Alex Levin is the co-founder and CEO of REGAL | AI-Powered Contact Center Software . He leads the GTM teams. Prior to that, Alex was a product manager at Personal and Thomson Reuters, and then joined Handy (acquired by ANGI in 2018) as an early employee. At Handy and then ANGI, Alex led growth and marketing. Alex grew up in New York and received his BA from Harvard.
What you will learn
- How AI agents are being used to automate and improve customer interactions in contact centers.
- The importance of building tools around AI models, rather than just focusing on the models themselves.
- How real-time data and A/B testing can be used to personalize and optimize customer journeys.
- How AI agents are increasingly handling routine phone tasks, freeing up human agents for more complex interactions.
- Why personalized conversations, even with AI, are crucial for building trust and understanding customer needs.
Transcript
Jeff Bullas
00:00:03 – 00:00:29
Hi everyone, and welcome to the Jeff Bullas show. Today I have with me Alex Levin. Alex is Co-Founder and CEO of REGAL | AI-Powered Contact Center Software . He leads the GTM teams. Prior to REGAL | AI-Powered Contact Center Software, Alex was a product manager at Personal and Thomson Reuters, and then joined Handy (acquired by ANGI in 2018) as an early employee. At Handy and then ANGI Alex led growth and marketing. Alex grew up in New York and received his BA from Harvard.
Jeff Bullas
00:00:29 – 00:01:16
Alex, welcome to the show. I’m looking forward to our conversations about how to use AI agents to help people in contact centers, and maybe do a chat around, uh, the general market of AI agents, which I certainly believe is the next big thing in AI if there’s ever a big thing in AI, but I think it’s one of the big things, and you obviously leaned into it, so. Tell me a little bit about it. How did you get to Harvard? I’m curious about that, because you wanted to go to an Ivy school league, um. Ivy League school Uh, was there a curiosity about AI, um, tell us.
Jeff Bullas
00:01:17 – 00:01:27
What happened years ago when you decided to go to Harvard because it’s a hard school to get into and it’s quite a, it’s quite, it’s quite the badge, isn’t it, to open doors for you.
Alex Levin
00:01:28 – 00:01:55
Yeah, just before I start, I’m not seeing it recording on my side, so I wanted to make sure it was, uh, OK, now I see it now. OK, never mind. So I’ll just answer the question. Yeah. Yeah, uh, so thank you for having me. I appreciate, you know, being here. So to go all the way back, you know, my parents really, I think, cared about education and, you know, I have two younger brothers and all of us, I think, did pretty well in school and all of us ended up going to, you know, top 10, top 20 schools in the country.
Alex Levin
00:01:56 – 00:02:16
You know, why Harvard specifically? Um, you know, I went and visited, you know, a number of schools that I thought were interesting, and I actually, uh, nearly went to a different school, another Ivy League school, and I had finished the application, was completely ready and like at the last second, like, you know, I had been in, in, um, Cambridge visiting friends and really, you know, liked Harvard and decided to go there instead.
Alex Levin
00:02:16 – 00:02:37
Or applied there instead, you know, I got in and was lucky enough to go. I think to your point, you know, one of the advantages always with those tops for 20 schools is it is a stamp that allows you to, you know, perhaps get started in your career a little bit more easily. You know, perhaps it is like, you know, uh, something someone looks at and says, oh, well, they must be at least a little bit smart. So that is helpful for sure.
Alex Levin
00:02:37 – 00:02:59
You know, but at the end of the day, the actual experience of college is entirely what you make of it. I had, you know, friends, roommates who did nothing in college at Harvard and got nothing out of it. And then I had other friends who, you know, went on to become Fulbright scholars and got a lot out of it. So, you know, you had to choose, like Harvard was not holding your hand and say, oh, do this next. Like some small
Alex Levin
00:02:59 – 00:03:21
art schools do that, but Harvard doesn’t care. Like, you do what you want. Um, so I think that’s one thing to think of. And the other thing to remember is Harvard, similar to other schools, has kids who are very smart, kids who are not so smart. Like, you know, they, there’s every sort that gets in. So, you know, uh, the range is pretty big. Um, and so, you know, in the end,
Alex Levin
00:03:21 – 00:03:42
of the day, like, same point, I suppose. Like if you’re gonna hire somebody or if you’re thinking about where to go, like, decide what’s right for you, not just sort of what the brand name is. And ultimately, you know, I studied, uh, liberal arts, which I highly recommend to anybody because I think it is sort of a good way of just learning and engaging. And then when I was going to graduate, I decided I didn’t want to go into academia.
Alex Levin
00:03:43 – 00:04:11
Like I studied philosophy and psychology in school, so I decided I didn’t want to be a philosopher or a psychologist. And instead, uh, decided every company would be a technology company, and I had to figure out how to go and, you know, learn about building technology. And so I was lucky enough to join some early stage companies as a product manager and start learning about, you know, what it was like to work with engineers, how you worked with customers, how you decided what features to build, like, you know, all those sorts of things, which have, you know, now served me very well.
Jeff Bullas
00:04:12 – 00:04:41
Yeah, it’s interesting, um, that you mentioned philosophy, and I’ve actually, I’ve heard that if you want to get into AI, you really need to start, you should be studying philosophy. Um, because I think AI is actually asking humanity big questions. Uh, what does it mean to be human if the machine can do everything, like write a story, write poetry, uh, write a great image. I’d be interested in your thoughts on that in terms of what the big questions that you think AI is asking of humanity is asking anything of humanity.
Alex Levin
00:04:41 – 00:05:06
Yeah, so it’s even stranger than me just studying philosophy. So I worked for a guy whose name is Steven Pinker, who started as a linguist, and then eventually became sort of very interested in evolution because of, you know, how did humans evolve language? How did humans evolve the right, the ability to think? And I ultimately studied within that even a smaller area, which is something that’s called philosophy of science or specifically like I studied the underpinnings of consciousness.
Alex Levin
00:05:06 – 00:05:30
So how do you know humans are conscious and how do you tell if a human is conscious or not and what does it mean conscious? That’s a very good thing that’s very conscious, you know, I think I’m pretty, I’m a pretty hardliner when it comes to this stuff. What I mean is, to me, like consciousness is sometimes described as not just thinking, it is the feeling of thinking. So sometimes.
Alex Levin
00:05:30 – 00:05:55
It’s described not as smelling, but as the smelliness of something. It shouldn’t make sense. So there are many computers that can smell. That doesn’t make them conscious. What makes humans conscious is that they can smell and appreciate that they are smelling. It’s the reflective sense that makes you conscious. So, you know, do I think, uh, you know, we’re anywhere close to machines being conscious? No. Like, do I think there are other animals that are conscious? No. Do I think there are many animals in
Alex Levin
00:05:55 – 00:06:19
Machines that can do things that are unbelievable, or even they’re better than humans that smell, see, think, and process. Yeah, for sure. Just the fact that they’re not conscious doesn’t mean that they’re bad or not useful or anything like that. So, the fact that they become conscious doesn’t automatically make them bad or anything like that. So it’s just that, that sort of um transition to the human species that we haven’t seen any other species.
Alex Levin
00:06:19 – 00:06:53
It is very unusual and very badly understood and unlikely to be replicated in AI at least anytime soon. Now, the really tricky part of it, if you want to get into the philosophy of it, is, well, how do you know if they’re conscious or not? Right? There was a big philosophical experiment where you’d say, OK, pretend there’s a zombie Alex. Exactly like Alex in every way, and when you ask it, are you conscious, the zombie Alex responds, yes. But you know for a fact that it’s not conscious. How do you tell? Right? So if, if, if there’s no way for it to self-report that it is reflexive.
Alex Levin
00:06:54 – 00:07:20
Yeah, there’s no way for you to know whether or not it’s conscious. So it’s kind of a moot point. So I think people should kind of forget the consciousness argument, ignore that, put it to the side, and focus on the practicality of how this stuff can be used. I think there’s really good safety and security arguments and I think ultimately they’re likely to be, whether it’s regulation or self-regulations, whether it’s government or self-imposed, constructs put around this so that people are not abusing it in the same way there’s constructs around.
Alex Levin
00:07:20 – 00:07:36
You know, other parts of the internet, other parts of technology, but be sort of the only thing that slows us down. Short of that, like, there’s, yeah, and I mean explosion of the use of this sort of more sophisticated reasoning capability in more and more places and more and more industries.
Jeff Bullas
00:07:37 – 00:08:32
Yeah. Yeah, what blew me away when I sort of came across Chat GBT, um, which is basically the friendly interface to AI. Um, In, you know, December 2022. Uh, I was just blown away and then looking at the data, just blown away by the velocity of this technology, and I’ve been in technology a long time. I’d, you know, my first sales job, a marketing job, was actually in, uh, at the start of the PC revolution. And I work for one of, you know, the largest IBM dealerships in Australia and that was exciting times. It was the wild west of personal computers. And uh, but AI’s velocity and acceptance by people is just amazing. I, I really, it’s, I think it’s gonna be the biggest change in human evolution that we’ve ever seen. Yeah, you know,
Alex Levin
00:08:34 – 00:09:02
so you can remember back to like, you know, a point when, you know, people went from, you know, not having a computer work, having a computer work, right, and the enormous change and you know, I remember when people went from no cell phone to cell phone, right, or a smartphone. Really, you know, and now we’re all gonna remember this moment we went from no generative AI to gene of AI. And yeah, it may happen even faster, but, you know, it is just amazing the things that I can do now that, you know, my father couldn’t do, you know, whatever it was 40 years ago when he was working.
Jeff Bullas
00:09:02 – 00:09:12
Yeah. Yeah, and for me too is that, then we had the PC, then we had PCs getting connected. That was called the internet and the web. And
Alex Levin
00:09:12 – 00:09:39
yeah, so I remember that. It’s funny, my first year in high school, basically there wasn’t, there was dial-up internet, but like nothing material by my second year in high school. It was really material, that was it. And so like I remember when we went to like a world of the World Wide Web and like everything was available. And so, yeah, it’s, you know, the generation that’s my youngest brother generation, like, yeah, had basically smartphones from the beginning. And so it’s a different starting point. Yeah.
Jeff Bullas
00:09:39 – 00:10:27
And I think I’m reflecting on that too as if we didn’t have the internet and the web, we wouldn’t have AI because AI needs large language models to really teach itself, amongst many other things. But uh it’s like the intersection of all these technologies to produce where AI exploded, which is what happened when we gave it to Friendly face in 2022 through Open AI. So, let’s lean into um but I do love the philosophical discussion because um I actually ended up interviewing um, Nick Bostrom, who wrote, um, Deep Utopia and Superintelligence, um, 10 years earlier, and, uh, he’s a polymath out of Oxford, and that was reading his books and having a conversation with him on the podcast was just amazing. So,
Jeff Bullas
00:10:28 – 00:10:54
His ideas of what could happen, um, from a big brain, and I felt it was really good to talk to a big brain. And sometimes I feel like my brain’s not quite as big as his, but, you know, we’re all different. But, um, it was, I, I’d spent a week actually researching for him and, uh, to, to speak to him and, uh, talk about philosophy, about, you know, being human and being a machine. It was really fascinating. So,
Jeff Bullas
00:10:54 – 00:11:21
But anyway, I digress, but the philosophy intrigues me. It’s sort of one of my curiosities amongst many others, but it’s a big one. So, Alex, you did all this work, you worked for these companies and you’ve learned in tech companies. So where did the idea for Regal.io to basically provide AI agents to help manage and run and be assistant to the humans in the contact centers, where did this idea come from?
Alex Levin
00:11:22 – 00:11:59
Yeah, so, you know, my background working with mostly as a marketer and as a product person, um, and I joined a big uh company called Angie that owns, um, you know, I’d say their best known brands are Angie’s List, HomeAdvisor, all the big home services brands. And we thought our business model was to just move everything for, you know, buying a fence or building a new home on the internet. So you’d never have to talk to another service provider again. And we’re wrong. So, yes, we’ve bought a lot of it online, but there are still huge parts of that flow where a person to person conversation, a voice conversation led to much better outcomes, built trust, built understanding why.
Alex Levin
00:11:59 – 00:12:23
You know, better quotes, better outcomes, so on and so forth. And so actually I ended up, uh, playing, you know, more and more of a role with the contact center, even though it wasn’t a world that I knew. And I was shocked at how far behind the contact center technology was compared to what we’re using in marketing or in product. And, you know, there were fantastic ideas among the people running contact centers, but the actual software was holding them back. So ultimately,
Alex Levin
00:12:23 – 00:12:46
Like we tried to do a little bit, but you know, and I even went to the current context companies and I said, hey, please work with us, you know, we were a big client. We’re probably a $30 million a year client, you know, so work with us, you know, no interest. So we ultimately started Regal because we were just so disappointed with the level of innovation in the contact center industry, and we focused at the beginning on, uh, how we use it.
Alex Levin
00:12:46 – 00:13:24
Two capabilities that the contact center doesn’t have yet. One, using real-time data to drive personalization or perfection of every interaction. So instead of just saying the same thing to everyone, make sure the conversation is specific to that person. And two, how to use AB testing in the contact center. So try version A, try version B, see what works better, which are novel concepts in the contact center, um. So that was the beginning and largely as human agents, because even though we had a lot of data, the generative AI wasn’t yet good enough for us to start using it in natural language processing, which was like the old version of AI definitely wasn’t good enough.
Alex Levin
00:13:24 – 00:13:49
To start actually, you know, having to do it instead of the human. It wasn’t until really about a year ago that we started seeing generative AI was good enough that, particularly in voice examples where we could shift instead of having the human as the tip of the spear trying to do something, uh the AI agent with the script and the objections and whatever it is. So, uh, we’ve started making the transition. I’d say most customers now are moving, I don’t know, 10 to 20% of their voice volume over.
Alex Levin
00:13:49 – 00:14:14
To, uh, AI agents to start instead of human agents and things, you know, going up from there. So it’s definitely not every interaction, but it’s the easier ones, the, the more time-bounded ones, the more clear outcome ones like that are getting switched. So, you know, scheduling and rescheduling, asking basic questions about my account, you know, helping with refunds or collecting payments from people, basic FA.
Alex Levin
00:14:14 – 00:14:58
Use basic troubleshooting, like those are great examples, you know, equal qualifying leads, where the AI does very well. And so, uh, now I’d say it’s, you know, the same two original points using data and doing AB testing, but instead of humans, like, you know, how do you have an omni agent world? And to us, that means both humans and AI agents side by side. Originally, I think I’d say when we started, my vision was that there would be a tool that helped the human agent. And I thought that’s where it was gonna go, but I was just wrong. The AI has gotten better much faster, and so now you don’t need a tool to help the human, which we have, you literally can just replace the human altogether with something else. So, in some industries like um healthcare or like
Alex Levin
00:14:58 – 00:15:14
Uh, the legal industry, they still just have tools to help the human because the AI is not yet good enough to do it all. But in the contact center, the AI is good enough to do it all. So, uh, we’re very quickly gonna see that transition away from human agents and towards, you know, AI agents handling the entire interaction.
Jeff Bullas
00:15:14 – 00:15:43
Well. Let me wind it back a little bit. So you’ve got the idea, you see that, uh, basically the contact center industry is ripe for disruption using technology. OK, you saw that. So where did you, where did you take the leap to cross the threshold? I’m using a Joseph Campbell, um, hero’s journey story arc here. Mm Where did you find the courage and how did you cross the threshold into starting Regal.io?
Alex Levin
00:15:43 – 00:16:07
Yeah, you know, my, uh, path was that I’d been in a lot of early stage startups. So I’d seen the sausage being made, so to speak. And I knew that, you know, that, uh, I probably had the skill sets or I thought I had the skill sets that the founders had. And ultimately, like, I hadn’t been in a place where I felt strongly about a business, an idea that I wanted to pursue for, you know, 10+ years. And so I went to work.
Alex Levin
00:16:07 – 00:16:32
For other people. But it came to a point where, you know, I had to make a decision. Do I want to go work for somebody else yet again, because I, I like early stage companies as a number 2 or #3. Or did I want to go and make a bet and spend time in a specific industry for the next 10 years of my life and really see where that and ultimately decided, obviously on the latter, you know, a big part of it was, uh, uh, my co-founder.
Alex Levin
00:16:32 – 00:16:56
Rebecca and I did it at the same time. So it wasn’t like I made a jump without anybody. So we made it at the same time. And we had a little bit of financial security because we had just had a pretty good financial outcome from the last company that we’ve been at. So, you know, doing with somebody having a little bit of financial cushion helps massively. It wasn’t like I was sitting there eating ramen out of a cup. Like me, I would like to keep a roof over my head for some period of time.
Alex Levin
00:16:57 – 00:17:34
And then, you know, you know, it’s about spending time. So we spent a lot of time digging into it and really convincing ourselves that it was a good business and that, you know, we listed all the big objections and the reasons we thought it was gonna fail, and immediately went and tried to convince ourselves whether it was going to succeed or fail. And when we had convinced ourselves it was a big business, that’s when we went out and started raising money, hiring people, getting customers, whatever. But I think people often do it backwards. They often think that the big shiny object is being a founder, so they want to be a founder, even if they’re not ready. They go and raise money before they’ve convinced themselves that it’s a thing. They get a couple customers and they go, oh shit.
Alex Levin
00:17:34 – 00:17:50
You know, I don’t really like being a founder, but I have the free people’s money, and I don’t know what’s it, you know, so that’s a bad situation they’d be in. So just do it in the right order and I think in the end, you know, we were lucky, like it was, we were right. It was a business. We got customers, we started growing, we started hiring great people, and started.
Alex Levin
00:17:50 – 00:18:12
you know, raising money, and, you know, have, have sort of continued to progress. You know, the, the fun part for me and like the, the maddening part is, you know, every day you come in and you, you know, look under a rock and there’s a new opportunity, and you have to be that personality type that just isn’t, you know, uh, uh, excited by that, enthused by that. And you go, oh, OK, that’s great. There’s another opportunity, we’re gonna go and do that.
Alex Levin
00:18:13 – 00:18:30
And if you have that personality, yeah, it’s very good to be in an early stage company. If you over, you know, turn over the rock and you look again, there’s another problem. Oh my God, when will it just end? Like don’t go to early stage companies. I know a big company where everything is solved and you just have to keep the, you know, the planes running on time. Yeah.
Jeff Bullas
00:18:31 – 00:18:58
So what are the biggest challenges you’ve had, um, you know, you obviously bootstrapped it to a point, and then you said we need some more money to actually make it. I, I suppose, to be number one in the industry, I suppose it would be your goal. Um, what were the biggest challenges you’ve had along the way, um, so you cross the threshold, you’ve got money, you can put, you know, food on the table, roof over your head. Um, so what are the biggest challenges you’ve had since that start?
Alex Levin
00:18:59 – 00:19:33
Yeah, you know, I’d say, uh, the journey as a founder is always sort of, you know, an interesting one. You know, there’s, it’s a roller coaster. There’s high highs and there’s low lows, and you have to learn how to like, you know, even it out and make sure you’re continuing to move in the right direction. Uh, you know, there’s also with the founder of the Real difficult decisions between when do you listen to feedback and when you ignore it and say, no, I know better and I’m gonna keep going, right? So you go to a customer and you show them a thing and they say, I don’t like it. Do you stop doing it? Or do you go, Well, no, I, that was the wrong customer or I pitched it wrong or whatever other reason, and I’m gonna keep doing it.
Alex Levin
00:19:33 – 00:20:00
You know, that combination of certainty and listening is really hard to get right, um, but important. And then, you know, at some point when you’re a little bit bigger, obviously then there’s other challenges, and we’ve had all of them, like, you know, the wrong hiring or the wrong uh customer type or, you know, building the wrong features sometimes, or, you know, having the wrong priorities internally and I, I’d love to say that like as a founder that we’ve, you know, me and my co-founder have gotten it all right, but we certainly haven’t.
Alex Levin
00:20:00 – 00:20:36
Uh, you know, and I think it’s the speed at which you’re able to recognize that you made a mistake and the ability to acknowledge that it was, you know, our fault, and, and ultimately nobody else’s, like, you know, it’s like bad happens, we’re the ones that caused it, and then the ability to quickly have sort of change management internally where you can move into the right direction. And the people who are able to fail fast and more often and are willing to get smacked in the face and like to keep on doing it again, are the ones who are great founders, right? You know, the ones who are too afraid to try it or, you know, don’t, you know, don’t wanna move it even though they know it’s not quite working right, yeah, that’s where it doesn’t work as well.
Jeff Bullas
00:20:36 – 00:20:58
Mm. So the other question I have around that is, uh, you come up with the idea and then you design a product and uh of course there’s the lovely acronym MVP minimal viable product. uh. So, did you build one of those? In other words, you created what you thought were the key elements, or you went all in and built feature rich from day one?
Alex Levin
00:20:59 – 00:21:19
Yeah, so we’re even more extreme. So not only did we not build feature rich, we didn’t even build MVP. The first version was like literally I’d made a deck with like 8 to 10 slides with screenshots or basically designs of what I thought the product would be. Because for all purposes, that’s all I needed to be able to go and have conversations with customers. I didn’t need a live working product.
Alex Levin
00:21:19 – 00:21:39
So forget MVP like the customers who spend a year or the companies who spend a year making an MVP only to show that to customers later, you know, I’ve wasted a year because you’re never going to get it right. So, you know, I come back every day from having talked to customers and go, oh yeah, like Rebecca, my co-founder like, you know, I learned this, I learned that, or this is working, or here’s a different customer type or whatever else.
Alex Levin
00:21:40 – 00:22:18
And you know, it’s the way I always explain to people if you’re lucky on the 8 or 10 slides, there’s one slide where people are going, that, like I need that today, give it to me, you know, no matter the price. Now you have a business. If, if nobody is reaching across the table and saying that, you don’t have a business, go start from scratch. Like, if they’re being faintly nice or maybe there’s something interesting or I’ll introduce you to somebody else. Yeah, you’re not, you don’t have a business and you know, don’t go to friends at the beginning, like go to people acquaintances will be honest with you and tell you, do you have that thing or somebody wants to buy it, um, but yeah, I think, uh.
Alex Levin
00:22:19 – 00:22:58
Yeah, the less you can build from the beginning, the better. And like we took that approach to an extreme, you know, but then you have to very quickly focus on building that MVP, the first version of it getting in people’s hands and entering from there. And in our industry where there are competitors that are 2030 years old, to get to parity is hard because even if we are underlying infrastructure in the way we approach it is quite different. We still need to be able to check the box and say, yes, we have SMS. Yes, we have voices, yes, we have whatever features. And so it took us, you know, 3 years, let’s say, to get to parity, maybe even 3.5 years to get to parity with those platforms because they had 30 years of building things.
Jeff Bullas
00:22:58 – 00:23:25
Yeah. But reaching, competing against the big guys has both its advantages and disadvantages, doesn’t it? So they’ve got the money and the market share. Yeah but they’re basically dinosaurs, which is what you saw. So let’s move on to the next little thing that curious me about what you did with OK. You showed 8 features and one or two features were where they almost reached across the desk and said, I want this now.
Alex Levin
00:23:26 – 00:24:04
Very early, I think people understood the idea of like, how do we use, we had a concept of the journey builder, the journey builder to, uh, you know, change what type of calls were happening and texts and, you know. Some are gonna be branded names, and some are gonna come from a, you know, different number, and some are gonna have different messaging, like people like that concept very early on, you know, and then we had to, uh, at the beginning, like, there wasn’t a self-serve interface for customers to do that. So my girlfriend and I were the ones building journeys for customers like we could do on the back end, but there wasn’t even a self-serve product for the first like year or year and a half basically.
Alex Levin
00:24:05 – 00:24:33
Um, which is funny to think about, you know, and then I’d say more recently, you know, you see the same thing with, you know, AI voice agents where, you know, people are going, yes, I know I need that. Like, the only question is like, which vendor am I gonna use? Like, that’s the place you want to be in, where they’re in market, they want the thing, you know, and you get the opportunity to demonstrate why you do it in a different way than other vendors in the market, and then the customer can make a decision about what approach they want, and whether they think your approach is the best.
Jeff Bullas
00:24:34 – 00:24:56
Right. So, in terms of AI agents, are there any platforms you use, uh, because basically AI it’s almost being broken up into very small niches or verticals, aren’t they, really? So it’s being applied to just a niche. What technology did you choose to actually build your technology on an app, um, or is there a range?
Alex Levin
00:24:57 – 00:25:23
Yeah, so we built the end to end, you know, uh sometimes we call it the contact center as a software or the operating system like we built a CAS platform, which is the core software, whether it’s a human or an AI agent that you need to decide, you know, who’s making the call or taking the call, uh, you know what are they saying on it, what is, what happens afterwards, you know, what are next steps, all that stuff. Um, we then built uh
Alex Levin
00:25:23 – 00:26:08
You know, a lot of the capabilities around like what the AI agent does, but we don’t like that we are not an LLM provider, right? So we are not in the business of raising millions of dollars to go and try to optimize an LM. That’s what, you know, meta and Anthropic and Google and OpenAI are doing. We partner with them, so we have contracts with all those players. And our customers can use any of them too, you know, and there’s the ability to customize them, obviously, for specific industries, but that is a big advantage. Like in the same way, like we don’t run fiber optic, you know, under the ground to like our customers, we use the existing fiber players and our customers sometimes bring LLMs with us, with them. But so we use existing providers.
Alex Levin
00:26:08 – 00:26:34
for that. What we’ve really focused on is, you know, what are the tools you need to build AI agents, to test AI agents, get them in production, you know, AB test, see what’s working, see what’s not, you know, manage them. Those are things that we do very well. And those are things that the LM providers are not as focused on, you know, they’re not making what are the tools for contact centers to use AI agents and movies and whatever other 10 other examples. They’re focused on.
Alex Levin
00:26:35 – 00:26:58
You know, what is the API endpoint that we offer everybody that they can all use. So I do think there’s a lot of opportunity for workflow specific tools. Even within that, you know, we focus on certain industries and certain examples within those industries. So it gets very, you know, when a customer comes in and they’re an insurance customer, there’s 3 or 4 things that we think AIG does amazingly well in their use case and we focus on.
Jeff Bullas
00:26:59 – 00:27:32
Right. So one of the big challenges that’s been raised too now is uh, Basically proprietary IP that’s basically locked within a company. Um, OpenAI is open AI. In other words, it sits in the cloud, you can access it and so if you share it with the cloud, it becomes everyone’s, uh, IP. How do you manage that? Because a lot of, I’m sure some of your clients must be saying, I want to make sure that my IP is not lost, is not given away. How do you do that? Because I think we’re realizing now that we need to make sure that, Um,
Jeff Bullas
00:27:33 – 00:27:50
You know, with the web, we gave everything away to the web, basically or most of it. Um, but, you know, that’s the easy part. Basically making it works the other part. But how do you protect your customer’s IP, uh, within your industries, um, and then I’m gonna ask you which industries you’re focusing on, but that’d be interesting.
Alex Levin
00:27:50 – 00:28:16
Yeah, yeah, it took a pretty strong line. So all of our agreements with the LM providers are very clear that they’re not allowed to use our customers’ data to train their models and clear that while during the interaction we’re giving EI agent a specific piece of information like where are the customers from, all that gets erased as soon as the interaction is done. They don’t store the data in any way. So we are truly just using them as a provider of a specific service. OK.
Alex Levin
00:28:17 – 00:29:01
And that, you know, alleviates a lot of concerns because yeah, customers don’t, are very worried that their customers’ health data or security data or whatever will accidentally end up in the LM and then like it’d be a bad thing. So yeah, we just took a very strong line. There are some providers that will still, you know, say, hey, we, you know, we want to use your data to train our models, but honestly like it’s not necessary anymore and it’s very expensive and so that sort of model of Working where you use the existing calls to train the models as going out and sort of the new sort of ways like no pre-training, like we, we’ve already done enough with the models to get them in a place where we don’t need to pre-train the model specifically to your company.
Jeff Bullas
00:29:01 – 00:29:26
Right. So the other big question raised, which sort of comes out of social media’s history, is do you, um, everyone trusted Facebook until they didn’t. In other words, Cambridge Analytica is a perfect example of that. So they’re telling you they’re raising the data. Um, do you think, obviously you could have trust about that. So, is there any way you can test that?
Alex Levin
00:29:26 – 00:30:06
Yeah, well, so contractually like their, uh, Facebook’s contracts were very different because it wasn’t a contract with the company, it was a terms of use with an individual, and they changed it so they could use the data. So was Facebook actually violating their terms of use from an end user? I don’t know. In this case, we have a contract with a company, so it’s company to company, it’s very clear, it’s a different standard. If they were to violate that, there would be huge lawsuits around it because of that. So there’s some legal financial protection, you know, do we check like we can and most of the contracts we have the right, like we could go and do an audit and check, but I’d say like,
Alex Levin
00:30:06 – 00:30:29
We believe that these actors like Google or Facebook or whatever, are doing the right thing in this area. Like I don’t think that like they are trying to go and do the wrong thing, you know, at the very beginning where they sometimes like scraping YouTube when they probably shouldn’t have, sure, but like they got their hands slapped and now they’re like their contracts are much clearer, what are they allowed to use and what are they not allowed to use for training?
Jeff Bullas
00:30:30 – 00:31:01
Yeah. It is a fascinating area. It’s one of trust and uh. And the big boys might go, well, you know, we’re, we’ve got, we’re worth $1 trillion sue us. But anyway, that’s another story. Um, so AI agents, let’s go back to that, um, generally, so AI agents, tell me if I’m right or wrong here, AI agents are something that actually can come up with an idea, use data and actually then act on it. So in other words, it’s not just information curation, collaboration, distillation.
Alex Levin
00:31:02 – 00:31:30
Yeah, I go even beyond that. There have been systems that can act on data for a long time. The difference with the agents is the historical, historic systems were deterministic where you had a rule. You said, when I give you 1, you do B. When I give you 2, you do Z. It was deterministic, right? And just, you had to be really specific with it. Generative A is not deterministic, it’s stochastic, right? If I give it 1, it might.
Alex Levin
00:31:30 – 00:32:07
Give me different things, right? And it’s like a human, you can teach it that the right answer is B, but it’s not always gonna just do B. It might say, well, let me think about that. I’m not sure there is an answer. Well, if the answer it could be a B or C and well, you know, whatever. It is, you know, a different way of approaching the problem. And so I think. You know, there’s some worries that come with that like what if it gets it wrong? What if the training is bad, you know, uh, what are the legal implications of like, you know, not being deterministic, so on and so forth. But the consequences when you’re a human interacting with it, it feels much more natural. Like it’s not the
Alex Levin
00:32:07 – 00:32:26
Uh, you know, IVRs of days passed, we would call in and it would go, press 1 for this, press 2 for that. Like, and if you said one syllable wrong, you go, I didn’t understand you, right? The new ones are much better at understanding and much better at comprehending and then having an interactive response with you. Yeah.
Jeff Bullas
00:32:26 – 00:32:35
So what other industries do you think AI agents and uh AI agents with what you’re doing, I suppose just the start of this change, isn’t it, in terms of
Alex Levin
00:32:36 – 00:32:58
there’s two sorts of things happening. One is, what are the industries where AI is truly replacing humans period. And I’d say the contact center is one of the best examples for that, because it’s already been highly, highly, uh, um, analyzed and we have tons of historic data showing. like when the customer says this, here’s what the AI agent’s supposed to do, and we, you know, have very bounded interactions.
Jeff Bullas
00:33:01 – 00:33:01
Son, isn’t it? Yeah,
Alex Levin
00:33:02 – 00:33:46
it’s a perfect thing for AI agents to do it instead of humans, um, and, you know, allow humans to do other work instead. There’s a lot of other industries where the actual interactions that are very valuable require humans, but there are all sorts of things. Humans are so expensive that you don’t want them to do simple things. So take lawyers or take, you know, software engineers. A lawyer, you know, needs to make judgment calls about whether, you know, certain indemnity is OK or not, right? But what they’re spending all their time doing is marking up two different documents to see what the differences are and trying to manage comments. And if they spend an hour on that, it’s an hour they can’t spend thinking about the actual business implications of that change. So having AI handle all of the markups.
Alex Levin
00:33:47 – 00:34:11
Ups and, you know, call outs or whatever, so the lawyer can actually think about the more complicated question, uh, save a lot of money, and, you know, make sure that they’re working on actual important hard problems. You know, same with computer, you know, engineers, like if an engineer spends an hour looking up some function that they forgot, it’s a waste of time when they could just type it into like, you know, some system and say, hey, can you remind me how do I do a date function that does that.
Alex Levin
00:34:11 – 00:34:36
X Y and Z and it just spits out the thing. It may not be good enough for you to just put that in production, but you can certainly use it as a way of improving what you’re doing. So, you know, I think there are industries that split. Ultimately, you know, I’m optimistic that AI can do more and more of these things that we still think humans are the only ones capable of doing. And most people, you know, most companies, most people who work in companies will be like a giant HR department.
Alex Levin
00:34:36 – 00:35:00
Yeah, right, I mean is they’re gonna be hiring AI training AI, you know, firing AI, coaching AI, like, what do you call that a manager or HR, like that’s gonna be the function, not just of IT but of everybody in the company, like a great software engineer will know how to manage an AI software engineer the same way they might know how to manage a human software engineer today. Yeah.
Jeff Bullas
00:35:00 – 00:35:09
Cool. And Context senses, like you said, are a perfect recipe for that disruption. And, uh, have you got any numbers on how many people work in contact centers globally?
Alex Levin
00:35:09 – 00:35:26
Yeah, there’s $11 trillion spent on that, you know, versus there’s probably $40 billion spent on the software. So much more, you know, many more dollars are spent on the actual labor than on the software.
Jeff Bullas
00:35:26 – 00:36:01
Yeah. So the other question we talked, sort of, in my mind, we’re thinking a bit more about inbound, you know, contact center customer service that the AI is gonna help you with, which is gonna be different answers that, you know, basically helps the human provide the right answer, because, you know, AI has got almost a perfect memory. Um, so let’s talk a bit about outbound in terms of outbound sales calls and maybe outbound text marketing. Do you guys get involved with that yet or is that, yeah, so,
Alex Levin
00:36:02 – 00:36:42
and I ask about industries we work specifically in industries that are higher consideration, um, high touch, so think of again financial services, um, think of healthcare, uh, local services, education. These are not like buying pencils where you can just return it if you get one. Like you’re actually making a decision to go to some university and pay a lot of money, spend a lot of your time trying to, you know, improve yourself. So, in those industries, communication and interaction, whether it’s with a human or an AI, is so important, you know, and it’s not going away anytime soon. So like those are the types of use cases we’re very focused on, where, yeah, AI is enormously valuable to what’s going on.
Jeff Bullas
00:36:43 – 00:36:52
Are there any rules around outbound, uh, calls in the USA regarding, you know, and text messaging? Um, in other words, outbound sales calls and,
Alex Levin
00:36:53 – 00:37:31
very, very strict. So at least, on the consumer side, it’s very strict, right? The consumer needs to give you what’s called Expressri consent. So when they give you the phone number, they have to check a little box and say you have the right to text and call me uh from an automated telephone dialing system and so on and so forth. And only then are you allowed to go outbound. What’s hysterical? Less hysterical for me, but I can appreciate, uh, uh, how, uh, you know, weird it is, is on the business side, there’s no rules. So, because I’m considered a business person, when they get my cell phone on the business side, they could call me all day long, and, you know, they, even if I tell them not to, and I
Alex Levin
00:37:31 – 00:37:49
I have no control whatsoever. So the, the rules of the United States were built around consumer companies and there’s this gray area in businesses that needs to be fixed and eventually will be where, you know, you’re not gonna be allowed to just randomly spam somebody unless uh you have their permission.
Jeff Bullas
00:37:49 – 00:38:04
Right. We have similar rules here, uh, I think maybe might be a little bit tighter because uh in Australia than the USA in terms of uh, Do not call uh list, I suppose, but you do have to, if you tell them verbally I think that’s enough or or a text on
Alex Levin
00:38:04 – 00:38:17
on the consumer side it’s all very strict in the US. It’s just, you know, like I say on the business side. You know, I get a million calls a day because my, I’m the, you know, I run a business. My phone number’s out there.
Jeff Bullas
00:38:18 – 00:38:31
Cool. All right, just maybe a couple of hours of questions just to wrap it up. But, um, what does success look like for Regal.io? And what does success look like for you as well?
Alex Levin
00:38:31 – 00:39:02
Yeah, so it is an important conversation to have with your founder or co-founders in the beginning. You know, we want to make sure like your expectations are aligned and like one of you isn’t looking for X and the other is looking for Y. Um, so, you know, because remember, the #1 reason companies fail is because of co-founder, uh, breakup, and maybe the #2 reason is running out of money and then number 3 is everything else, but you know, co-founder breakup is up there, so be, be aligned, you know, for, for the company, you know, we really are out to help our customers, so we track all kinds of metrics around how much it cost.
Alex Levin
00:39:02 – 00:39:23
Customers are able to use our product to drive better outcomes, save money, drive more revenue, drive more customers. So that’s ultimately like why we’re in this is we think it’s an interesting industry where we can really have a major impact on the way our customers are actually engaging with their end users. Personally, like, yeah, you know, I think Michael Finn and I enjoy the day to day.
Alex Levin
00:39:23 – 00:39:52
Work, like, we wanna make sure that like we as a company are moving forward, you know, we think there will be a financial outcome at the end for all the investors and employees involved. And so personally, that’s important to us is that the people who are along the journey with us, like, are able to get something out of it. Um, but yeah, those are the sorts of things that matter to us. I think, you know, as long as we like, we continue to serve customers well and innovate and, you know, we continue to have fun with the team like we’re in a good spot.
Jeff Bullas
00:39:52 – 00:40:26
Cool. One last question, um. In terms of uh. You know, as an entrepreneur, um, you cross the threshold, it was easy, sort of, um, in terms of you had enough money, you, you had, you know, basically a good idea, a disruption of a legacy industry. What are some of the things you learned along the way as, uh, an entrepreneur that you could share with our audience? Top 2 or 3, just, um, what’s your biggest learnings, um, that you’ve learned along the way?
Alex Levin
00:40:27 – 00:41:04
Yeah, always, you know, do this with the co-founder. I don’t know how people do it all by themselves and like try to do, you know, whether they’re the only person. Like, my co-founder and I are fifty-fifty and everything. We have different parts of the business that we run. We sort of trust each other to make decisions and move more quickly. Wouldn’t work otherwise. So that’s definitely one. I’d say, you know, 21 of the bigger mistakes, and it’s funny like we always prided ourselves on hiring, but there were times where hiring was hard and we couldn’t find the right people, and we ultimately settled for certain, you know, hires, and it it always, always, always bit us, never settle.
Alex Levin
00:41:04 – 00:41:39
You just don’t do it. Just like, continue to slow down or continue to not do the new thing, like, do not hire the wrong person because having the wrong person means you’re having the wrong culture internally. It means you’re not having somebody who has the same, uh, desires and the same sort of work ethic. And so, ultimately, it hurts everybody else when that sort of bad apple is in the bunch. So just whatever is your standard on hiring, don’t lower it purely. Like, figure out how to offer more money, uh, change the role description, whatever it is, but figure out how to get the right quality talent.
Jeff Bullas
00:41:40 – 00:41:42
How do you know when you’ve got the right person in front of you?
Alex Levin
00:41:43 – 00:42:11
Yeah, you have to have a bar. So like you, you know, you know, in every case on our side, it wasn’t a question of whether we knew what the bar was. We knew what the bar was, we couldn’t find somebody who met the bar, and so we lowered the bar, and that’s where you put yourself into trouble. Look, if you don’t even know where the bar is, go, you know, go to a company where you have the opportunity to hire and fire and learn and figure that out. But again, our problem wasn’t knowing where the bar was, it was. Lowering it instead of waiting till we found the right person.
Jeff Bullas
00:42:12 – 00:42:14
So being patient’s very important.
Alex Levin
00:42:14 – 00:42:22
Particularly with hiring, there’s many things you want to be impatient about, but the hiring just kills you, yeah, when you have that wrong person in the seat.
Jeff Bullas
00:42:23 – 00:42:25
And it’s not fun firing people, is it?
Alex Levin
00:42:26 – 00:42:36
No, it’s, you know, it’s not even if they leave of their own volition cause they understand it’s not the right fit, it’s just, it hurts so much and creates so much delay when you have the wrong person in that seat.
Jeff Bullas
00:42:37 – 00:42:44
Yeah. So I’ve got one final question, which is more a personal one. If you had all the money in the world, what would you do every day?
Alex Levin
00:42:45 – 00:43:16
Uh, so my co-founder sort of asks me sometimes or she tells me like, hey, even if we were to sell this, like you would go back and do it again, wouldn’t you, Alex? It’s probably true. Like, yeah, I’d take it a month off, 2 months off, whatever, but I, yeah, I would probably have to go back and go back to an early stage company either as the founder or as one of the executives. That’s just what I like doing and, you know, what I find, you know, engaging. I think like I’ve always told myself like, yeah, I’d like to go do X, Y and Z. That’s not work, but I think in the end, I kind of like work.
Jeff Bullas
00:43:17 – 00:43:39
Well, the right work actually fulfills us as human beings, I think. And uh. Um, and for me, it’s also I. I think the question I ask myself is what I’m most curious about and then follow that and uh then work’s no longer work, it’s actually play, and that’s for me. So what are you most curious about,
Alex Levin
00:43:40 – 00:44:10
Alex? Yeah, I think right now, I’m very interested in AI agents, like how this is gonna change the way companies interact with customers, particularly when it comes to voice interactions. So that’s obviously a lot of my mindshare. You know, there’s other areas, you know, outside of that, like, I’m very interested in international trade and like how we improve what’s going on and sort of manufacturing, you know, I’m very interested in, uh, you know, a lot of the asset like business models that you see out there that are transitioning sort of the the the the.
Alex Levin
00:44:10 – 00:44:28
Way in which we use assets. So instead of like somebody’s car sitting around doing nothing for two months, maybe that person allows someone else to borrow their car. Like that’s actually a better use for everyone. So, yeah, there’s other business models, you know, and, and lines that I find interesting, but at the moment, yeah, I’m pretty focused on one.
Jeff Bullas
00:44:28 – 00:44:56
Well, it sounds like you’re making a roaring success of it, and, um, look forward to hearing updates about, uh, how things are going, but it sounds like you’re actually doing very, very well. And, uh, thanks, Alex, for sharing your story and, um, what, uh, gets you up in the morning and what keeps you curious during the day. So thank you very much for sharing your stories and, uh, your wisdom and, and experiences. And, uh, thank you very much. It’s been an absolute blast.
Alex Levin
00:44:56 – 00:44:57
Thank you for having me.