Nikola Mrkšić, Cofounder and CEO @ PolyAI

0:00Nikola Mrkšić:All right, we are back. We are here with an amazing guest. We just had Victor from Zendizia, another amazing AI founder. We're now here with Nikola from PolyAI. How are you doing today? Thank you for joining me. I'm good. I'm great. Thank you for having me. It's a pleasure. Why don't you just give a quick introduction for those who haven't heard. What is PolyAI? Poly AI builds a voice agent, and AI agents for customer service, where we help large enterprises become the best version of themselves in every customer conversation. So we work with the likes of Marriott, Caesars, you know, things like companies like Unicredit, Allstate, PG &E, really large companies with thousands and thousands of agents.

0:37You know, we help them pick up any phone call,

0:40Nikola Mrkšić:chat interaction anytime of the internet and just be that best version of themselves. And you are currently targeting, like in the companies you mentioned very large enterprises how has that process been of kind of getting your foot into the door and selling innovative new technologies to quite old-fashioned or traditional industries oh yeah it's hard look and i think that uh we started a few years back straight out of our phds at cambridge where really it felt like we were selling science fiction like you can imagine the reaction of someone in a company where with a kid coming and saying well you know we're going to automate we can automate half we can automate 90 percent what are you on about, right?

1:16So that was very different. And we really had to sell the dream and look for those like extreme early adopters, really innovative people. I think the child GPT moment changed it all, where now people are kind of like factoring in and everyone's thinking about where we will be in a year, in two years, in five years. Often, I think, you know, I mean, it's hard to know whether you're overshooting or not, but there's a lot of optimism and excitement now. So you get more opportunities to prove out something that may be an outright ridiculous or at least a very ambitious idea. So yeah, I mean, you know, winning the first few is really, really, really hard, right?

1:49You need to find something, you need to find the right person and something they really care about. You know, what propelled us to fame and, you know, I think of all of our competitors, we're known as like the premium, most branded option, the most expensive one, right? We've always been fairly unapologetic about that. And, you know, the thing that's really interesting is you go and you say that, you got to have something people really want. So when we started working with Caesars, that I think was one of the key moments because Caesars never automated any customer experience. They refused it.

2:18They were famous for saying no to it. And we got the angle of like, we'll have exactly the kinds of voices you want. Hell, we'll have a carousel, 10 different ones. It'll sound exactly the way you want them to. And the moment they decide to work with you, that goes from being something that companies that don't really care about their customers do to save money into something that the companies that care most about their customers invest in to have an even better experience. right so when you flip the script like that then you get known for it and then the rest follows

2:45Nikola Mrkšić:super interesting so i guess yeah the the the sellers not not only you automating things but actually improving the quality of the customer experience you're improving customer outcomes and it's no longer just about yeah automation reducing costs it's about making the customer experience better yeah exactly right like you know first and foremost you improve customer experience you grow revenue you never miss revenue generating calls and then you put these systems in places where they can do things better than humans and what we found is that there is a surprising number of use cases already where we're able to like seriously outperform a human contact center so believe it or not like in cases where people talk about bereavement and inheritance where you know a loved one passed away we found that we universally get better csat scores than humans because it's actually easier to speak to someone who is not a human without invoking a kind of like emotional backdrop of like you lost a loved one you have to talk and then they don't know how to respond do they ask you about them or do they just kind of dodge the topic and keep conversation going of course with AI it's just transactional and that just makes it better so you have a better experience that's so interesting yeah yeah it's like there's real instances where actually talking to an AI is actually preferable to talking to a real person because of the emotional weight or baggage that's so interesting well and you know people talk about like um you know preference for there was a generation of companies and there still are those that push chat with that kind of like premise like Gen Z they don't want to pick up the phone and well no one really wants to pick on when they when their internet doesn't work or something that happened with a service or something like you don't want to have that interaction to begin with because you don't want to have that problem to begin with yeah once you do people want live connectivity with the company that can help them right and their voice still trumps everything and everyone for every age group right amazing that's so interesting i also wanted to ask about the talking about you know the early days of selling to enterprises before the chat gbt moment has there have you seen a real noticeable change in the appetite and demand from enterprises for this you know when you look at your icps whether they're like a head of innovation or head of customer success and services are you seeing that they are that are they feeling the pressure to adopt ai solutions in a way that is just is only going up extremely i mean i think what i'm for your you will have heard from others is this is a rare time where enterprises are moving faster than kind of like smd and that part of the market because um everyone's aware of this kind of like avalanche things coming so they want to get ahead of others.

5:24I think it's also the past experience of some companies being fast and most of them being quite slow with cloud adoption and then kind of like paying for the non-accumulated interest on innovation that they could have had, whereas their competitors might have realized it through using cloud products and everything that came with it. Everyone's very keen to not be that guy who didn't adopt cloud early enough and is still going through a cloud transition 10 years later so there's almost like you know a posture which is like by default you must and you will right the question then just becomes like from whom right and there you know what we've seen previously and we see a bit less of now is the kind of like you know the the flight to safety of hyperscalers in very large companies that are able to um you know you don't get fired for buying ibm right yeah or google or microsoft etc right there is still quite a lot of that you know like these execs and others will still, you know, largely look to cover their ass when making large purchases and the larger they are, the more they will hedge.

6:27But I think, you know, it's also recognized in AI that those guys are not the ones that have like the high-end products, the ability to deliver. And because of that, I think we see an increased openness to working with, you know, well, large startups, startups that have kind of like made it, companies at our stage.

6:45Nikola Mrkšić:true and has is that different when you speak to u.s companies versus european companies is there a great appetite to try these tools out in u.s companies yeah i think you know this whole like european and like what are we doing wrong i think we often ask ourselves in the context of like why america is doing better and my only answer and look we're 82 american revenue right america is you know like there are cultural things that have to do with america being you know a bit more empowered at the organizational level to make standalone decisions, try it out for a year and move away. There's definitely a bit less of that consensus decision-making in America, but it's also just a much bigger market and it's a much richer market than us.

7:26Yeah, yeah. Because we build European companies, it's actually immaterial, right? We're on the same side. We're selling to American companies. That's going to be a home market as will be Europe, right? And depending on what you have and how well it suits the needs. And hell, it might be the difference between you hiring a phenomenal French seller who knocks back five enterprises and gives you the case studies to get another 10. In France, maybe you won't. You probably won't, right? Unless you're French, then you're probably a bit more likely to do it. We're just like many and we're different, right?

7:55We're a continent with rich history. And that makes it a bit harder for some of these things to resonate as quickly as they do in America. So however you flip it, America's the largest market. You're going to have to do America well. So I think you just do it all, right?

8:09Nikola Mrkšić:And how are you thinking about building a footprint in America? We are predominantly American, right? Most of them are in the US and they're half the time. So I never built it as a company that had, say, the first 10 British logos and then expanded to America by hiring someone, etc. I've just been working for, well, ever since we had our first clients probably about four years ago. You know, I work till 2 a.m. daily. I'm effectively in America. Like where I am, people don't really know, right? I just set up camera lighting. Well, they can't really tell. I don't know if it's there or not where you are right now, right?

8:42Yeah. So with that, you know, just go and do work. It doesn't matter where you are, right? You have all you need to succeed in London or in America or like wherever. You're just going to have to do a lot of work. That's true. Yeah. If you're in America or if you're here.

8:56Nikola Mrkšić:Yeah, yeah, yeah. No, no, 100%. Very true. And I want to touch on your revenue. You said 82 % is coming from the US. You've seen amazing revenue growth over the last few years. I saw a post saying that you're on track to be at 50 million ARR at the end of this year. Is that right? Are you on track to hit that milestone? No, absolutely. That's amazing. And where are you, I guess, where are you seeing that come from predominantly? Is it large enterprise deals? Or are you seeing a combination of both large enterprise deals, but also some smaller players with a wider range? Where's it coming from?

9:29Look, it's a very interesting question. I'm a nerd. I look at our like, you know, splices of like pegments, verticals, everything. It works everywhere. That's actually the hardest thing about running a company like this, right? Where you could scale an SMB proposition. You could do upper mid-market, lower enterprise, strategic enterprise. We're well done for the kind of like strategic enterprise rollouts. You know, we work with the likes of FedEx. We work with Unicredit, PG &E, the largest utility company in America. So those we do really well. And it's really worth doing that kind of forward deployed model where, you know, your teams live and breathe, PG &E or FedEx, et cetera, to make sure that they, you know, I think we're the only one of our editors that currently has systems that do the work of at least a thousand full-time employees on behalf of a single app.

10:18Right. So I think that to me was, I've been to contact centers with a thousand people and, you know, it takes a few truckloads of food to bring lunch over for those people. Right. So now you think there's like a server humming and it does that work. That to me is like, that's real. Yeah, we get hype. Like, sure, there's some hype and we can talk about whether NVIDIA is priced right or not, like whatever. Like that thing doing the work of a thousand people, that's really real. So I love that segment. But it's equally true. You know, I think we have something like 3 ,000 different restaurants and pubs using us to be like kind of a client facing phone line for that.

10:54Nikola Mrkšić:3 ,000, did you say? Yeah. Wow. Wow, that's insane. Yeah. No, no, some of the biggest, yeah. Yeah, wow, that's amazing. Gosh, that's a huge number of people. I can't, do you have any data about the number of people who are speaking to your agents on a daily or weekly or yearly basis? Yeah, yeah. I mean, like, at the moment, I think we're on track for, like, that number. And it's always hard to identify and track and repeat callers and stuff like that. But, you know, I think well over 100 million people. speak to us every year right and that's way more calls than that because many of them are like repeat calls and could be higher i just can't you know there's like enterprise safety data and stuff where i don't of course get to look at it for every client in the right way but it's in that scale of things that's phenomenal i remember two years ago we crossed kind of like the call count that was like the british population at the time so like 69 million yeah so the next one was the us population i think we're right about that so yeah that's amazing and i want to ask you know you said that there's a lot of hype in the market at the moment you know you raised i think a 500 million valuation in your series c over a year ago i think it was may 2024 things at the moment are frothy right it's a crazy market valuations are sky high everybody's looking to get access not just to category leaders but categories in general you've seen amazing success right you've grown at a time when the market is also really popping off has that changed about the way that you're thinking about fundraising maybe you know thinking about raising more money sooner than you had anticipated to try and hire the top talent as talent becomes more competitive give you opportunities to do secondaries has it changed the way that you're thinking about fundraising well it hasn't really changed it i think you know frankly i think i've been very fortunate in that ai has always been hot i think previously it might have not been available to to all the people but if you came out of a strong research group and wanted to raise money it was always there right yeah so like the exact multiple is less relevant but yeah it's getting hotter and hotter um we're doing something now i can't share all the details now but expect to hear something soon but yeah no it's very hot right now amazing okay so it sounds like something is something is going on in the background but nothing you can share amazing very cool and other than the thing cooking in the background like what else are you working on like what is the top priority for you at the moment when you get up like what is the thing that you're looking to achieve either in the next few days weeks months man it's so exciting so i think you know for us like the the real suite of things around the agent itself you know we have clients where we do like 90 of their like calls or like text interactions automatically like what i'm really kind of like obsessed by is this notion of what will a contact center look like at scale in like three years from now and you know what it's probably like a tenth or at least at most one piff of the people half of them are going to be earning way more you know i think that we get people promoted we track this religiously internally like you know over half of our deployments have led to a promotion i think a quarter have led to a double promotion over the past years so like i think it's you know the contact center is this like often forgotten often neglected often abused part of the company that doesn't get the credit for like keeping the whole thing up and running right and i think as they get the right tools they really become the central point.

14:10They turn from a contact center to a command center where these AI agents are now doing all the work, right? And they're there sitting like air traffic control saying, okay, like the number of calls to Newcastle about people being overcharged have skyrocketed by like a factor of 2 ,000. We have an office issue there. And like this data has always been there. It's just been impossible to access it with kind of like distributed human agents. So now that you have AI doing it, you're able to just monetize. You know, most of the time when we automate, I think like at least two thirds of the calls, inevitably the next piece of work becomes sales, right?

14:44Like one of our top champions has recently moved from contact center to like sales there, right? And that's not something that was really written in the stars for a lot of these people. Because now it's like, okay, when the call is done, I'm going to program this thing to see how I can generate leads, upsell things, suggest new products, test interest for new products. and like guess what that person running that is now in a position to do it way better than so you know head of sales with 10 people right so the world's going to change in a very massive way and the people who've been learning how to do it for a year or two now are now worth a lot more than the people who have not been doing it for two years that's so interesting says people who

15:21Nikola Mrkšić:who made the decision to bring on you know poly ai as a provider you're seeing their careers take a different trajectory because of the impact that your product is having absolutely i mean look they're now like they've been given super you know they have an infinite workforce that is really high quality at their fingertips and they can do whatever right so they are no longer just there to hire just enough people for the whole thing to not collapse they're now they're asking questions of the org like what else do you want me to do you want me to sell you want me to upsell you want me to, you know, like fill out that CRM?

15:55You want me to fix up that backend data that's always been broken? Well, let me call everyone and figure it out. Now you can. Again, there's a limiter on just like how many initiatives you can be running, but you can do a whole lot more.

16:06Nikola Mrkšić:Yeah, when you think about the technology and where it's at today, how good do you think it is relative to how good it could get? I don't think there's a limit on it, right? I think we're far from like the best that it can be, but just in terms of like say conversational quality and stuff, you know, we're like pretty much the human level. The limit is not hallucinations. Hallucinations are not like a real thing. There's just bad context that you're not fed into the application, right? And then past that, you know, it's really just like what can you think of doing and trying it out, right? And does it make sense?

16:40And the other limiter, of course, is your enterprise architecture, you know, your integrations, availability of things like MCP servers for your agents to hook into. we see a huge wave of investment in companies large and small into getting ready for that future i mean that future is already here so uh you know there is a backlog there are limits on old systems that will continue to exist and be there especially in mission critical areas you know one reason that airlines have always been hard is that you know you often see two screens and the other one is that like you know like green screen with like something that's probably been running before you and i were born right and maybe much longer than that actually so you know those things are limiters to some of this technology changing the world but i think even there i think a lot of innovation will come from computer use and other things where we'll just be able to like again punch through those things with the power of ai and when you think about the future of poly ai are you

17:38Nikola Mrkšić:thinking are you starting to think or have dreams of other use cases for this type of technology or are you thinking this the whole customer experience market is so massive that we just need to go deeper and deeper into this technology in this market and this is where we build a multi multi-billion dollar company it's funny people like like like i'm not a contact center guy right i never really was our previous company was acquired by apple to make siri more conversational and before that i was a machine learning researcher right so to me i've the reason i committed poly are to doing a lot of the customer experience stuff is that it is the largest thing addressable by company that doesn't have its own hardware, right?

18:17And now that we have that volume of data, you know, we're like, we're not maybe a great power yet, but we're a regional power. We're serious, right? We have an amount of data, people, models, everything that is putting us in a position to just own part of that like communication. Really, it is conversations wherever they're happening. And you'll be speaking to your hardware. You'll be speaking to your software. where you'll be speaking to companies and enterprises. That's kind of where we're focused now. But, you know, 40 % of the interactions that we power have something to do with general service provision and revenue, be they like appointment making or recruitment or outbound sales, where it's like legally allowed, right?

18:58So when you do those things, you're really helping companies make money. And, you know, a dollar earned is a lot more enticing to a lot of buyers than a dollar saved. So that's growing a lot faster, right? And then, you know, I think that there's like a much deeper integration with a lot of verticals where, you know, the limit again is just like the ecosystem. And, you know, we like in the last year, we've seen our healthcare revenue nearly 10x. Wow. Like it went from being nothing to being close to number one vertical. And almost definitely by the end of this year, it will be the number one vertical.

19:31Right. And if it continues at that rate, it's going to be more than half our revenue by kind of like this time next year, maybe even sooner. So there are pockets of, and that's also surprising, kind of like what we said about enterprise being usually slower but faster now. Now we see healthcare just ripping our hand out with the level of interest we're seeing there. So you kind of follow the market. And again, the real limit is just how many things we can do at once because both the model capabilities and the interest from the market is such that you can do whatever and it's going to be fairly successful.

20:05Nikola Mrkšić:no that's super interesting to see that healthcare is um yeah being the early adopter and and and driving that change in such a yeah and such a unusual outcome you know this is not what people would have predicted five years ago i don't think well i mean that's the thing you know to be optimistic about because it will like mean that we eventually get a lot better at it and then we'll live in a land of abundance where at least you know healthcare will be like even even if we're on UBI maybe we'll help like other healthcare universally right yeah yeah amazing well look we're out of time um but thank you so much for joining me I love chatting I'm a huge fan of what you're building and I'm gonna be following your journey on and you know whenever you've got something to announce just let me know because I'd love to talk to you about it when the time is right yeah we'll do we'll do and thanks for having me always a pleasure lovely right let's speak soon bye-bye