$40M/Tech Founder Reveals the Smart Way to Run & Grow Your Company With AI in 2026

I think we're completely wrong to fear AI, but I think that humanity will change completely. AI can be better than humans, substantially better at many aspects of customer experience. I think, you know, we, especially as we fear the oncoming advance, further advancement of AI, we think about that human touch as the one thing left. I think it's really quite different than that. Historically, it was hard for smaller businesses to benefit from this tech because the setup costs and the cost of that ongoing managed service was really, really high. And that's no longer the case. It's definitely It's definitely possible to do more with less. What's really interesting now is to look at the adoption curve in any enterprise, and I think people are losing their minds. Most people have not yet learned how to use these tools very, very effectively.

What's the biggest constraint you see in the business at the moment?

It's a good question. Hi, my name is Nikola Mrkšić. I'm the CEO and co-founder of Pauly AI.

Brilliant. And what do you do? What does Polly AI do? And where are you in terms of your lifecycle and gestation?

Yeah, yeah. So we are a Series D company operating in the UK and North America. We're spun out from the University of Cambridge, where I met my co-founders and the whole kind of like senior research team. And we had worked on building really good voice agents for pretty much all of our lives. Professional lives and academic lives. So really the business started with an idea of proving to people that voice technology can be a good thing in their lives, not just this pesky thing that gets you to not speak to a human when you really want to speak to a human. Our previous company was acquired by Apple to make Siri better. That proved pretty hard at that scale. So I decided to start Polly and really focus on just improving customer service, large enterprise by large enterprise until I proved that we can do this at scale. And then now the business is much more of a platform business where both our teams with large enterprises, but also IT teams at large and smaller enterprises, as well as systems integrators, ISVs, and others use our platform to deploy really, really good AI agents that automate customer service.

Fab. Nikola, what is the What is it that you think is true that other people probably think is bollocks or vice versa?

Yeah, I mean, in the context of our business, I think it's really that, you know, AI can be better than humans, substantially better at many aspects of customer experience. I think, you know, we, especially as we fear the oncoming advance, further advancement of AI, we think about that human touch as the one thing left. I think it's really Quite different than that. I think we still have, you know, our brains are fascinating and extremely capable. And also, yeah, that human aspect matters in some places, not others. What's true in customer service is that we gave up that humanity when it comes to most companies a long, long time ago, right?

Okay.

Most customer service reps are bots. They're robots. They're reading a script. They're not allowed to use that humanity in any shape or form. They're just there doing what a bot would do. following the same instructions. And if you're going to do that, that same human is not reporting live why that's going south, why it's not working, what's wrong with it, whereas an AI does. So if you're really going to standardize, because it's really about that act of standardization of your customer service at scale, if you're going to do that, you're best off doing that with AI. And it's going to be a lot of work for the humans behind the AI, both to build it and to support it on an ongoing basis. So the humans aren't really going anywhere. Right. But they will be doing very different things. And then that AI that's front-facing is going to be better than what we have.

And is that the AI experience for the users, for us, is better than the average? Or does it also have the potential to make the better best?

Both. Both.

Okay.

You know, the best is the best if they know a lot about you, if they know a lot about your problem, and if they have like the right angle to help you out at the right time. And it depends what you're calling about, right?

If it's— because I would, because I would say, I would say the best, the, the, the politeness of the person is great, right? So you could probably, you know, make that work. But quite often, you know, you ring up and, and where it's really frustrating is where maybe you're the edge case, right? And the computer says no, and the human says, no, no, that's fine. I'll, I'll, I'll work around the edges. And like, and that then you feel as though—

You know yourself. In those edge cases, for the most part, what happens is you get bounced around between 6 different humans who are all trying to find that one human who knows enough and is empowered enough to deal with your edge case.

Yeah, totally.

It's rare, very rare that you get to the standard front-facing human agent of a company. I'll give you an example, right? I'm a vMO2 customer. It was the 1st of January, and had a lot of family over. The internet wasn't working. Now, small kids, and then from grandparents to small kids. You can imagine what happens in a household like that when you don't even have internet to calm the children using Netflix, right? It's like a true 1st of January, right? Not Christmas, but almost as comical, right? So, what do I do? I start looking through my email and I realized that my bill, the whole kind of payment when it was set up, I got a confirmation email with someone else's name on it. And I'm like, wait, this looks like I've committed fraud, like I stole someone's identity and put it in so they pay my internet bill. I had done no such thing, but I didn't check that email 'cause I just set it up. And I don't know how this happened. It's honestly none of my business. I called and spent, I think, 2 hours on the phone. And through the virtue of my job where I've now worked with a lot of large companies, I almost take delight when I map the journey and see what's going wrong, right?

Yeah.

6, 7 people between me, including sales, support, technical troubleshooting. And then the internet came back. So honestly, if I hadn't called, I wouldn't have even known about any of this. Finally, after those 2 hours, a ticket was filed. And like 5 days later, I was just messaged saying it's all been dealt with. Your details are now in and you've been retroactively charged for the last 7 months. I'm like, I shouldn't have even called, A. be. I'm glad it was resolved so that it doesn't look like I committed identity theft, right? But you think about that process there, it's not really AI or humans, it's really like the edge case-ness of the edge case. And those humans were as flabbergasted as I was with the situation. So I think that when you look at AI, if you're a business, you start identifying, does this happen often or is it like every time panic and then it gets lost through the contact center? If this happens like 10 times a day, then it's probably worth looking into, because there's going to be some major fraud case stemming from it at some point, right? Or like a huge fine they're going to have to pay. If it happened once in— well, if I was the only person the whole year, then maybe okay, like you deprioritize the fix, because I'm sure they've got plenty of edge cases to deal with. But with humans, it's very rare that they would actually have like that taxonomy of like, this is happening and that is happening and that is happening. So Let's go solve one at a time. 'Cause each of those issues is something that a team of humans has to work cross-functionally on to figure out inside that company, right? So yeah, I think that the whole, is it a human or not, the issue is really just like, the more complex the business, the bigger the failure modes. And there's some very large businesses out there.

What's your favorite, I was gonna say best, but what's your favorite OurAI does customer service better than humans could implementation?

Ooh, we've got a bunch, right? I think that we built our brand on Las Vegas and on hospitality more broadly, right? So we work with Gordon Ramsay, for instance. He is the brand face of Polly AI. We did a big Super Bowl ad in January, February with him. We work with Caesars. We work with Pokeraton, which just closed their first deal with the Rosewood. So, you know, to me, those are precious because it's really about like those guys don't mess around with customer experience. If they're implementing this, it's because it's better, right? Like you hear those names and you're just like, yeah, like they didn't go there to like penny pinch, right? And then, you know, we have some other examples, including like financial services where we have this like phone line that deals with A lot of customer issues. I won't name the client, but you know a large portion of them are to do with bereavement, and our first recommendation was no, no, no. Those calls immediate escalation to the most proficient human we can reach, right? Because that makes sense for technical reasons. That was really hard to do, and they just said, "Hey, do your best," and then you know, like if you have to hand off, hand off. But they'll be in the same. queue as everyone else. So we put a lot of work in. There's a tone change, there's like a really, really, really good bot that, you know, goes through kind of like what the procedure is for submitting the claim, because people have to do it physically.

Yeah.

It's not something that can be done on the phone regardless. And what was really interesting is we got higher customer satisfaction scores for that call type than humans, substantially higher. And I was just like, this can't be true. I was like, this is like my team, like, you know, rose-tinted glasses. Something wasn't considered. I didn't believe it for the longest of times, but the data held and we looked into it and started thinking about it. And then you realize when you're a human customer service rep and you have to deal with 1 in 8 calls being about someone losing their loved ones, my God, like it's not a great conversation for either person on the phone because You know, on the caller side, you are already going through something terrible, and you don't really want to talk to someone about it. And you've probably had to call 10 different numbers to deal with, you know, a bank and a telco and memberships of different things. God knows, right? Like, you don't want to have that conversation. And if it's not with a human, if it's with technology, the emotional register, you just don't have to go through that. And if you're the customer service rep, then it's like, well, Like, oh, what do I say? How do I deal with this? And you know, they had coaching, serious coaching for this. But, and I've since taken a great interest in this. There's no right answer, right? Do you ask a question? Do you say something nice? How can you say something nice? You don't know much about the person. You don't know anything.

But that's because, because, because you're the coaching for the humans and the training for the bot is you go in the same place, aren't you? From a, from a, like, what is the psychology of this conversation? What is the what is the optimal way to drive the conversation for the person who's—

Well, it's fairly, right? Because as humans, we have expectations from each other of like what common decency is. And you know what, even cross-culturally, like it's quite different how you would deal with this in like a British context, Caribbean context, or whatever. So like, you know, we don't all really behave the same, right? What taste is and what, you know, good behavior is varies across cultures, right? In some, it would be expected to ask something about the person. In others, it would just be like, okay, acknowledge, move on. And that's probably the safest thing to do more broadly. But then with an AI, they just don't care. Humans don't care. They're like, oh great, I don't need to now go through that awkward part of the conversation before we just get on with it, right? And if they, like the pressure on the human is interesting, 'cause if they just want to get on with it, like do they not care about the person who died? Are they just scrolling to like— inherit their money? Well, no, they have to make that call, right? But then that all just goes away. The emotional buildup of the call is gone with AI. And we've seen it since in healthcare. We've seen it with people especially talking about embarrassing health conditions. So think lines about STDs and stuff. And then more broadly, I think the number one use case for ChatGPT is as a therapist, right? People will tell AI their deepest secrets and, you know, like insecurities, hesitations, just because they feel like they're not being judged. Right. So that's like a place where AI can do more for humanity than humanity can.

Right. Right.

We are repressed. I mean, we don't easily come out to other people about these kinds of things. And with AI, I think that with ChatGPT, for better or worse, right, like what it has turned into for probably like 10% of the human population is already a bespoke therapist, right? And not everyone can afford a therapist. Not everyone. I mean, I think very few people have therapists, right? So.

Okay. What else is it you believe that other people think is mental or you think other people are mental for believing?

Hmm. I mean, to go a bit more broadly in the sphere of what we do, but not exactly what we do. I think we're completely wrong to fear AI, but I think that humanity will change completely. I think that with the smartphone already, we went closer to becoming this hive mind where a meme in London can, you know, spread and be a meme in California by the evening, right? So we already have created something that would be unheard of, you know, like 30 years ago. And, you know, we can be nostalgic about that world, and there's great things about it, and there's great things about today's world. But I think with AI, we're like accelerating the pace of information where we're just becoming more of that hive mind, right? Companies are aware of everything instantly. It's becoming more competitive, it's becoming exhausting, but we are becoming different humans, right? So the same way you have your smartphone, you're going to have your agent, right? And you and your agent will be you. And you know, you'll probably pay for extremely expensive retreats to go away and be without your agent for weeks at a time, the same way that people drop their phone. But you know, you also I don't want to not have my smartphone. It is an essential tool for me doing most of the things I do. So we can't really imagine what all those things are just yet. But I think we're in for something really exciting as humanity evolves. But it will be like a pretty radical transformation of the human experience.

Very good. And how is AI impacting your workplace? You've got what, 300 employees? And so what are you, is, does, are you, do you have, has AI meant that you can do a higher level of turnover with fewer people yet, or are you just doing more work?

It's definitely possible to do more with less, right? I think what's really interesting now is to look at the adoption curve in any enterprise and I think people are losing their minds in fear of AI psychosis-induced CEOs thumping around and pushing AI through their org. And I think most people have not yet learned how to use these tools very, very effectively. There are different frameworks that people used to talk about, just kind of like how activated, so to speak, you are, or how literate you are with respect to AI. Yeah. Most people are not really past that kind of like level 1 where they're like, oh, like rewrite me this email or this or that, because it really takes just kind of like adapting to your workflow. And it's quite different for a lot of people. I think that's why you see a lot of different companies emerge and go up and down really quickly, you know, whether it's for like development, like, you know, Cursor or different harnesses used for, for different tooling. But yeah, I mean, you can definitely do more work yourself. Now I think what's interesting is what that allows businesses to do in the way that they're applying themselves, right? And how do you advance your product roadmap? You know, it's just like it's compressed the timeline and the cost of doing some things in the evolution of your company. And then there are others where you just can't do as much, right? I mean, you can do a lot in almost every department of the business with AI, but I think that that lens of like more with less, sure. I think that like if I look at PolyAI in 3 years from now and PolyAI now, I think we'll probably grow headcount. a bit less quickly than we are now, we're still hiring like mad, right? And especially like on that human side, people buy from people, right?

Yeah.

So like in go-to-market, we don't see much of a change. I think we see a big change in the tooling we need to use for marketing, for sales, et cetera. But at the end of the day, you know, we're a B2B business, so that part is not changing, right? And then one of our PMs told me something very interesting. It was like, You almost need more PMs, not fewer than before, because he's like, whatever I write, you know, it would've taken like the engineering team like 4 weeks. They come back 4 days later and they're like, here you go. He's like, what the hell? I'm not ready for like the next piece. Okay. So like the pace has really, really increased. And that's interesting, right? So I think on the engineering side, again, there was, I think 3 months ago, there was consensus that coding jobs are going away and blah, blah, blah, right? The truth is, it couldn't be further from the truth. There's, I think, then a month ago you had that like research that said that there are more advertised software engineering jobs now than ever before. Right. And I think like really what's going to happen is the same thing that's always happened with technology, just probably even faster than ever before, where people who don't adapt to using these tools are going to get eviscerated by younger people who have higher plasticity, And are just able to learn new tools faster. But that's no different than ever before. I remember, you know, my mother worked at the National Bank of Serbia right as Milosevic fell and like the whole like post-communist transition then properly happened. Like it was a daunting time for her and her colleagues because computers, Excel, everything that didn't used to be done in that way was starting to be done that way. Right. And she was, I think, of the 40-person department. It was my mother and one other colleague that weathered that transition. And, you know, lo and behold, 10 years later, it was still like, I don't know, a 40-person department. I don't know if it was 30 or 50, but it was the same sized department, right? So I think, you know, we for a while heard people complaining that like there are no job opportunities for young people. I think this is the shot. I think this is where, you know, like kind of like the Gen Z generation stuff. Really comes into its own with tooling that they are much better at using than the older generation that's, you know, and I mean, I'm a millennial, right? So it's not like I'm threatening an older generation here. I think even my generation is a bit old already and we see it, we see it at work, right? I think the younger the people are, the faster they tend to be at adopting these tools, which doesn't mean that, you know, I think our CMO who is not Gen Z, you know, he's like number 3 on our like token leaderboard, right? We don't push it publicly, we don't use anything, but it was just really funny when we vibecoded the dashboard and looked at it. We were like, well, our CMO, who, you know, I mean, he is technical now.

What are you tracking in the dashboard?

Oh, I was just curious to see which tools we're using. I think my co-founder, Sean, who's the CTO, has, you know, willingly or unwillingly become the CIO of PolyAI, where it's like, okay, what tools are we using? How much are we spending? You know, everyone's looking at the spend.

And token usage on Claude Code, for example?

Well, Claude Code is one of the many that we use. It is number one, of course. I mean, but many others, the different Anthropic ones, the GPT ones, and everything else, right?

And are you— one of the people I was talking to the other day said, I'm not hiring junior developers, I'm hiring senior developers, because senior developers and the right tooling, 10x. Junior developers and the right tooling, not so much of an impact on the team. Is that, you know, you were saying, look, this is an opportunity for young people, but how do you, do they have to be self-taught then if they're not going to get opportunities?

We love narratives, don't we, as a society, right? Like, and pushing these things and these, like, you know, I feel like every era has its version of, you know, Like Gen X versus boomers, millennials versus Gen X, etc., etc. We just love that stuff. We can't get enough of it, right? It's all hogwash, right? I mean, like that same senior that you— what did you and I talk about before we started this? You know, school runs for kids, etc., etc. There is a 21-year-old who, unlike Nicola and Dominic, does not have to do a school run, right?

Yeah.

And, you know, on the weekends they're not going to Peppa Pig World. And look, I love doing that. It's a great part of my life, right? But, you know, they're sitting there and, you know, maybe they're like off their faces drunk, probably a bit, right? That's what you do in your early 20s. But equally, we hear they're drinking less and less. So, they're probably using Claude code and learning how to use Claude code, right?

Yeah.

And getting better at it. And, you know, they've got more energy, more plasticity. Plasticity just goes down as you age, right?

Yeah.

So, your ability as a human being to learn new things as your age goes down. That is like a law of nature. So there is absolutely nothing that can convince me that these seasoned 20-year-old— sorry, 20 years of experience developers will keep up as fast as the kids. Like there is just— that's never happened before. It will not be true in the future. The kids will be better at using these tools. Now, I think that what's also true is that what's funny about this one, unlike some of the other maybe kind of like new waves of technology, is that it was pushed top down, right? It's a very top-down push from the CEOs, from the boards, because of that narrative around efficiency, right? So as it got pushed down, I think actually some of the mid-senior developers were probably pushed to use these things before the junior ones, right? But I think that, you know, we're really talking about, you know, people who come in and they are like AI native and they grow up with ChatGPT. We don't have that generation just yet, right? And ChatGPT is just the beginning, right? Like when you think of, and you know, we will adapt to it. Like it's good enough and powerful enough that we are all learning these tools, but like the kids have always learned them faster and that'll be true.

But you know, I was chatting to my 10-year-old the other day and I said, oh, what are you doing? She said, oh, well, that's Google AI mode. That's how I search. I don't just use Google. I'm like, oh, okay. And over the summer holidays, I said, you're not allowed to play games you didn't write yourself. And so I think they will definitely be of that AI-first generation as they join the workforce in a few years' time. Yeah.

And can you imagine like then you and I competing with them? Right? When their whole mental model is wired around that and, you know, you're like thinking of this world where you type in keywords and they pop up. It's—

Well, it'll be— it's as ridiculous as a cassette tape. Although again, my 10-year-old is very retro, so she's made me buy some CDs and a CD player because she thinks the whole idea of having a CD player is fascinating.

No, but you know, it's really funny. You look at the world today and just because, you know, we're still growing, You know, touch wood, and, you know, the GDPs are growing and we have more than we've ever had before, right? We all live the lifestyle of medieval kings, if you think about it. You know, we can wish any food to be delivered to us and it will be delivered to us anywhere in London or wherever, right? Like, so, um, I think, you know, the retro goods, the cool things of the past are just going to grow in value, right? Because, you know, we got to spend money on something.

One bad hire could cost you £50,000 and 6 months of momentum. And if you're turning over more than £5 million, you've definitely lived this. Not because the person was bad, because look, you've hired off a vague spec and you've trusted your gut feel in the interview. We all have. You bring them in, you carry them for months, and then you're back to square one. CEO fixes that before you post the role. You tell it where the bottleneck is. It gives you a role profile, what to test in the interview, and what good looks like in the first 90 days. One bad hire is going to cost you £50 grand. CEO Friend is going to cost you £145 a month. Get your first month free and have a play. stwtrialceofriend.ai. Link's in the description. How has it changed the way you run the business, do you think? You know, if you think back a few years ago and what you're doing now, are you, um, are you, you know, are you manager mode? Are you delegating? Are you in the detail, in the weeds? What, are you working harder than ever? I mean, you are working. I'm sure, I'm sure you're not relaxed. In a fast-growing AI business competing on a global stage. But has it changed? Have the tools given you more to do, less to do, different?

Yeah, I mean, look, I think they've definitely given me like more, but like better as well. I think, you know, they, one of my teams said this to me and I think it's really, really interesting, right? Like, you know, it basically gives you this ability to like become much better at the things that you were worst at before, right?

Uh-huh.

So, you know, I think one, one instance was someone who is like very technically strong in our team, but not much of a natural seller, not a sales role, but a client-facing role. I rocked up into this meeting with like a public company, C-level, and this person was presenting and I looked at the slides and I know their usual slides and they're good, but I'm used to kind of like having to like Steer it a bit towards, you know, like a more commercially oriented thing. I looked at the slides and I was like, whoa, what happened? It's like nothing. Like the guy wrote a skill that just like kind of like beats the previous, the first draft he has with questions that usually, you know, are not answered in his stuff. And then he added it all and he provided information and iterated and created something much better. So I think similarly, you know, to me, I was always very like founder mode in that you know, the highest leverage point in the business. I go there, forget everything else, then I emerge, like look around. And I think like what's been possible with AI now is just to have a constant feed and organization around this stuff where it's kind of like a chief of staff reminding me like, oh, you said you would do this 4 days ago. And like every morning I'm going to just look in my like daily briefing. It emerges and goes like, you've been radio silent with this person for 3 days. And like, it's enough to make me feel bad that I will probably then at 11 p.m. do something about it, right? That nagging thing doesn't come after me the next morning and the next morning, right? So I think that's just like a very, very powerful tool to make a business run better because, you know, my weakness always has been like, you know, dropping mid-priority things because I'm working on the highest priority ones. And, you know, a little bit of work, you know this, right? Like the CEO can be a blocker. Right. And like, people are, no matter how much you tell them, like, hey, nudge me, whatever, like, they're not really going to say, hey, you irresponsible prick, like, I've been blocked here for 4 days. No, they're going to go and do something else or they're going to plan for both optionalities. And maybe they have to because you genuinely are uncertain about something, or maybe you're just being bad and you haven't told them a crucial data point you have or you arrive at 2 days later. You should tell them immediately and you might forget. And now I don't because, you know, I've got most of my meetings like transcribed and it's just ingesting, ingesting, saying, okay, you refer to that to this person. I don't see you've sent it. I'm like, ah, you didn't capture, capture this Slack exchange. The bot's like, okay, cool. Okay, you're right. Or like, no, I haven't. I'm going to go and do it now. Right? And it's like that leverage, honestly, I've always— and I'm sure you felt the same building a business, but you always feel like there's something in the back of your mind that you forgot. I didn't get back to that person. Yeah, you're laughing. You know the feeling.

Because you think to yourself, because you just go and you go, that thing, make a note because I had forgotten it's 2 days ago. And yes. Yeah.

And you know, like how liberating is it? I mean, I do work more than ever before, but also frankly, because this is really fun, like building these harnesses and systems is really fun. And my co-founder is, you know, I think he's like a stage further into the AI psychosis than I am. So, I'm just perpetually trying to catch up with him. But it really gives you superpowers. And it's also really funny in that it's quite hard to bring people with you, depending on how good you are at articulating these things. So, you build these phenomenally effective tools, but they're really mostly tools for yourself. I think that's a different thing maybe to smartphones. And I mean, even they are highly custom. use their iPhone in the same way, right? But here you figure out something that adapts. It's like a well-broken-in shoe. It's your shoe, right? If you take the same model and you put on someone else's shoe, you're like, this is not my shoe. Like this, you know, I can't really wear this.

And in the past, you'd have used a piece of software and you'd have had to adapt to it. And now you don't. So it's that, Everything's unique. What's the biggest constraint you see in the business at the moment?

It's a good question. I think what we've been focused on this year is really just kind of like going from this really high-end, because you can think of us and what we've built so far as the most premium player in our space where you care about your brand. You care about doing customer experience even better. You're willing to give AI a shot because what you're doing right now is not giving you the results you need. And then you work with us and we'd send these forward deploy teams to work with you and to really kind of like wrap themselves around your operation and just figure out how AI can give you the most leverage, right? So if you think about like the whole piece, it's really been both this like consulting motion and tech implementation motion and the platform behind it. Right now, the more you do it, the more kind of like you build it into the platform. So that effort is lower and the more you learn around like what tends to work for what kind of business. So we're now really focused on just becoming much more of a platform business where, you know, you can go onto Polly AI and create an account, build an agent for yourself, integrate it on your website, put it on your phones, and just go and work on your own, right? Now, that really means that if we were a carmaker, right, I might struggle to find the right model here, but BMW owns Rolls-Royce, right? So I think, so the model is like there's a Rolls-Royce for the large enterprise where we'll go in and work with them really, really closely, including with our deployment teams who are the most experienced people in the world for implementing this kind of AI to put on your phones and your websites, to drive both CSAT improvements and OpEx reductions in places where it's possible. And we're very good at telling people where it's not possible, or where they're just like, having a pipe dream around what can be done. And we're very well known for being very forward. about that. And then the other business, you know, that was Rolls-Royce. The BMW is like, hey, like, it's still great. It's very, very strong, but it's something that's a bit more off the shelf. It's not something that's handmade and assembled.

Mm-hmm.

It's something that, you know, like you can go and use on your own to give yourself a fleet of these autonomous customer experience workers that will work together with you. And, you know, we for the longest of times have been constrained to large enterprises. Rather than the wider economy, exactly because that nature of deploying human teams to do this work was expensive, right? And it meant that, you know, if you're not talking like at least mid-six figures and probably low seven figures, it wasn't really realistic for you to have this thing. It was like, you know, me deciding with my two small kids that I want a like Porsche 911. Like if I bought that. and my wife saw me park in front of the house, I could pitch her on putting kids in the back, but I think I'd be driving back very quickly. So similarly, historically, it was hard for smaller businesses to benefit from this tech because the setup costs and the cost of that ongoing managed service was really high, and that's no longer the case. So our orientation this year has just been to strengthen the platform business underneath both the BMW and the Rolls-Royce by just making it better and better and better and doing those iterations more and more quickly.

Historically, B2B SaaS businesses have been really hard in Europe. B2C have worked, but B2B typically, unless you're in the US, it's just too hard. Is this wave of AI technology, is that different? Do you feel like you're winning?

We're winning spectacularly in America.

Yeah, yeah, yeah, yeah. So I think, you know, but even that, even that has historically been hard for so many companies. You know, it's like we started in Europe and we go the other way and it just, it's hard.

Yeah, I never went there. I was there from the start in that, you know, I knew of that, right? So my go-to-market was always built with, you know, an America-first perspective.

Yeah.

And, you know, we have me and all of our, like, GTM leaders who are still in Europe when they're not on a plane back and forth from America. We work late hours here. Right. And that's just the curse of our generation because, you know, like, it's a good thing that the West is still integrated and, you know, one political alliance. And America is just a spectacular country. Its economy is just unbelievable. Right? You know, I was looking at a random like graph of how its growth has just accelerated since World War II, and like the place defies gravity, right? You look at the markets today, and it's just, you know, you'd be foolish to bet on any other Western market over America if you want to build a global champion, right? So that's just like the rules of physics. That hasn't changed. I think the other part is just also the European markets, even though they're large, like we are not growing as fast. And that just doesn't give people the liberty that they need to experiment with technology, right? It's much easier for an American, you know, business line owner to commit to doing something with, with a vendor and spend like $100,000 than it is for a British one or a French one or a German one, right? Because the companies are bigger there and they're richer and they're growing faster. So all of that is a lower risk. So when they say they're risk takers, I'd really like to see those same Americans in European companies, no matter where in Europe, take the same decisions. It would be harder because they have less resources, so they couldn't, right? And I think that, you know, again, the same way we like waging wars between different generations, you know, this is something where the Europeans love doing that, like, you know, and we do too much of the whole like self-punishing, criticizing, um, compared to, you know, our American Colleagues, and you know Americans love you know oh the lazy Europeans, you know on holiday for two months, and you know I I get I get no end of this because we are a delightfully Anglo-American company run by a Serb. Yeah, and you know I think that like it's just really you know now it's become just a meme where you know I can promise you that I wish we took two-month holidays in Europe. We do not. In fact, I think in many ways we now work harder, and on average, sadly, get less. And that's that structural problem with the economy in Europe. How we fix that, that's above my pay grade, right? But I think for software companies, if you're going to be B2B, like, you're going to have to plan for America. And the best case, you get like a delightful Anglo-American company with, you know, people joking about lazy Europeans as those same Europeans take calls at 11 PM. But that's just the curse of our generation, right?

Well, I spent most of my working life working for North American companies. And then when I went and lived in the US for a bit, it was like, oh, it stops at 5 o'clock. No, no, it just, it just stopped at 5 o'clock. It's like, what? I spent, I've spent 10 years being up till 12.

It's a shame, man. It's a complete shame, right? I mean, you know, like you look at the Bay Area and like at 3 PM, like, you know, you can't, you wouldn't be able to cycle through the traffic from the number of people going home to beat the traffic. And it's like, well, You know, like, I know, I understand that you are very hardcore for doing your first call at 7:00 AM, but, you know, like, the fact that you're taking your last call at 3:00 PM from a car, you're not working harder. You just have an inherent advantage, which you should be very grateful for. And, you know, it is what it is. Well done, America, right? But yeah.

And what about the other meme, which is, you know, great companies get built by immigrants in whatever country? And what, and is that, do you think that, do you think the fact that you're a Serb in the UK has something to do with your success? Or even maybe work ethic versus the, your colleagues or our colleagues in the UK?

Definitely something to do with immigration as a process, right? It has nothing to do with Serbs, Brits, Americans. I think that's like flawed. I think, you know, we can like do like profiling and talk about what Brits or Serbs are good at and what their failure modes are. They're not the same. I think you know both nations are fascinating and have fascinating failure modes, but it's not really about that. I think Brits are phenomenal entrepreneurs. I mean, you look at the British Empire and what it built, like hey, that's quite an entrepreneurial endeavor, right? I mean, it scaled to be bigger than anything else the planet has seen. So it's it's not that. I think that like immigrants in countries just you know they go through a lot of. They go through a lot, right? Like you leave home, that hardens you a bit. You know, you build a new life, you learn to build just through the act of building a new life, right? So it's not surprising, by the way, how many Brits in America are America's, you know, like top entrepreneurs, right? And, you know, many of the largest American companies are either built or run by immigrants, whether they're Europeans or they're Indian or Chinese, Taiwanese, right? Like all sorts, right? So I think there's definitely something to where an immigrant has a bit of an advantage, right? Like inherently you grow up with 2 cultures, right? So if you're gonna be a manager of a large team building it from scratch or taking it over later, you know, if you're honestly Serb in the UK, it's not even that interesting 'cause we're just like, you know, like white European Christian, right? But you look at like Indians in America, right? I mean, India by itself is a multilingual wonder. Right?

Yeah.

Like, it's a continent of its own that we forget about. We model it as like one European country, right? But then you look at like someone like that who then grows up in an American setting, like they've just been exposed to a lot. So they know how to manage different cultures and how to get people to work together well. So I think the advantages just stack up when you grow up like that, right? And I don't think— like, also, I think like if you just compare averages, Immigrants are those people in a society that are willing to go through a lot to achieve a lot, right? So I don't think there's a difference. I mean, there are differences, but Serbs in the UK are like mid-range immigrants. It's not that far. You can be home on a Wizz Air flight in 2.5 hours, right? I think long-distance immigration is an even better proxy to sample entrepreneurs from, right? But you look at Brits in America, I think they kill, they're really, really good, right? So, and also like, you know, they're less encumbered by like the number of ties they have, so they can just work nonstop. So yeah, I think there's definitely something to it, but you know, I wouldn't really overthink it, 'cause some of the best companies were built by people that are like, you know, X generations from a country in their country, 'cause their advantage is that they know the system well, and they know how to work in that system.

And What books have you read along the way that have made a difference to you, inspired you, gifted to others?

It's a good question. I'm not much of a book giver and stuff. I think it's quite— people process information in very different ways. So, you know, like, what's the famous one about, you know, Radical Candor, right?

Yeah, yeah.

The number of times— I mean, you know, I think it's an interesting framework and it's a good framework, but, you know, it's basically every other time someone reads that, they think they've gotten a mandate to be an asshole to everyone, right? So, you know, we had an early, you know, we liked the framework and we channeled it through the company. And I'm like, you know what, after a while I wasn't sure that it actually made the company better.

Right, right.

So I'm quite careful with it. I think tactically, to anyone like starting off in like commercial roles and negotiating, I think Chris Voss' Never Split the Difference—that is just a textbook where I see no problem with that book or any way that it can like backfire in that it just teaches you to listen, right? It's like about negotiation. The guy is a former FBI hostage negotiator, and you know, I think he also writes delightfully well. So. That's the book I'd recommend to everyone. It's tactical in that it's really about negotiation, but more broadly, it's about like, listen, listen and find angles to give people something that costs you nothing. That's a good lesson for life overall.

Yeah. And why did you end up interested in the thing you did at Cambridge? Like, why voice?

Complete accident, if I'm honest with you. I grew up with a heavy kind of like STEM, and I competed in like programming, maths, physics, went to a really kind of like fairly typical kind of like elite grammar school in Belgrade. They exist in every Eastern European country. They're a kind of like communist invention where, you know, People who you take the biggest nerds, you stop them in a room basically, and you teach them like university-level maths in high school because why not? And because of that, I ended up at Cambridge doing computer science and maths again—a very similar environment.

But how did you how did you end up at Cambridge?

Oh, well, I mean, look, I grew up in Serbia through like a civil war, Yugoslavia breaking up, and you know, I think my mother spent half of her savings. Sending me to learn English so that I can do this if I want to. And then, you know, you compete in maths, physics, programming to win international awards because then you are more likely to get a scholarship at a leading Western university, which is really your ticket into doing what I do now, right? Or whatever, or working at Google or being a professor at a top university, right? It's just, The systems have been built to use these things as a conduit, so it pays to get in, right? So, through that, I kind of like applied for a lot of American universities, and then my friends, who a lot of them didn't really wanna leave Europe, were like, Cambridge is the best math school. Now, my school was the kind of school where like what is best really, really mattered, right?

Yeah.

And Cambridge is the best math school in the world. Was, is, will be forever probably, right? Trinity College, Cambridge is, you know, as they proudly said on the first day I was there, We still have more Nobel Prizes than the entire country of France, which is a little British fascination, right? Whenever I see articles about how we have more unicorns in London than all of France, I just giggle because it's just on repeat, right?

Yeah.

The English and the French. But it's, you know, I applied with them just because when you apply to Cambridge, you have to go and do an interview there. So I was like, well, okay, before I go to America, and I was accepted at a few top schools there, I was like, before I go to America, I'll just go and, you know, visit the UK with my friends. We'll go do the interviews at Cambridge, then we'll go visit London. You know, we had saved enough from our scholarships to pay for that trip. So I was like, it's a bit of a boondoggle, really, right? And I ended up getting a full scholarship. In fact, of the 8 of us that went from my high school, 7 of us got full scholarships for Trinity College, Cambridge, which is— we had more people accepted that year into Trinity Than Eton, right? And that was quite like an upset, and it was a national sensation. Not that anyone in Serbia really knows what Trinity College Cambridge is, but you know, like it was funny, right? All but one ended up going there. My friend Dushan ended up going to MIT anyway. He stuck to the American plan. Then later he found his way back to Europe and London anyways. But I couldn't say no because you know. My best friend to this day now lives, you know, seven minutes away from me in London. We're both here. We've built a life, and I think that's the British advantage in that London is the one global metropolis the continent has, right? So guys like me will naturally gravitate here and then build here over going somewhere else. And then at the end of my undergrad, I almost went into banking. I met my wife while interning at Credit Suisse.

Okay.

And that was me rebelling against being the nerds, you know, and I didn't want to just go and work at a hedge fund or, you know, hedge fund as a prop trader or, or quant. Right. So to me, rebellion equated to like, I'm going to be like a suit-wearing banker, which really shows you how not far I was capable of going because I didn't really realize that those guys are just nerds who are not as good at maths and they're a bit better with people, which kind of describes me relative to a lot of my friends growing up. Because they were really, really out there. And then right before I was about to kind of graduate and go there, my master supervisor, Zubin Garamani, was like, what are you doing? Are you insane? And I respect him greatly, but I was like, what do you mean? He's like, you want to be a PowerPoint monkey? I'm like, well, yeah, I do. I do. That's the Serbian side. I do. I absolutely want to be a PowerPoint monkey. Right? He goes like, and you know, the guy's seen everyone and everything. He's one of the best people in the world at Whatever he wants to do, right? He just chuckled and he was like, "Okay, okay. If you don't want to do a PhD, why don't I just introduce you to a few startups around here and you decide?" Was it cool? And I meet these guys. They were starting this like dialogue systems company. They were like three guys—one like South African, the other person was Indian—and the Indian guy Vishal, he knew everything about Yugoslav history. I went in for an interview with him about this company where he was going to turn this nerd into being their first engineer. And we just talked about, you know, like communism in both countries, monarchists, different battles, Soviet battles in Finland where his wife's from. And I was like, I like these guys. I like them a lot. I want to work with them, right? Yeah. And I think that's like life writes very interesting stories where I ended up being their first hire. And, you know, like that company got acquired by Apple. I ended up doing a PhD with Steve Young, the other co-founder of that company, who is a legendary professor in Cambridge. He's worked on this stuff his whole life, and you know one thing led to another, and I started a company, stayed here, built this. But yeah, it's um I'd never planned on voice. I was trained much more rigidly around you know maths and very rigid things, whereas voice voice is messy. It doesn't work. It still doesn't really work, right? You know this from your like daily experiences with it. So it's a deep problem that you can like wrap your life around. And maybe it won't be in 10 years from now, but it's still very much like we're in the thick of it, where like it works well enough to bring tremendous value to companies and build a pretty good experience, but it's not solved just yet, right?

So, um, is that why, do you think there's a, you've got to differentiate, you know, you've got a gap there between, you know, OpenAI and Anthropic because it's just hard and they've got some other big hard problems to solve, leaving You out on your own?

Yeah, I think that's like a false dilemma, right? Like, I think that underlying technologies improve and give you, like, they solve some of your problems and they don't solve others. And, you know, you are what you eat, right? And you are who you spend your time with, right? And, you know, OpenAI, if they make it through this right now, right? I mean, they're a consumer app, right? They killed it with their consumer app. That writes a particular kind of DNA, right? Then Anthropic is more enterprise-oriented, and we really admire them, right? Like, we don't see any competition with them. In fact, we spend an enormous amount of money with Anthropic, right?

Mm-hmm.

And, you know, I think that they're out to really challenge the old guard of the very large tech companies, right? And that just means that they're like distribution businesses for a lot of things, right? Now, the gap we're really solving is, and the bet that you have to believe in to believe that Poly is gonna be successful, is do you believe that Anthropic will provide all software to all businesses, right? And that it'll just, you'll tell Claude to write all this, you'll vibe code an agent in a day and you can do it today, right? And then you hope that your IT department and between the CIO and the COO, you hope that your vibe coded thing by your 10 engineers who have never built this before won't completely crumble your NPS score. And at the end of the day, what are they buying when they do Pauly AI instead of making that bet? It's a throat to choke because if something goes sideways and they call me at 2:17 AM, I will be on the phone. Well, maybe it won't be me, but it'll be one of my lieutenants. And like, it is our life's work to make them prosper with AI on this side of their business, right? So, you know, is it a smaller bet than an Anthropic-grade thing? I guess for now. Is it still a huge bet in that, you know, 2% of the British population works in customer service, 2% of American population? Yeah. They do, and they'll continue to work in something adjacent to it, right? I don't think those jobs are going away, but I think the companies that don't use this to become much better to their customers, they'll be wiped out. And new challengers that use AI well will mop the floor with them. So we're here to make sure that those enterprises can benefit from AI. And I don't think that in that there's anyone really like us. And we've cared for a long time. And I think also our DNA is deeply technical. So we are really good at passing the value from the model layer all the way through platform into kind of like the whole like solution, the application layer that the enterprise needs. So that's complicated. And you know, you've worked with large companies, you know how complicated they are. Like if at the end of the day you can, you know, why did Google never conquer this space? We've been fighting them left and right. We've won deals, lost deals to them. I can promise you that in the 5 years of winning big battles against Google and losing big battles to Google, Whoever went the other way still doesn't have a good voice agent implemented, whereas all those that went with us do. And then a lot of those guys, 2, 3, 4 years later come back, right? They may come back to us, they might come back to others, but it won't be Google solving their problem because they don't care about it. In the words of one of my largest clients, he's like, look, they care about ads, they care about compute use. You have to go very down their list of priorities. And, you know, like their best people, do they want to do what we do? Maybe some do, but I think they're much more likely to find themselves applying for a job at Pauly AI if they want to make an impact to how this technology rewires the planet than by convincing Google to shift their priorities, right?

That I'm gonna— we're gonna— I'm gonna leave that there. Thank you. That it's your life's work, and that's where that sort of passion I think is, is that's where small companies always win against big companies because they can harness that, you know, that spark.

Yeah. Thank you. And you know, like, you don't, you don't win every battle in that war, but you win the war because like it is, it is your war to win.

It's, it's that, you know, if you look at the wars that America have walked away from, it's that, it's insurgents with belief.

Yeah, exactly.

That sort of asymmetric warfare. Brilliant. Well, look, thank you very much indeed for coming on the show this morning. It's been fabulous talking to you.

Yeah, it was my pleasure. Thank you for having me.