broader scale across a lot of different functions, like brand and creative. I don't want to always have a design dependency for everything I'm building. Hey design, can you give me the rules you follow? Can you write down what the definition of good is, so I can do it myself? I think we're past this age of needing a team of people who write reports. Hey, I'm going to build something, go write me a report so I know if it works. I think that's going to be handed over and be more autonomous too.
I hope so. I think, however, that rather than defining what good is, what AI right now is able to produce is something that's good enough. And that's where some of the tension lies. This is good enough for someone who isn't a designer, but the designer looks at it and says, that's not good.
Yeah.
And there is an inherent danger of optimizing for good enough, sociologically.
I think that's absolutely true, and you're never going to get those touch points down to nothing. But the important part is to encode the thinking, start to democratize it, and come in when it's the human high-leverage work that you need to have, not all the low-leverage work, to your point, just getting to good enough. And by the way, from a product standpoint, sometimes good enough is good enough, because the goal is to learn, to put something quick out there just so we can get feedback and decide the direction.
And then, to end here, that's probably the point: you don't have to define what's good, you have to define what's good enough. And the best companies start with the basic principle that it's never good enough, but is it as good as I can make it with current constraints, finding the constraints, whether they're financial, time, or talent, and that's where you get "this is good enough in current circumstances."
Totally agree. Thank you so much, this was fun.
Be part of the
conversation.
Whether you're a CTO who wants to be featured, a company looking to sponsor, or an engineering leader wanting a seat in the room — there's a place for you here.