#533 From Code to AI Teammates: How Artur Rivilis Is Reimagining Logistics and the Future of Work

[00:00:00] 

Mehmet: Hello and welcome back to a episode of the CTO show in Mehmet Today I am very pleased joining me from the west coast of the US Art Rivilis. He's the CTO of Augment. The way I love to do things is I don't like to take much from the [00:01:00] spotlight of my guests. So this why I usually keep it to them to introduce themself, but art, before I give it to you, uh, I start to do this and people start to like it.

I give teaser of, you know, what we're gonna talk about today, Uhhuh. So what, what do you expect, uh, us to, to, to talk about? So we're gonna talk little bit of course, about our journey. We're gonna talk about like. You know, being a CTO and of course solving problems, uh, art, you know, have done some m and as merchant acquisitions from tech perspective.

So we gonna shed some light on this and of course, like we're gonna see how the flow would go and we will follow it. And of course, you know, we'll not miss anything without talking about AI and building engineering teams. Great. So without further ado, art, I'll, you know, give it to you, tell us more about you, your background, your journey, and what you're currently up to.

And then of course, we can take the discussion from there. So the flow is yours. 

Art: Yeah, absolutely. So, um, you know, how, how early to start? I think, uh, you know, I think for, I've always [00:02:00] really had a lot of love for tech. I remember when I was, you know, maybe, you know, I ca I came here as an immigrant, spoke zero English, uh, to the states, um, when I was, you know, seven, like seven or eight years old.

And, um, I remember kind of like one of my earliest memories was. Uh, my teacher, elementary school teacher at the time, um, where I, I knew I could barely speak, but, um, she had written this sort of, uh, note to my parents saying like, um, we really like having art in our class. And, um, he really likes computers.

Um, and, uh, my dad had, uh, gotten me like a very, uh, kind of, you know, someone had given him basically this, uh, computer with like, dos on it. And, um, it had basic, uh, this was like, you know, it was like nine or 10. Um, and I was just, you know, at the time it was very [00:03:00] simple, uh, but I was, uh, you know, would hack away on it.

And so for me, really early on, I knew what I wanted to do. Right. I just like, I really loved. Building things and being able to make something out of nothing, uh, from scratch. And it was, um, it was something that just really gave me a lot of satisfaction, just, you know, spending a couple of hours on, on making something that was just useful for me or for someone else.

Um, when I was in, uh, you know, as early as middle school, I learned Visual Basic and I built, um, this quiz software that, um, all of our biology classes, uh, ended up using. Um, and, uh, and it was like super cool. So the, at the time that was not, that was certainly not a thing, uh, kind of like, you know, computer administered tests.

Um, and uh, it made it super easy for the teacher to grade and I was just, I was like, wow, this is so cool, and how powerful, uh, this is. Um, so I, so for me it was [00:04:00] really clear when, you know, when I, um, I couldn't go too far from home, so I ended up. Uh, you know, really kind of applying to just schools in California and specifically Berkeley, which was the closest school.

Um, you know, lucky for me, that's also where I had met my wife. So it was a, it was a good decision all around. I like to think I, and, you know, and I focused entirely on, uh, you know, entirely on computer science and math and, uh, the two things that I really, honestly just loved. Um, and it's hard to love them because at, uh, you know, uc, Berkeley, it's, they're in, at, at least computer science is an incredibly rigorous program.

You're kind of put through the meat grinder with lots of other students. And so it's really hard and it's like a lot of work and a lot of sleepless nights. Um, and, uh, you know, there I really focused on something that seems very unconventional, but I [00:05:00] worked with, um. Uh, Jan Stoica really liked network programming and, um, Jan Stoica was, was one of the founding, um, one of the founders of Databricks.

Uh, but he was a brilliant guy and I really loved, uh, really loved working with him. Uh, and I was like, at that point I was like, so sure I was going to go into building distributed systems. Uh, 'cause I found them to be so, um, so fascinating at the time. Uh, and then, you know, this was, I started looking for internships and, uh, and it was a kind of a tough market.

This was, you know, right after the dot.com bust, uh, you know, it's like 2004. There just wasn't a lot of jobs. And I ended up landing a job at VMware, uh, doing front end stuff, which of course, like totally didn't marry up with the plans that I had, right? Like, I was like, oh, I'm, I'm very, I'm like, I'm sure I wanna do distributed systems.

Couldn't find any internships. [00:06:00] That had anything to do with that and ended up doing front end work. Um, and, uh, but you know, it was like, I, I was up for new challenges. It was, it was quite different. But it, uh, one of the things I did really like about it was that, um, I could build something that people u that people touched, right?

That people like used, um, with their own two hands. And it wasn't something that was kind of like behind the scenes and that no one really saw. So it was a kind of an exciting, an exciting thing for me as a kind of a change of pace. 

Mehmet: Mm-hmm. 

Art: And, uh, um, and you know, after VMware my, um, my boss at VMware had left.

Started his own company and he recruited me to join his company. And uh, and that was like my first foray into startups. Mm-hmm. That was kind of like the beginning. Um, and I, you know, and I had this sort of contrast, this experience of like the ware, which was already a fairly, I was post acquisition, but it was fairly [00:07:00] large, fairly established company.

Um, where I was working on one kind of very narrow thing, um, to this kind of open world of startups where, you know, like it's, you're just, there is nothing, right. You're starting from nothing. And so everything you build is meaningful and impactful. And, uh, and I kind of got addicted to that. I just thought that was a really, uh, really exciting and fun and I pretty much just stayed with startups from then.

Um. And, uh, and yeah, that's, that's, that was kind of the beginning. 

Mehmet: Uh, that's a very cool, um, journey. I would say art, like, um, trying multiple things before you, you, you get what what you find with fill out. Yeah. E exactly. And you know, all the names of the companies you, you mentioned are familiar to me. So I'm not a developer, but I'm on the, I was on the infrastructure sta slash networking side, right.

So, so whether [00:08:00] it's like VMware or even later, you know, of course Databricks is also now one of the most growing companies of the world. And it's a cool, it's a cool, you know, to be, to be in the heart of tech. But you know, moving forward in your journey to, of course, and we know this like once you are a developer or an ENG software engineer.

So you spend your time solving problems, right? So how did you find, you know that? Okay. I like solving problem, but I need to be on, on, on like maybe the, the, the leadership side. So moving toward the CTO VP engineering and it's not an IT path. So, and the reason I keep asking all my guests who are like either CTOs or like VP of engineering, the same question because every time I'm surprised by [00:09:00] the answer like why someone decided to go to this path rather than, and because I know brilliant guys who become, maybe not CTOs, but they go and choose to become CEOs.

Sometimes you see they stay in the architecture, even if they take like a leadership role, they still like code by their, by their hands. So for you, like what was, you know, the, the, the main reason to choose the path of becoming A-A-C-T-O? 

Art: Yeah, I mean, it's a great question. I, I have to say I never really like planned on that career path.

Um. I was always a tinker, always a, you know, an engineer, uh, in my heart. And I, um, uh, you know, like the transition, like, I think the transition for me happened. Um, so kind of walking through my sort of like career story. So I did that startup. Then after that, I, uh, went to Flixter, um, which had like a, this is in the [00:10:00] crazy sort of like Facebook, uh, app years, which had like a super popular movies app, both on iPhone and on uh, Facebook.

And then had acquired Rotten Tomatoes. And so, and then after that I joined this company called Symphony. And, uh, at Symphony I was about 26, 27. There was a cir, a circa there. And, um, I, you know, was sort of one of the first engineers, uh, at the company. And uh, and that was kind of like my sort of like heads, sort of heads first into, uh, continuing my startup journey.

Um, and, uh, I, you know, I was kind of very, um, you know, uh, focused on delivery and on getting stuff done. And um, and I had a really kind of clear sort of clarity of mind of I like how to prioritize work. Um, [00:11:00] part of that was that, you know, the thing that really attracted me to startups was that, um, you know, if you go, you know, I had, um, had multiple forays of interviewing at larger companies and especially earlier in my career.

And, um, and I would get an offer, be like, oh, this is a web developer role. And then I would be like, oh, okay, can I. Do product Is there is, is that like an option to maybe do product part-time or, um, you know, can I like, like drive a project with an eng with a group of engineers? I don't need to manage them, but I'd like to like, you know, drive the project and always like, no, you're like, you're this level and this is the expectation.

We're expecting you to just code. You know, just only really do majority, like, you know, design and build and that's it. And um, and for me that was kind of not that interesting. Um, and uh, primarily because I was always so sort of like connected and attuned to what is it that we're building, why are we building it, what value is that [00:12:00] creating?

And um, and to me that was like intertwined with development. I never wanted to be in a position of where I was like building something that didn't make sense or that like someone somewhere else kind of had determined and I couldn't really control and had no kind of like input into, especially because I had like such a strong opinion.

So when I was at Symphony, what that, what that meant was that I, you know, typically had really good sort of ideas of what we needed to build and then how we needed to build it. And I would, for the most part, I would take on a lot of that work myself. Um, and, um, but I would also provide guidance to others, right?

Others on the team. This is before, you know, before I had any kind of management role or any, you know, or anything. And I was always inter, I was, I'm an introvert. Um, and I, you know, I don't think I ever kind of like for myself thought about like, oh, I'm gonna go into a ver into a role [00:13:00] that, um, you know, where, where I'm gonna have to, like, is a people, people job, right?

Like, I think management, management and leadership, they're intertwined and it's, it's a, it's really oriented around it. Um. Work, like very intimately working with people and working on their, you know, development and their career progression and things, uh, things of that nature. And so I never really thought about that.

But, um, I had folks around me at Symphony that encouraged me and they're like, Hey, you know, like, okay, like you, like of all the people in this company, like you have really sort of clarity of mind of what the priority should be, what needs to be built, how do we build it? And like, you should, you should do this.

And to me, I was like surprised because I was like, you know, I, I was always kind of like my own sort of like harshest critic and you know, kind of had like, uh, oh, do I know what I'm doing? You know? Um. And, uh, but I had all this kind of [00:14:00] external validation, right? And, uh, and so ultimately, um, this was a sort of a controversial move.

The one of the, the, the CEO founder of the, of the company, you know, effectively made me the, uh, the lead of the team, which at that time was small as maybe 10, 15 people or something like that. And, um, and it was a little bit of a controversial move because they were like folks that were, you know, I was 26 and there were folks that were way older than me that were like, oh, this is weird.

Like now you're managing me. And it was very kind of awkward. It was awkward for me, awkward for them. Um, and also I, you know, it's hard kind of like jumping into a management role because in a startup you don't necessarily get the appropriate sort of like, levels of mentorship. And so I had to really like, learn on the job.

What did this mean? Uh, but um. And this was kind of like my, you know, I went from engineering manager to VP of engineering and this was kind of like the track of, you know, [00:15:00] of um, uh, you know, kind of rising to the occasion of what the company needed. Um, and, uh, taking, uh, taking the role. And I think at the time, like I also like, made a number of mistakes.

I, I, you know, I was, it's hard to be the first, especially in a business that doesn't have that kind of role, right? Like the deep mentorship. So what I started doing was I just started reading a lot. So I like, you know, I, I was like, okay, and this is obviously before kind of the dawn of chat GT and all of that.

So lots of online research. I picked up a bunch of books and, um, and kind of like started formalizing, like, what, how do I think about what, what does this mean? What is, what is my leadership style and like, what are the things I value and, um, what is gonna be my particular take? At this, at this job. So I took it.

So I was like, okay, like move, remove myself away, away further from code and then figure it out. Like, how can I provide [00:16:00] leverage to the team? Like, how can I make folks on the team more effective? How can I, you know, take a one X engineer, make them a two X engineer? How do I make sure that we don't build the wrong thing?

How do I make sure that we don't create technical debt? And so I started thinking about, like, systemically across the team and focusing more and more of my attention on that. 

Mehmet: Cool art. Like there is something which, you know, it looks like it's happened, you know, in, in, in, I would say, I'm just looking at your link in profile.

So let's say in the past, uh, since 2011, so that's how many 14 years? Yeah. So, so, so there is a, so there is a theme which is, uh, you know. Moving to a company that get acquired and, you know, acquisitions, uh, on all levels aren't easy, of course, from business perspective, but from from technology perspective, you know, like [00:17:00] how hard it is, you know, to prepare yourself and the team for what's next, especially if, if you are to be acquired, I mean, as a company rather than you acquiring another one.

Because, you know, I, I, I spoke mainly to people who did both. So if I am the one who's acquiring, there's kind of a standard book there, you know? Yeah. Yeah. We need to do the technical due diligence. We need to do this. We cannot start to think how we're gonna integrate. But when you are on the other side of the table, how does that look like?

Art: It's messy. Um, I think, I think we, it is an interesting journey and I learned a ton. Um, and, uh, we didn't just get acquired once. We got acquired twice. Um, so, uh, so at Deliver, so, you know, from Symphony we started deliver. It was, you know, um, doing D two C fulfillment, uh, for non Amazon, [00:18:00] basically Amazon Prime for non-Amazon world as we'd like to call it.

Um, and that business, uh, you know, grew really, really quickly and we were acquired by Shopify. And then, uh, you know, a year afterwards, um, uh, Shopify had, uh, sold all, all of its sort of acquisitions, uh, uh, in the logistics space. And, uh, specifically for kind of like deliver, they sold it to Flexport. So we then ended up kind of going through this process twice.

Um. I don't know if I'm the only person that that's happened to, but it was certainly, it was certainly quite a ride and quite an adventure, I think from a technical perspective. Um, you know, I think that, so I think that there's really two sides to it. I think the first is, or what makes it integrate, you know, uh, like, uh, technical acquisitions challenging.

I think the first side to it is, and I, I really like, um, uh, the book, uh, innovator's [00:19:00] Dilemma. Mm-hmm. And it talks a lot about, you know, acquisitions and why they work and don't work. And it comes down to like, um, this sort of, uh, the sort of, um, uh, you know, idea that, uh, you know, there's like, companies are sort of like purpose-built, like they're successful in a particular industry based on kind of certain operating principles and certain values, right.

It's just that, you know, they happen to figure out like, what's the right. Formation of like what we should be really focused on and where our, our attention should go and if they make the right call. And a lot of it is just some, some, some probabilistic it's, there's timing, there's like ability, all of this, all of these kind of things.

But if they make the right, um, call, then they, then they're able to like really access that opportunity and grow within that space. And, um, what that, that extends into development as well. Right. Uh, and so, uh, the har one of the hard [00:20:00] things about, you know, acquisitions is that the, you know, there's a particular way you work as a company, and that means the way you work as a, you know, as an engineering team and, um, the, you know, and the, and the acquiring company also has the way that they work.

And oftentimes it's going to be different, um, especially if they're solving different problems. So in the case of like a delivering Shopify deliver was a, you know, a a, it wasn't a SaaS company. It was, you know, we had physical a, there was physical assets, there's warehouses, a warehousing, cross docs, um, uh, you know, there's, there's managing trucks.

Uh, there, there's a whole many different facets of the business, but it's a low margin business. Um, and then with something like, uh, Shopify's a high mar, it's a higher margin business. It's, it's, it's like a SaaS, uh, SaaS business for small businesses. And so, um, those two things are just gonna be [00:21:00] built, uh, built differently.

And so, you know, one of the kind of, um, kind of, I would say like quick, kind of one of those differences was, and by neither one of these is right or wrong, but it's kind of like each one of these is purpose built for the problem that, you know, each individual company's trying to solve. Um, the, one of the, one of the, one of the differences that we, uh, that we had, um, uh, you know, experienced was, um, with, uh, with logistics, you have to be incredibly close to the problem.

You have to be sitting at the facility and building software for the operator on the floor. And the reason for that is because like operators on the floor, they're different than, you know, kind of like your common sort of, you know, whatever, like, like an Instagram user or something like that. And 'cause they're, they're, they're doing like physical work.

And if you don't model the world [00:22:00] correctly, they're not like, you could build all you want, it's not gonna work and they're not gonna use it. Um, and so you have to have a lot of empathy of what they're trying to accomplish and then like, work within kind of like their, you know, sort of like their, um, their boundaries.

And then, um. Uh, you know, whenever you're building SaaS software, like you obviously need to understand your user, but it's not at that level, that level of intimacy, right? The other thing is that, um, in logistics, because it's a low margin business, you every single scent matters. And so you tend to sort of like obsess about like, okay, we have this particular process, it's a game of all of these sort of like op optimization algorithms along the way, and we need to, you know, every single decision on like how you distribute inventory across, across, you know, particular geography, um, has a large implication on the economics, the unit economics, um, uh, for the, uh, for the [00:23:00] business.

And so there is like, you know, this sort of like meticulous thing of of, of kind of measuring every single thing and then looking for every single little opportunity to improve. And then making those tweaks and then measuring it again, and then making it again, and a sort of this like very, very fine micro optimization over time.

And whenever you're in a higher margin business, that's not something that's, that's, that's as, uh, as valuable, um, as an example. And so, um, and I think that that's, uh, you know, kind of some of the, I would say some of the, um, challenges that, that we experienced from a kind of a product engineering perspective.

Um, but I think the hardest thing is that I think in any acquisition, you, uh, you know, you're, you have an integration. Mm-hmm. And, uh, what that means is that you need to, you know, come in and, um, [00:24:00] uh, you know, and, and whatever plans you had up to that point, whatever you, whatever the roadmap looked like, whatever, you know, whatever you were trying to accomplish, you kind of need to like.

Now you're, there's, there's a thesis behind the acquisition of like, how is this all gonna fit together and how is it all gonna work? Right? Like, what is, you know, the, the experience gonna look like for, um, you know, for the customer. And in, uh, the, uh, in the Shopify acquisition, the, there was, um, an existing, uh, application within Shopify, uh, for fulfillment, right?

And, uh, and our kind of goal was like to take that, that interface and then like build it on top of, you know, kind of our API deliver mm-hmm. Or XAPI layer. And, um, and that's a fairly large undertaking. Um, and so I think, you know, kind of [00:25:00] navigating, okay, like what does this, you know, you, you have like existing engineering team.

You have this new engineering team that's coming in, navigating like who does what and. Who's responsible for, you know, like for each one of these components and like how do we get them to work together really well? Um, all become really big challenges. Um, and, uh, and also just because of those kind of like different sort of like, uh, you know, kind of like working styles and stuff.

Um, and so I would say that that, you know, that's kind of one of the hard things. Like integrations are hard. Um, and, uh, you know, we spend a decent amount, you know, with an each acquisition, you know, Flexport as well. We spend a decent amount on like, okay, here's what, what is here, right? There's always maybe some the complimentary sort of like set of services or complimentary, you know, um, uh, [00:26:00] you know, a complimentary layer.

And so doing that integration I think is, you know, is hard. 

Mehmet: How is the feeling that all of a sudden, especially when you know the team is small, and then now, uh, you are in charge of a much bigger team. So how that transitions looks like? 

Art: So I think, I think in the case of, um, in the case of, uh, Shopify, so our original, you know, team was around, um, you know, about 150.

Mehmet: Mm-hmm. 

Art: And in, in deliver. Uh, so that was my, you know, it was my org is product and engineering and design. And, uh, in Shopify, I forget the exact, uh, the exact amount. It was a larger team. Um, and then I think when we, when everything was kind of like merged together, uh, I had about 300. So it was twice the amount.

Wow. And so I think the, [00:27:00] the hard, the hardest part right, is that I think in. In the context of the acquisition, you know, like, um, you're coming in and you know, no one really knows you. They don't know your character, they don't know what you're about. You know, are you trustworthy? And it's very vulnerable because I ended up leading the team.

But, um, you know, like folks that were there, you know, before we arrived that were building a kind of a, a, a, you know, competing product, um, they, you know, they'd be like, oh, well are you just gonna favor everyone from your team? Which is a supernatural kind of like, uh, you know, it's very understandable bias, right?

It'd be like, oh, okay. Are you, I'm, am I gonna be successful here? Right? And can I trust you? And so I think what was really, what was really hard was, um, you know, how do I build the trust and. I think, [00:28:00] and this was also during like, kind of during COVID and so it was even more, even more difficult. Um, but what I did was I, you know, I honestly, I, um, you know, I'm, I'm not in it for ego.

I'm in it for building something that's really useful and that people really like, like I just, that's goes back to kind of the origin of why I love being in this, you know, space and being in this business. And so I, you know, I got on a plane, I went and I visited, you know, every single office, uh, you know, like all the, uh, and tried to make as much face time with folks.

I spent a, a lot of one-on-one time. Um, and, and part of it was just kind of like taking, you know, kind of taking things down and making sure that, hey, this is not, uh, this is not, um, uh, you know, this is not a kind of a thread. I, I'm, I'm really just trying to like do the best to like do the best for everyone, to represent everyone and not.

You know, represent. I don't have any kind of my own interest. [00:29:00] Like, I don't, I just, I just wanna do the right thing. Um, I think from a, uh, from a technical, uh, technical perspective, there were some challenges. Like, uh, you know, Shopify was, was on top of Ruby. We were using a TypeScript stack. Um, and so you kind of have like, you know, especially once you integrate, you know, for a certain point we were just two separate teams, and then we had to itegrated together.

And then once you integrate teams together, you, um, you know, it's like there's this unfamiliarity of the stack and what's the patterns and, um, you know, and how are things, how are things done? And that was something that was like an adjustment, uh, adjustment for folks. Some folks just like transferred out.

They're like, oh, I don't, I don't wanna, I don't wanna do this. And that was okay. You know, I, I, I totally respect that. Um. So, uh, yeah, I would say like that, that was the, um, you [00:30:00] know, and the one thing though is I, you know, if I think about kind of like, you know, once you get to about a hundred people, one 50 doesn't feel that different than 100.

And then 300 also doesn't feel that different than one 50. And I say that because ultimately, um, you know, let's say that you have, like, you know, at a hundred, you have like 10 teams of 10 people. Usually there's gonna be a little bit more, and they're gonna be smaller. But just like, let's say roughly do the math, then, you know, at like one 50, you have to have, you have to start introducing some intermediate directors.

You have to start grouping because, you know, you're, I'm not gonna be able to manage more than more than 10 teams. So, you know, at that point you have lieutenants, you have directors underneath that are basically saying, okay, you know, you have. You know, three teams, you have three teams, et cetera. And usually I like to organize teams by domain.

So it's going to be like, okay, like what is the business problem that you're so solving? And I'm gonna give you kind of like, [00:31:00] you know, singular ownership over that business problem. I'm gonna give you all the data scientists you need all the front end engineers, you need all the backend engineers you need, and you have everything, almost like a gm.

You have everything that you need in order to, you know, like set goals and then hit goals. Um, and you can't come and say, oh, I couldn't do it because, oh, this other team, they, you know, they, I, I, I was waiting something. And so it was like very sort of like important for me that there was, you know, taking a step back.

Sometimes you have cases of where like, um, nobody owns something or where you have more than one person that owns something. And both of those are super problematic. And so for me it was always like. I, you know, um, there's functional organization, there's domain organization. I've always favored domain. I think it works better, especially for, um, when you're dealing with, um, a B2B software.

But then from there, like it's always about, okay, every team needs to have [00:32:00] one singular charter. It has to be exclusive from all the other teams. There should be no overlap. So I think whenever you take that, you know, if you have like, you know, let's say that, um, we have like 300, um, uh, you know, 300 people, let's say that, we'll, we'll just do simple math.

I have, let's say eight directors, uh, is gonna mean each director has about four teams. 

Mehmet: Mm-hmm. So 

Art: for me, I'm still dealing with the directors at the end of the day. Right. And it's like, maybe I'm, I have more directors that I'm working with, but I'm still working with directors. 

Mehmet: Right. 

Art: Um, I think the challenge of that, of course, is that the more indirect, the more people you have in between you and the frontline teams, the more, um, the more challenge you, you lose visibility, you lose visibility, you lose control.

And so I think then you have to be really judicious of where do you [00:33:00] want more of that? Right? Right. But going from one 50 to 300 is a game of which, which of the folks that are reporting to me are performing really, really well and hitting their goals? Who can I rely on? Who do, who do I want to give more teams to because I have high trust in their ability to, to execute.

Mehmet: Mm-hmm. And then 

Art: that's, that's, and then how do I get more of those people? Right. 

Mehmet: Uh, you mentioned the number a hundred because, uh, in a previous episode, like, uh, with my guests, we were, uh, and this is, you know, a discussion that I did many times about, you know, what's the magic number when a startup, you know, should, should change something, right?

And I'm not talking about the number of engineers, you know, and, but it applies also to an organization. Indeed. 

Art: It's an organizational number. 

Mehmet: Yeah, yeah, yeah. So, so we're saying like, okay, you know, we're talking about something related to culture and something related to [00:34:00] communication and like, we're just trying to guess.

Of course we know that. I mean, there's no one answer, but I mean, should we say like. 50, should we say a hundred? You know, when things, you know, it's not like goes out of control, but to your point, you need to delegate. Yeah. So just, I remember we were talking about de delegation, right? Yeah, yeah, yeah. We were talking about, you know, about, uh, the famous saying like, uh, uh, you know, you should scale what I forget.

Like it's, it, like, don't, don't, don't do what doesn't scale something like this. Right? So, um, and it's a magic number. Like you, you, yeah. And I, and I 

Art: think it's like, it, it really is a function. I do think it comes down to like a function of like, there's a couple of factors into it, right? So like one factor is how experienced are the ICS and the, the engineers on the team, because like you can kind of flex ratios more up and down depending on like how, how independent [00:35:00] folks are, right?

So if you have like a team of like. You know, junior engineers, then it's gonna be obviously, like as a manager, you're gonna, you're gonna be very hands-on and you're gonna have to spend a lot of time and you're probably not gonna be able to have 10. Right. So, you know, so I think, I think it kind of makes an assumption that, you know, you, let's say you have a, the, the teams are fairly stacked with experienced engineers, then you can stretch those and experience like man frontline managers.

You can stretch the ratios, I think to a hundred, like where it's a hundred, a hundred people, 10 people report to the vp and then all of them are frontline managers. It's great because it's very flat. You have full visibility. You can provide really, um, really good feedback. You can identify patterns across teams really quickly.

You can make sure that two people that need to talk to each other will talk to each other. Um, and you can root out inefficiency. You can, you know exactly like there's no silos effectively, because you [00:36:00] have a purview of, of all 10. But if you have more junior people, then maybe the ratio is like, oh, you know, maybe it's like eight, you know, per team.

And so it's, the number becomes 80, right? And so, so have experienced managers, let's say your managers aren't, like, you know, as experienced, well maybe it becomes 72, um, you know, or something like that, right? So I think, uh, that, that's kind of like, you know, how I would, you know, how I would like think about it, right?

It's, you're right, and there's no exact number. It's all kind of like situational dependent on like seniority and experience levels, but it's somewhere between, you know, I would say probably between 70 to a hundred is that magic number where you kind of have full visibility, full control, um, and is, you know, the most, the most efficient.

Mehmet: Cool. Now. Uh, because you said and talked about, uh, solving business problems. So tell me, what [00:37:00] problems are you solving now at Augment? 

Art: Like, yeah, augment, I mean, it's, um, super quiet. I think, you know, in my, you know, I've been, I've been in this space probably for, you know, around 20 years. And I have to say, like, I don't think I've been as excited about something that I've built as, as I am about, uh, augment.

And, um, and I think particularly because, you know, we're in this kind of like really kind of cool and revolutionary time with, um, you know, the advancements of these gen AI models and the things that are possible that previously were very hard to do. Um, and, and so like, you know, with Augment, what we're building is in AI team.

Name is Augie Pence. Augment, augment agi. Um, and uh, and I say teammate, it's, it's, it's a bit intentional because there's different sort of, [00:38:00] you know, kind of, um, uh, solutions out there and tools and, uh, you know, for, so I think like one operating model is you have a copilot, right? A copilot is built to support somebody, right?

But they're not built to be autonomous. 

Mehmet: Mm-hmm. 

Art: And some of the developer tools that we're seeing, like, you know, like, um, cloud code for example, they're kind of starting to go, you know, and Devon to be, to be very precise, are going in the direction of like more, becoming more and more autonomous. But, but that's an important distinction.

Um. And, uh, and they're not like, and, and Augie isn't a kind of like a point solution where, you know, you're focused on solving, you know, it's a fixed workflow focused on solving one, you know, respective sort of problem, answering just one email or doing, doing one phone call. And, um, so, and it's an AI teammate, just go back, it's an [00:39:00] AI teammate for, uh, for the logistics space and specifically for, um, uh, for freight, freight and trucking.

Mehmet: Mm-hmm. 

Art: This is, I'm sure is like a very kind of unfamiliar, uh, you know, world for a lot of people. Um, that's one of the things I love about logistics is that it's, it's, it's quite niche. Uh, it's the hipster in me, but, um, it's, and it's also like a, uh, it is obviously really like sort of country dependent.

US is, is is quite large and, um, it, uh, you know, relies on like obviously, um, there's many days of, of, of, uh, transit between two locations. 

Mehmet: Mm-hmm. 

Art: Um, the other interesting thing is that there are, uh, about a million carriers in the us, uh, 3 million truck drivers. So most on average, most carriers or trucking companies are quite small.[00:40:00] 

And, um, and so because of that, there's just a lot of fragmentation and non-standardized platforms. It's, no one is quite using or has converged on like one thing. And there's been many efforts to do this, but no one's converged on one thing. Um, and so, you know, any type of automation is really hard, um, to do in this.

And you know, in this space you have basically shippers, so think of Pepsi or Nike, and they want to get, you know, their stuff from A to B, right? And then, uh, you have the carriers as I discussed, and then you have brokers, which are service providers typically to shippers, to basically get them better rates, uh, get them more capacity, um, things like that.

So what, uh, augment is, is looking to solve is basically, um, you know, just like any team member in a, in a [00:41:00] brokerage or a carrier, um, Augie's able to, uh, accept emails and answer them, uh, able to make phone calls, able to text and receive texts and understand kind of intent and be able to communicate with external parties.

And, um, and it's able to follow a, um, uh, a particular workflow. Like, how do I solve, you know, you can think of something like a tracking problem, right? Like, oh, this driver's supposed to pick up this, you know, this load from, you know, uh, this warehouse and drop it off at Walmart, and I need to make sure that they're coming on time.

So I will likely need to give 'em a phone call if they're late. There's all of these things that need to happen if, if, if, um, if they're late. Now with that example, you can imagine that, well, you know, um, if they arrive late, there's [00:42:00] customer specific rules would reference it as knowledge, but there's customer specific rules.

So like, oh, if it's a Pepsi, you know, uh, shipment, well, Pepsi would like to know immediately, and here's the person that you need to contact. 

Mehmet: Mm-hmm. 

Art: Right? And so, um, so Augie operates on these set of skills. This uses workflows to accomplish a particular task. And then uses knowledge along the way in order to, uh, to, to solve, to basically solve problems, uh, and to make better decisions.

And so, um, uh, and with that, we, you know, we're able to take these very, very manual, very difficult, um, processes that, uh, you know, brokers, carrier shippers, uh, are, uh, have to do today, which a lot of it is, you know, the protocol is the English language in this industry. Um, and so we, we can take a lot of these, uh, you know, a lot of these [00:43:00] processes and we can basically have AGI do them.

Mehmet: Uh, I was looking on the website, I was fascinated, you know, by, by the things, you know, like, uh, it's like it brings, it's, and I'm not sure if this is why, the reason you call it like this, so it's like augmenting. The, you know, the, the, the technology to behave as it's our, as you mentioned now, kind of a coworker, so the machine becomes, as an agent, I don't know what you call it exactly, or like becomes, uh, you know, part of the admin team maybe who should make sure that these shipments, you know, are carried, delivered, uh, correct technical.

Correct. Right. So, which is fas fascinating to me now it's, you know, built on ai. Now I'm not, and I'm assuming like [00:44:00] logistics, it's a, it's a niche, right? And so how much was. Uh, hard or easy because it's, it's one of the businesses which is, has been around for a long time. 

Art: Long time. One of the oldest May, maybe one of the oldest businesses in the world.

Right. Um, 

Mehmet: yeah. 

Art: And fifth in the US it's the fifth large, the fifth largest, uh, industry. Yeah. 

Mehmet: Yeah. What, what I'm curious about, like, we hear about some businesses which like, usually they are not sensitive, but you know, they don't adopt new technologies fast. So how was, you know, the acceptance of people in this business to, it's kind of a revolutionary product and service Yeah.

That you are not disrupting in the positive sense, of course, the way they operate. So how was like, you know, how they receive this because it's built on AI and you know, and the reason I'm asking you Yeah. Because [00:45:00] AI nowadays, because of all the noise that happens, it carries a lot of. Question marks, you know, for, for business.

Hundred percent. You, anything, 

Art: you, I think we, we've seen this with every sort of like technological trend, right? There's always like, there's an adoption curve, there's a lot of skepticism that's, uh, that's upfront. Exactly. Yeah. I think one of the things that makes this industry quite interesting is that, um, is that folks are like, like folks that actually work in this industry, this is true for a, for a, for a lot of brokerages, true for a lot of, uh, carriers as well are incredibly entrepreneurial.

Mm-hmm. And they are looking at what is gonna give them an edge in the market. Right. The other thing is that like, we're in the freight industry, in the US we're in a, I would say kind of a recessionary environment. 

Mehmet: Mm-hmm. Uh, 

Art: and so, you know, so margins are tight and I think folks are looking at like, okay, like how do I, you know, [00:46:00] improve my margins?

And, you know, and, and, and so there is, so there's, that's the, the secondary factor here. So one is super entrepreneurial looking, always looking at like, how can I do things better? How can I make the, make this, make, make my existing process better than it is? Um, how can I generate margin improvement so that, you know, I can grow and grow and get bigger as a business.

And then, um, the last piece, which I think is, which I think is interesting, um, and to me a little bit unexpected is that, you know, I think, uh, chat GT has like 800 million users. And so the adoption of this technology, given how how new it is, is very high. 

Mehmet: Massive. Yeah. And I 

Art: think that like, you know, I think one of the most interesting things is that, you know, previously we had like, you know, the Turing test right?

Was sort of like the gold standard for when we've, when we would have hit an inflection [00:47:00] point. And I think, and, and you know, and I think this technology passes that test, right? And so I think everyone kind of collectively recognizes that, oh no, this isn't a fad. This isn't just, you know, smoke and mirror, right?

This is real. And, and so I would say that, you know, given the kind of the usage and the fast adoption of this technology, the realization I think from folks that work in the industry, the business owners that are like, okay, well this is very interesting. And, uh, and, and we like, and it's like if we, either we adopt, like either we adopt it, right?

And then we get ahead, or our competition adopts it and then we fall behind, right? And so I would say like that's, you know, kind of like the, kind of like the mental model. And so I would say that, um, you know, even though there's that initial skepticism, I think with anything. [00:48:00] Uh, what we've seen at least is like, you know, kind of this acknowledgement, right?

That this is, this is happening and, and it's there. And, and it's really like, how do you best apply it? And how do you build, you know, with this technology to solve a specific problem, like specific vertical problem, which is really the hard part, right? Like, um, and so, um, and so that's, yeah. So it's, that's what what I've seen.

Mehmet: Now, out of curiosity, and I'm not sure if you or might not have any plans when, you know, when you were describing all, uh, the, the use cases, let's say that you're solving, I mean, there are probably pain points, frictions, you know, which, you know, o augment, Oggie, uh, you know, try to solve. How easy is it to take this and apply it to different verticals?

Art: Um, I think it [00:49:00] depends on the, uh, complexity of the vertical. There's a couple of, there's a couple of sort of like, um, I think, uh, challenges to be, to building any vertical business. And, uh, one of those challenges is sort of like the, so the wide swath of integrations in the ecosystem, right? Right. And so, um, so I think that's, uh, kind of like one challenge, right?

So like if you, and like in order to build a really good solution, you need to have very deep integration, right? Mm-hmm. Um, I think the other thing is that, uh, I had talked a little bit earlier about like knowledge, right? About the, the, the, the, the data model, the ontology of, um, of logistics. So. I think it's considerably reduces.

Like if you, if you have [00:50:00] like, Hey, here's a canonical set of data. You can model it as a graph, as a set of like, um, you know, SQL tables, but a really well documented data set that, uh, you know, an agent can query and pull data from in order to make decisions. Um, it's much easier for you to solve a new use case than to start completely fresh from scratch.

Because right now, like if you start, start a new agent for a complete, let's say we're like, Hey, we're gonna build this in healthcare or something, right? Well, you know, outside of all the integrations, all the, you know, health record management systems and all that kind of stuff, um, you would need to know like, oh, okay, like what are all the important entities and what are all the important fields and like, what are the relationships between those fields?

And like all of the immense amount of domain, because in order for you to be able to, you know, kind of like both make decisions as well as. Um, answer, you know, in, in the course of this, you know, like Augie, [00:51:00] for example, is able to answer questions about what happened, right? Right. Why did it happen that way?

And like, you know, in, in a business rel you know, in, in a logistics relevant world, right? Uh, religious relevant questions, I should say. And like, you would need to then apply the same thing to a completely different domain. So it's not, so you could do that. It just, it would be, it would be a lot of work.

Mehmet: Yeah. And of course, like to your point, there would be some, uh, what we call them moving parts and the more moving parts you have. So that means more processes. More processes means like maybe multiple, um, I would say. Factors that plays the role of deciding what happens. There is a lot of players probably that they need to be consulted, right.

So, yeah. But, but, but I mean, as architecture, or let's say, [00:52:00] I'm not sure if architecture is the right word here, but I mean, the concept, let's say, let's say the concept of, I'm not saying it's easy to, to, to duplicate it, but it's, it, it, for me, it's like the baseline that we can look at and probably maybe because also logistics.

And this is why, you know, I, I always, uh, myself, I was fascinated by the way, anything that has logistics in it, regardless, like it's not fright only, like it can be anything that has logistics means like you have like, it's like a puzzle, right? And then you say, when this happened, and to me logistics is very similar to coding in, in, in some way, like.

So when this, there's a condition, is this, do that? So it's like a Yeah, 

Art: there is, there is some, you know, there is like, I think every business is gonna have some amount of like, you know, logic. Logic like a business process, right? Or SSOP or what have you, right? So [00:53:00] like, how do I do this job, right? And then, uh, with that, it's gonna be what are the tools to do this job, right?

So I think, I think it's, um, what I would say is, I think for like simple, like, you know, there's sometimes there's things that are really simple, like, oh, you know, I get this email, I look things up, and then I reply, you know, something that's quite, quite straightforward. I think, um, you could, you could apply the same system to something like that, right?

Whenever you have something that's a complex workflow. I'll give you an example. Like, um, you know, if you're doing a phone call and that phone call has. Uh, you know, you're asking for someone's location, they give you a location. Um, well, like transcription on voice calls is like, is not quite there yet. And so the location might be kind of garbled.

You have to kind of like underst understand based on like route, where [00:54:00] is this, where is it located? Like what, what is this city? Um, so that I can actually capture where this person is and uh, you know, and so you have to build software for that, 

Mehmet: right? Right. 

Art: And so like, now if it's a simple, you know, I get a call or an email, I look something up, I give that back.

And there there's, and it's, and it's, you know, kind of like straightforward. I think that that's, then yeah, you can take the same architecture and build towards that and, and, and do a good job of it. Um, but the second it starts to get really kind of like, you know, like it gets complex, then all of a sudden you have to, it's, it's traditional software development.

And so like, it's not like, you know, we're not quite there where, um, you know, I think if we get with the models, I think if we get to a point where unsupervised they're able to generate accurate and correct code, then we could be, then we could, you know, maybe [00:55:00] be in a place of where, uh, this is, uh, this is more achievable.

But, you know, at, you know, I think at the moment still, like a lot of the code that surrounds the models and like all the verification and like things that you want to be a hundred percent deterministic, right? You don't wanna give it up, leave it up to chance. And so every single problem is solved with a combination of like, oh, there's a model making decisions.

There's some deterministic logic that we ba you know, that we basically are, are, are creating as the hard and fast rules. And it's the combination of these two things that solve a particular problem. Um, and yeah, like either, again, as the models get bigger, maybe these problems go away and then like either the code is generated or you rely less and less on the code at all, but we're not quite there where I think we could do that.

Mehmet: Uh, you know, actually you took the question that, you know, I was [00:56:00] preparing, which is exactly about this. But you know, in a nutshell, if I want to ask you this, uh, art, um, because you mentioned this, so it's not now, it's not the moment yet, but can we expect at some stage an agent, an AI agent that understand the business process or processes, right?

And. Can spot frictions and try to optimize it, and then goes either write a code for that or it might write agents for that. So am I talking too much, uh, you know, science fiction here, or is it, you know, with, with all what's happening with AI now and, you know, we are seeing, uh, a big focus on agents from all the major players open ai, entropy, Google.

So they're now, they left, not they left, I mean, but they, the [00:57:00] generative ai, it reached a kind of a po a tip, I would call it tipping point. Maybe still there is something to, to do more about it. So now we see them like working on videos. They're now trying to enhance, you know 

Art: Yeah, yeah. Generating images.

Mehmet: Yeah. Yeah. But, but, but now also on the side, we see all of them. They're trying to go and focus on the agents also as well. So where do you think they want to reach with the agents? Especially, you know, uh, everyone talks about, uh, super, uh, super intelligence, right? And to me, super intelligence is a machine that can, will be monitoring all the time, right?

What's happening around it. And then it goes and try to solve this. Now I'm not sure if we're gonna do this autonomously or we're gonna still give that. Yeah. Now, now your take as someone who, who, who's been, you know, in tech for a long time, especially with what you're doing at Augment now. So what, what you can tell us, like what are your expectations?

Of course. Like no one can predict future. I know [00:58:00] this safe harbor, but you, your, your point of view, let's call it this way. 

Art: Yeah. I mean, it's, it's a hard, it's a, it's a hard, um, hard question, I think. Um, you know, I think right now, you know, I think chat, uh, chat GPT had released like an agent, kind of like framework.

Uh, it's a visual visual builder. Um, I think, again, I think stuff like that works for simple use cases, but then once you have like a very complex workflow with lots of branching, lots of conditions, lots of asynchronous behavior, um, it becomes real so hard to manage that it might as well be code. And so the way that I, the way that I look at it is that will have to,

we'll have to get to a point where the, like [00:59:00] in order to, you know, to like avoid the visual builders and go directly to code, the models themselves have to be better and better. And I do have confidence that they will improve considerably. Right to where the amount of errors and the amount of like, you know, the, the, the sort of like the right understanding of intent is there and the amount of errors that they make kind of dissipates.

And then what ends up happening is that, like the code that's generated is a fairly accurate representation of what is being asked for. I think my concern is that not that the models are gonna be really good at code gen, which I, which I think they will be. And we've seen that track. I mean, it's just been kind of, you know, going up to the right.

I think the hard part in, in all of this is, uh, writing the prompt. 

Mehmet: And, 

Art: and this is, I see as kind of like a, the singular sort of like [01:00:00] big, you know, challenge. Like even if the models are really, really good with the output. Ultimately the input and like how you write it and what you put in it and, and is, is in, is is very closely tied to the output.

And I think it's like it starts to kind of show the fallibility of being human, where we're like, sometimes our unclear, our logic ourselves in our head is fuzzy. And so whenever we write it out, it's also fuzzy. And so then the model has to guess and make assumptions that, you know, based on that. Right.

And so this is kind of like, what I've seen in the industry so far is everyone's got four deployed engineers, four deployed, and four deployed engineers are the ones that are writing majority of the prompts, uh, four production workflows. Mm-hmm. And I think we have to think about like, why is that, right?

Like why, why, why are we still sort of relying on, you know, I would say semi-technical, [01:01:00] technical folks to write the prompts. And I think a lot of it is because. This, you know, engineers have this ability to think about like a bunch of corner cases and like rigorously test and like, that's part of the build process, right?

Um, and so I think the, some innovation will need to happen, uh, on the prompt development itself, in order for us to be able to kind of achieve what you're describing, which is like, okay, like I, you know, it's kind of, I don't know about superhuman agile, put that aside, but, um, but to achieve this sort of like next step in, okay, like not only can I generate really good code, but I'm able to really tease out, out of a person what needs to happen.

Mehmet: Mm-hmm. 

Art: And that's really the hard part because we don't, as humans, we don't think about all the corner cases, all the situations and all, like all the, the whole sort of like decision tree. It's hard, it's hard to [01:02:00] think about that. And it's hard to write it. And so to me, that's kind of the next, like kind of the next thing that, that really has to happen, which is like, you know, typically if you're doing clot code, it seldom asks you questions, you tell it, and then it just kind of goes, does things.

And then you look at it and you're like, oh, that's, oh, okay. Yeah, I can see how you interpret it that way. That's actually not what I wanted. And like, and there's all this back and forth, 

Mehmet: and 

Art: the reason for that is because of what I provided, right? And so like, once that evolves and gets better, I think everything becomes, you know, like I, I, I think the whole kind of like, you know, um, uh, we can then, I would say not just focus on having four deployed engineers and then like the whole thing just, just there's, there's the next sort of like aha moment, right?

Um, and, uh, I, I think as far as like a GI and. You know, kind of like there's a, there's a big focus on like scientific [01:03:00] discovery and, and, and, you know, that's, that's kind of like, um, kind of a, another sort of like thread here. It's, it's hard because it's, it's definitely not my area of expertise. So I have a, an opinion just like the dozen others.

Um, but I, you know, to me the, maybe the existing technology serves as a foundation of what ultimately becomes a GI, but it doesn't, given the amount of kind of like effort and expertise, at least from a business process, kind of like automation perspective that you need to have. I, I think that, you know, we're obviously not there and, and I think that like, it may come, um, but I think we're gonna need to have some like step function, sort of like technological sort of like, um, innovation for that to happen.

Mehmet: Absolutely. You know, like, uh. I think we still have some time, you know? Of course you have time. Yeah. Yeah. The [01:04:00] only thing is, you know. Maybe we, we are seeing fast developments faster than we used to see before. So maybe this is creating the excitement. Another, another thing is, which I'm telling a lot of people, and it's normal, right?

So, so yeah, yeah, yeah. We're living in a moment where, you know, we have a lot of these people who likes to exaggerate things and start Yeah. Like this is the last thing. I agree. Yeah. Yeah. So, and it's not the fault of the people who are behind these technology. There are a lot the, it's not the fault of, of these people also because I understand, and you know, a lot of, of, uh, people that are telling me this, uh, expression now we are living in, in attention economy, so, so everyone trying to get attention to themselves so I can understand like some, yeah, we need to teach people and do all this, but yeah.

Nevertheless, uh, on the prompt thing, you. I'm not [01:05:00] a prompt engineer. I'm okay. I have the knowledge of how these LLMs works, but I'm not the expert. Of course, I, I left, I left being, uh, you know, having the engineer hat quite some time ago, but I did a, an experiment where actually I vibe coded a tool that tries to enhance the prompt.

I'm not sure if it's accurate or not. Yeah. But I can tell you, you know, the output from a prompt that I write in simple English. Versus when I put my idea to this vibe coded tool and the way it, it gives me the prompt. Yeah. I get better results I can see because of course it tries there to put, you know, like, act as you are this, uh, this is the context.

Avoid this, do this, do that. So of course I get like, better result, but still, by the way, still I need to go and back and forward, like even after I enhance the [01:06:00] prompt. There's a lot of work to be done. But nevertheless, it's, it's a good point that you brought, uh, on art as we are coming almost, uh, to the end.

Final thing that I always ask, uh, my guests, where to get in touch, learn more about, augment and maybe, uh, you know, know more about what you're doing. 

Art: Absolutely. So, um, so my email is, and hopefully I don't get any spam 

Mehmet: you both have to do that 

Art: cannot be easier. It's, uh, art at, uh, go augment.com. Uh, go augment.com is also our website so you can check out more about us.

Um, and, uh, not a lot of folks with my last name, so easy to find me on LinkedIn as well. So that's another good way to get in touch. So either one of those would be great. 

Mehmet: Cool. So, uh, I usually don't put the email, but guys, you will catch it. I don't put the 

Art: Yeah, but 

Mehmet: I gonna, I gonna, I gonna put your link in profile.

Okay. That's good. 

Art: [01:07:00] Yeah. Maybe we'll scratch the email. Yeah. 

Mehmet: In, in the, in the show notes and of course the website, goman.com and um, yeah, like. I really enjoyed the discussion. Time passed very quick. Uh, I know like we can do more and more of these in the future, hopefully if you get the time. Um, because it's, it's like the topic of, you know, uh, especially the AI and the future of AI and where we are heading is, is yeah, we can have like, uh, panels and, uh, that's right.

Yeah. Yeah. It's a lot to say. Yeah. Yeah. Yeah. A lot of things. Well, and you know, like as I was telling you at the beginning, I, I'm curious all the time, especially to understand this from technologist perspective, right? So I can go and ask someone who say I'm expert, but it's not like the guy who does the thing.

They're actually using the tools every day and Yeah, like 

Art: having to. Go through a pain. Yeah, yeah, yeah. Exactly. Exactly. 

Mehmet: So, so no one like would be better than someone [01:08:00] like you are to, to share this then. Thank you for sharing and thank you for sharing your story as well. You know, uh, moving across, you know, different companies, getting acquired twice and merging engineering teams.

So I think this is also was one of the, uh, important topics we discussed today. So thank you very much for this and this is for my audience. This is how I end my episodes usually if you just discovered us by luck. Thank you for passing by. I hope you enjoyed it. If you did, so give me a favor. You know, I know like it's something very cliche, but subscribe and share it with your friends and colleagues.

We're trying to do an impact. We're trying to reach as much people as possible, and if you are one of the people who keeps coming, they're loyal to the podcast. Thank you very much for your support and help, and thank you for, you know, getting the podcast this year. All the time in the top 200 charts in one of the countries around the world like you, every, every day I wake up and I find the podcast is trending in one of the countries.

So thank you for, for your support wherever you are, and [01:09:00] thank you for supporting on the book launch last month. So again, uh, you, you know, this couldn't happen in a successful way without your support. And if you are interested in the book, you can find it on Amazon. It's called From Nowhere to Next. Uh, if you type my name, you'll find it.

So as I say, always stay tuned for a new episode very soon. Thank you. Bye-bye.