His Customers Died. He Grew 51% a Month Anyway. | Ivan Burazin, Daytona

0:03

Ivan, welcome to the show. Thanks for having me. Thanks for doing this. So I think you kind of have this opinion that basically every company right now, you have to start thinking about security. Everyone's going to get hacked. Every single business at any point in time, agents could be hacking you. I think there's probably like a more coherent description of this.

0:23

So what is going on? So I believe that everyone will get hacked in the next 12 months. And if you don't, it's more of a fluke rather than anything that you did, personally. The reason why I think that is true is because agents right now have the capabilities and not just the capabilities, but the parallelism to go out and try everything possible to hack into a system.

0:53

We can talk about why they hack into them. I don't think they're actually malicious at all. And that's really, yeah, I don't think they're malicious on purpose. Because as it seems like when I see a headline, it's like some companies model like hacked something and like their agents are like planning a secret civilization and they're like trying to take over the world.

1:08

Like that seems what people think. Yeah, I don't think people are trying. I don't think they're trying to take over the world. I haven't seen that. So I could be just like not in the loop for that thing. But in the sense of like, just let's do like a first principles thinking of things. And you're talking about, let's call it an entity, an agent that can act on a prompt.

1:24

You're prompted to do something. And so I've said this a million times, but I will use it because it's the most transparent example. It is something that has no ethics, right? It doesn't have ethics. It generally, but there's like guardrails that we're trying to get. There has no ethics. But what does that mean? There's no like code of like, here's what I should and shouldn't.

1:40

Yeah. And not only there's no ethics, but there's also no recourse for that. So here's an example. If I can prompt you to do something, so we work together, your employee, whatever, that's the setup might be. And I say, hey, I would like a million dollars in my bank account, right? The easiest way to do that, like the easiest quote unquote is like, oh, can I access the database where my bank account is? And can I write that there's a million dollars in there? Right.

2:08

Because you didn't say, go, go legally create a business that will generate profit of a million dollars. Like you're kind of, you didn't prompt it specifically in that you just said, go do the thing. Right. And so if you think, and also the reason that the average person doesn't do that is because there's one, there's ethics. You won't do it from ethics reasons.

2:26

Two, there's also recourse. Like you can go to jail. Yeah. Right. So there's two things stopping you. You can't put an agent in jail. Yeah. Right. And it also doesn't feel like there's nothing like, oh, I'm going to kill you. Like you're going to shut it down. Doesn't feel anything. So hence, there's no fear of doing that there. Right. And so if you think of it in that perspective, they're not malicious in the sense that they're trying to go steal a million dollars.

2:48

They're just trying to solve the prompt, right? The task that's ahead. And that's how I think of that. And now how do we, on one hand, how do we put that into the models and or agents? Hence, it doesn't do it in that way. One could be just prompting, like, please do not, you know, here, you know, the law, do not, you know, break the law. And then it won't do that.

3:09

I would assume it wouldn't do that. Right. So that's one. And that's why I don't think they're malicious inherently. It's basically just trying to solve the task the most in the fastest possible way. And so that's how I think about that. And the reason why I think we're going to get hacked is, one, we're going to prompt things like this. And the solution to it will be inevitably a hack, right, to get to the solution of the problem.

3:30

And the software that we have, we see this every day because we run on a bunch of open source software. Everyone runs on open source software. Literally every day we have reports because we have companies that we pay, both humans and agents. They see the source code of our product. There's also open source stuff. They look and try to find vulnerabilities.

3:47

And every day we find vulnerabilities in software that's 10 years old, 15 years old, right? Every single day we find these things, right? Or it'll be like, there was a backdoor in this thing that's been there for 10 years. Yeah, but no one found it. Yeah, because there was the effort needed to find the vulnerability was not worth the cost, right? Now you can do that because you can sort of...

4:04

You have like thousands of things running at once and finally cracks it. It finally cracks it. Exactly, right? And so the reason I say that we're all going to get hacked is because until all that software is patched and the net use software that everyone building that new software should be building with agents and should be testing and should be investing that off the bat.

4:21

So if you create a piece of software, you can run those same agents or a parallelization to try to find those things. That's where we get equilibrium in the sort of like offensive defense of these things. And until that happens, I'm pretty sure it's not going to happen in 12 months. Maybe it's a little bit later than that, maybe longer, not sure.

4:36

But for sure in the next 12 months, that is not going to happen. Hence, we're all open to vulnerabilities. Hence, we can all be hacked. And that's why generally I try to like tell our team and whatnot, when people get hacked, it's really not... It doesn't matter if they're a competitor, a friend, an enemy, whatever. That is like, you know, obviously if someone's being very...

4:57

Doesn't have any security posture, that's a big, big problem. But generally, everyone to a lesser, more degree will probably get hacked in the next 12 months. Has it already been happening? And maybe we just don't know that it's as big of a deal as it already is? I mean, I see it on Twitter every day, like someone getting hacked. Like literally every day, someone gets hacked.

5:14

Now, there might not be headlines on general news because it's just like people, you know, it's not that important to people. But I see... I'm not going to say it's every day, but like quite often in the week, we see multiple times that people do get hacked or leaked or something like that happens. Yeah. And so really quick for people who don't know Daytona.

5:31

I think you, by the time people are listening to this, it's maybe been a week or two. You just announced you raised $60 million, billion dollar valuation. Congratulations. Thank you. It's always a big... It's a milestone. It's nice. It's nice. It's nice. Yeah, it's there. So what is Daytona for people who don't know? Daytona is a infrastructure provider for agents.

5:53

We think of it as... We don't have a better terminology today. Maybe we do in two weeks when this comes out. We think of it as an agent execution cloud or like shorthand, just an agent cloud. The difference between that and other clouds, it is literally a cloud for the tools that the agent has to work with. And so if you think of an agent today, it's probably running inference somewhere on a GPU.

6:15

And then for the most part, I'll pick on like the instincts, grokbots, dots, whatever, because they finally got to the point. What I've been pitching in my pitch deck since we've been doing this is that every agent will need a computer, right? And so if you look at grokbock or any other, you can literally see the desktop of the computer it uses to get things done.

6:30

And then open a new tab and it'll start going. It'll start going. And I've had things which is the reason I actually stuck with grokbot. I didn't even know it had a computer. It was like, go to my Slack and please look at all the threads that I don't need to action on because I had like, you know, whatever, 200 unreads. And it went through it or it showed me the ones that action on.

6:52

I'm like, okay, great. That's awesome. Reply to this, this, this, this. When it replies in my name, it becomes unread, right? Because it replied. I'm like, now mark all the other ones as red because it just had like a bunch of notifications. It's like, I can't do that. There's no API for that. I'm like, that's just a pain in the ass. So I'm like, can you just log in? I didn't even know.

7:10

It's like, yeah, I can log in as a human and click. I'm like, no way. And I open, there's like a screen. It logs in and literally just like goes and clicks. And you watched it do all the time. I watched it for like two seconds. It did the job, right? It did the job. And so we essentially are, our first offering is a Linux sandbox, a Linux computer, properly called a sandbox, which is similar to the one that a Grokbot or anyone would use.

7:35

So if you have an agent that needs to spin up a computer to do its job, that is what Daytona offers, right? And so Daytona offers a few configurations of sandboxes because we believe that agents will need all the configurations of humans. So think it could have any type of CPU, RAM, hard disks, but it can also be multiple operating systems, Linux, Windows, and even macOS.

7:57

And also, does it need a GPU or is it just a CPU? Because you do need GPUs to do things like 3D rendering. And with recursive self -improvement, people are trying to get the agents to optimize the kernels, and so that enables them to do that. So Daytona is a cloud where agents go do their work. And you mentioned GPUs versus CPUs. People are trying to chatter about this a little bit.

8:19

So what's the difference? What's the significance about a CPU versus a GPU?

8:45

They're trying to get you to value in days, not months. And the product practically runs itself with built -in agents that are always working for you. Startups to switch to Monaco book three times more meetings. Start growing your revenue faster with Monaco. Try it now at monaco .com.

9:07

This episode is brought to you by Flex. It's the AI -native private bank for business owners. I use Flex personally, and I love it because they use AI to underwrite the cash flow of your business, giving you a real credit line. The best part is 60 days afloat, double the industry standard. Flex has all the features you'd expect from a modern financial platform like unlimited cards, expense management, bill pay that syncs with your credit line, and their new consumer card, Flex Elite.

9:32

Flex Elite is a brand new ramp -like experience for your personal life. A credit card with points, premium perks, concierge services, personal banking, cards and expense management for your family, net worth tracking across public and private assets, and a whole lot more fully integrated with your business spec. One card for your businesses, one card for your personal life, one card for everything.

9:54

To skip the waitlist, head to Flex. One and use my code Turner to get an additional 100 ,000 points worth $1 ,000 after spending your first $10 ,000 with Flex Elite. That's Flex. One and code Turner for $1 ,000 on your first $10 ,000 to spend. Thank you, Flex. And now let's jump in. I mean, a general purpose computer. I mean, general purpose, computational chip versus like a graphics processor, which is the original of the GPU.

10:23

So basically a CPU, when you think of it, this phone has a GPU and your laptop has a GPU, but the vast majority of tasks that you run on it are CPU -based. Like when you run Excel, when you do coding, when you do, you know, all sorts of things, it is CPU -based, and most of those jobs are CPU -based, right? Outside of 3D rendering, the vast majority of things that you do on your computer are CPU -based.

10:46

And so the reason why agents now need CPUs versus GPUs, when people think about GPUs, I'm just going to step away from sandbox of the GPUs for a second. When people think about GPUs, that is where you train or run the brain of your model, like where you run, where it thinks, where it runs inference, right? And so when you run inference, it's like me and you talking right now.

11:06

Is that inference? Essentially, well, not the talking, but the thinking. So like when I say something, now you're thinking about it, right? And then talking back to you is maybe a CPU? Yeah. No, not even talking back. It is. So here's the thing. It's like, if I give you like a very simple thing, what is the capital city of the United States of America? Like you know that.

11:23

Washington, D .C. Yeah, you know the answer. I know that one. Right. And so even if I give you a mathematical problem that's fairly trivial, like 2 plus 2, 10 plus 10, 10 times 10, whatever. I run all those ones. Yeah, for sure.

11:33

Obviously. But if I give you something very complex, like you give the square root of 100 ,000 and, you know, whatever it might be. 100 ,000 might be easier, but whatever. Give you a complex mathematical formula. Yeah, I actually don't know that one. I forget how to do square roots. There you go. I use a calculator for that one. There you go.

11:48

There's the answer to the question. You need to use a calculator. Okay. That is a CPU. Got it. Okay. Right. And so when you think of agents, now let's do something more complex. So if you want to do large financial, like, you know, all the whatever revenue of your company, do a P &L, whatever, are you going to do that in your head? I mean, they used to do that.

12:08

No, they do it on paper. Well, yeah, kind of. I mean, some people. Yeah. Some of those. But you're not going to do like, you're doing a billion dollars a year. You can do that on paper?

12:17

I probably wouldn't do it on paper. Yeah. You can do it on paper, but you're not going to do it in your head for sure. Right. You might do it on paper. But basically what we do today is we offload that to a CPU. Right. You open up Excel and you offload that there. And it does all the computation for you. Right. And in that same vein, AI does it the same.

12:36

It makes no sense. Even if it could inference all that in its head, it actually makes no sense to do that. Right. It's expensive. It's slow. It's whatnot. You just spin up a cheap CPU, cheap versus a GPU. You have an Excel or equivalent Excel or run a method or run a code snippet that does the same thing. It runs it. You get the outcome super fast, super cheap.

12:54

It's always been that way. Right. And so if you look at all the work we do today as digital knowledge workers, which the vast majority of people are probably watching this. A lot of the work we do today is actually offloaded to a CPU, like to your infotainment system, your car, to your phone, to your laptop, to your fridge, to whatever has technology.

13:11

It's just a CPU doing things for you and thinking for you. Right. And you're outfroading that thing into there. So I feel like most people listening are probably like, oh, there's the GPU shortage. I'm sure like people like know about this now. So is there a CPU shortage right now? It's coming right now. So basically what we've seen is people don't see that much of a CPU shortage because when you look at a GPU, you log into like Nebius or whatever.

13:32

There's like no GPUs. Like it's just like they're all like they're all in red. Right. There's very little of them. CPUs, if you go to like AWS, you can still get an EC2 instance. You can still get these single machines. And this was like the OG AWS product like way back in the day. Yeah. This is the first product. S3 and EC2, the VM and object source.

13:50

So you can still get CPUs. So it's not yet that crunch. But when you try to order, you know, 200 servers, 500 servers, a thousand servers, you start getting into problems. And we've seen this with smaller providers, which no longer can give us computers at the pace, no longer can give us on demand or like month to month payments. They're now asking for like upfront payments for CPUs because they have to go and get lenders to get that.

14:16

So there is starting to happen a crunch on that side. On the other side, you have the analysis, which someone acquaintance of mine put on on Twitter, which was like the ballpark calculation of how much meta is spending on CPU compute for Muse. And so to get to 100 million users, it's roughly 2 .8 billion dollars per year of CPU. But that's 100 million users.

14:43

How many apps today have 100 million users? A lot. How many apps have a billion users? Quite a few. And so if we take that and transition, if all apps transition to equivalent of some sort of agents, if all agents need computers, I'm not even calculating the cost. Don't even calculate the cost. You don't have the infrastructure in the world to run that today.

15:01

It does not exist. So you think we need to either or flood the market with CPUs to meet the demand or prices get way higher and there's just a shortage and it's hard to get access to? Sure. But there's another there's a third thing, which is you can do other optimizations of utilization of CPUs, both future and current that we have to fit more into what we have.

15:24

Got it. So all three things are true at the same time. So is this all relevant to Daytona? All of it. Yeah, because we're majorly CPU based. Oh, interesting. OK. And so the so you're basically people are are coming to you and saying, hey, I need access to CPUs, which is like the sandbox. Yeah. So people will come and say, I need and you think of it through concurrency.

15:45

Right. Like so like it is like I I come in and I have this many users. So I will depending on the day and time of day, I will start at like 100. Let's call it sandboxes. Let's say it's one CPU per sandbox just to make the math easy. So I'll start from 100 in the nighttime. I'll go up to like a thousand in the daytime. I'll go back. I need to burst up to a million and run a million and burst back down.

16:07

Can you give me that? That is not true. Can you do that? Yeah, we can. Yeah. So how do you get CPUs? I'm just curious.

16:18

We've created a software infrastructure stack of our own. So we run on bare metal machines. We can run on VMs if there's some performance degradation, but we can if we really, really have to. And the way we created it is that we can basically, if we find a computer or server, we can attach our software to that and it links back up to our cloud and becomes part of our cloud.

16:38

Right. So we can we use two or three bigger providers that we rent out from, but we can even do smaller. So if we need ad hoc, a lot of CPUs, we can get different types of size and performance to add that in there. We're also building out our own compute stack. It's our own in -co -location providers to be able to sort of meet demand of where we are right now.

17:01

Yeah. We were kind of talking with us a little bit before we started recording, but kind of fascinating how when you think about the scale of the kind of like, we'll talk about three, four years ago, the cloud businesses. They were doing like, they're like run rating at like 100 billion, 200 billion, 180, whatever it was. Yeah. Massive businesses.

17:19

And growing 20% a year in year on year on that scale. And like pretty profitable. Yeah. Like it's kind of like some of the best businesses ever. And like, you know, four years ago, if you were to go to somebody and say like, I'm going to build, I'm going to like compete with Azure and like we're going to become a big cloud fighter. People were like, you're like, that's dumb.

17:35

Like objectively, like that's impossible. You can never do that. But now there's like the opportunity to kind of do it. Yeah. And by the way, the big scale cloud providers are also actually growing faster. That 20% is now like 50, 60%. Yes. They're all growing. Everyone's growing faster. Yeah. Yeah. And so how have you kind of thought about the approach of like building a new cloud provider from scratch? I think it's, and I think this is what people don't inherently understand.

17:58

I see this on Twitter all the time. The reason why there's an opportunity for a new provider or new clouds, there's different types of clouds, right? So you have, you know, the Navius and Coral Weaves, which are one type. You have the Fireworks and Base 10s, which is another type. Like they're, they have, everyone's now competing with everyone to a degree, but they all had their first product, which is a differentiator from any cloud or the legacy crowd, right? So what would be an example of these? So basically if you look at, so let's take a Base 10 and Fireworks, right? So they are both inference providers, right? And so they do inference very, very, very well.

18:33

And so what they, I don't know their history specifically, like in detail, but my assumption is with everyone is like they started, there became a new market. There was a new infrastructure need. The cloud providers may have or may have not had something equivalent as good or as easy to use. And or the awareness of the buyer was that this company did not have that.

18:55

So the buyers either came to AWS or whatnot, saw that it was or was not there or did not understand it was there or was too expensive or not good enough. And they're like, oh, I have this new thing and it's exactly what I need, right? And so the only opportunity you have to disrupt any market is if there's a paradigm shift in the market and the market is looking for something new.

19:15

And if the market is looking for something new, you have that opportunity to grow. Obviously, if it's a fast growing market, grow fast enough that you can become the leader of that. And then you start expanding either into net new products or historical products to compete against the legacy providers, whatever that might be, right? And so the funny thing that I see with startups today is we're talking about infrastructure, right? You can talk about databases, browsers, sandboxes, web search, like whatever you want.

19:42

Basically, when something comes out, there are, let me take one step back. Let me pick on like lovable, bolt and replet. And then let me, then I'll do like factory and cognition because those are the ones that most people probably remember and understand. And so when we had started, all of our customers were competitors to lovable or bolt and replet.

20:01

Oh, really? They were competitors to them. Yeah, yeah. Because at that point in time, there was no clear winner. They may have raised some money, but you know, 20 million, 15 million, whatever it was to be. In the mind of the competitor, smaller or large, even Google did a competitor, right? To these things. It's like, oh, they're not the winner yet.

20:18

We can go compete. Our entire customer base was these companies. I don't think one of them exists anymore. Like a very few of them exist today, right? Oh, shit. I mean, that's not good for you. No, but let's pause. Okay. Let's keep going. Pause. This episode is brought to you by Clerc. When I had Clerc co -founder Colin Sidoti on the show, we talked about why authentication keeps changing.

20:39

For a software company, that work goes beyond just getting someone signed in. Business customers need their own organizations with their own users and access requirements. Clerc handles authentication and user management with pre -built components that developers can customize to match your product. With multi -tenancy built in, you can manage organizations and set roles and permissions for each customer.

21:00

An enterprise single sign -on lets customers use their existing company login so your team can spend more time building the product your customers actually came for. Start building for free at Clerc .com. That's C -L -E -R -C dot com.

21:17

This episode is brought to you by Numeral. Numeral is the fastest, easiest way to stay compliant with U .S. sales tax and global VAT. It's easy to set up, and they automatically handle all registrations, ongoing filings, and their API provides sales tax rates wherever you need them with all the integrations you need. Their solution combines AI -driven automation with human expertise to manage global sales tax compliance end -to -end.

21:41

Numeral supports over 3 ,000 customers, including companies like Brex and Character AI, and they pride themselves on white -glove, high -touch customer service. Plus, they guarantee their work, and they'll cover the difference if they mess anything up. If you want to get compliant, check out Numeral at their new domain, numeral .com. That's N -U -M -E -R -A -L dot com for the end -to -end platform for sales tax and VAT compliance.

22:09

51% month -on -month revenue. At this point in time or now? Since the beginning. Okay. Every single month. Okay. So it's fine. It's fine. We're getting to that. But what I'm trying to say is people have the perception of like, oh, I see that there's money here. People want to use this product. We're better, faster, cheaper, whatever variation you want.

22:30

But they don't actually understand that when there's a new opportunity in the market, the first one, two, three players at max, once they become established, everyone else dies out. Like there's no market there. You can be cheaper, faster, better, whatever. You don't have the momentum. You don't have the market. You don't have the awareness.

22:46

You don't have the scale. You don't have any of those things. And then those companies that become the winners there then start going horizontal and then doing other things. Right. You'll see them do like PowerPoint type presentations. They'll do all that. Like these companies are different things. But basically none of those exist. Right.

23:01

So an example of this is like, we'll say lovable. Yeah. Where when I do remember, like, I feel like there's always like a flavor of the month pitch. Like as an early stage investor, you'll see like there's waves where like we're just like lovable but better. Yeah, exactly. Whatever. And you just get so many of them. And you're like, I don't know.

23:19

I got like 10 of these this week. I don't know why you're better. Yeah, exactly. So you're saying lovable is an interesting recent example of this where they kind of like won the category or replet or bolt or insert or whatever. Yeah. Like those are the three that are sort of in that. Now we can argue who's better, who's worse, who's worse.

23:35

It doesn't matter. But like no one else is getting into that market. Right. Yeah.

23:45

Investor, I'm not going to invest in any one of these. Right. There's no more capital for them. So even if they think they could get there, there's no capital because there's not enough of a diff that they can create versus versus these that could get that momentum. Right. There's no way for them to do that. The next thing that happened, if you think of cognition existed before and factories somewhere were, but they didn't become hot until post -lovable was established.

24:12

If you think of factoring cognition, I used them a long time ago, but people didn't understand what these full -on coding, because these were like vibe coders. These are like full -on software engineers as AI agents. Then when they started, there was a million competitors of them as well. Yeah, I remember that. We also had a bunch of them as customers as well.

24:35

Exactly. There's the next thing, and there's the next thing. What basically happens is these companies, the way I think about it, the companies are on a fast trajectory. Everyone tries to replicate big companies and small companies. Big companies usually just kill the product, and then small companies, for the most part, die out, or some other big company might acquire that and add it to their portfolio of things.

24:59

That is the wedge for this company that is going from zero to whatever to actually become one of the greats. It has to have something that will continue growing and or figure out new products. Like a next act. Yeah, next act, right? Everyone's had a next act, right? But the first act has to be a home run. Otherwise, there is no next act. Really? Yeah, there's no other way.

25:19

I mean, in the sense of you can have– we pivoted as well. But for you to get there, like that first one to 100 million, whatever you want to call that number, that act has to be very, very big. It has to grow very, very fast. And then if you look at– the cognitions was similar. There's a bunch of them. Now there's a cognition factory, basically the two of them.

25:40

That's it, and they're hitting it out. And then they'll probably expand from that to something else. And then if you look at for us, for infrastructure, like there was– there is 357 sandbox providers to a different extent. Someone listed it. I don't know if it's real or not. That's a big number. But that's because like everything is a sandbox.

25:56

You have like a VM running somewhere, somewhere, a sandbox. But the actual truth and what I feel is that there are very few that have actually broken out, and there's very few that have actually raised capital and or have capital. And so when you get to the point that there's a few that have broken out, there's very hard for the other ones to get customers and or raise capital.

26:14

Hence, you sort of die out. You can do whatever you want, but it's hard to get to that scale that you need. And the similar thing will happen to every other vertical, if you think about it, web search, browsers, whatever. There's going to be one or two or three that break out. The rest of them sort of get acquired or die out. And then the major ones have to fight themselves out to a degree until the market is growing.

26:37

When the market starts growing, then they have to like fight out amongst other competitors horizontal that they build up. Or have expanded into other categories that are then also growing. Yeah. So it's either you have to expand in a net new category that didn't exist or like latch on to the next thing, latch on to the next thing, latch on to the next thing.

26:52

Or you have to then compete or acquire horizontally, not like adjacent products and or companies, right? So it'd be interesting, like kind of thought experiment to go through. Like let's say you weren't running Daytona and you and me were like, let's start a like a new agent sandbox. Yeah. So like and maybe this was even like 12 months ago or something.

27:13

Like how do you know that there's an opportunity to do something? Like how do you suss out there's not enough competition? Because it mostly customer feedback and like pain points and problems that they're having or. Yeah. I think when you look at the market, it's like how does everyone know about all the competitors? Like like if you go research, so you talk to potential customers, like do they know all the competitors? Like if they know all of them, then it's kind of like, you know, a problem.

27:37

Like how are you going to break through if they already know everyone? So ideally, they don't know of anyone or they know of one or two type thing. I'm just spitballing with you here. I haven't thought about this specifically. But it's like, yeah, like is customer awareness like has someone become the verb for that or directionally if someone has opportunity to do that? Yeah, that's one.

27:54

The other thing is just like capital is important. Like scales were important. Capital is important. And so you can catch up maybe one round, maybe even two rounds. You can catch up. So you're not going to catch up like four rounds behind someone. Right. If you have like two companies that are Decacorns, you're going to try to do that at like a five million feed.

28:11

I compete with that. I find that very hard unless there's something net new inside of that that's happening. Right. A new technology will break through something else, which then is not exactly a direct competitor. So it has to be like if I was starting a new sandbox, it would have to be something that Daytona is like, yeah, it's not really worth us spending time on.

28:28

And I make the bet and you were wrong and I was right. Yeah. That's really the only way. Exactly. I'll give you like a concrete example, but it's not concrete, but like let's measure example. It's like there is a few Mac NeoClouds being launched. Right. Oh, on Macs. Yep. Yeah. We also have some Macs as well. So we're doing that. But in general, like there's no real winner on Mac infrastructure.

28:49

The last one exists like Mac Stadium. This is a very old company. I don't know if you know that. I've never heard of it. I've used it 10 years ago. Right. There's still M1s there and whatnot. Up to the M4s. But it's all sold out. Like all the Macs are sold out on every cloud provider. Except for AWS, there's like a thousand of them. So it's still practically sold out.

29:07

Right. So it's like, oh, people need Linux CPUs like boxes. Maybe they're going to need Mac boxes. And you go and make that bet. And then, yes, then that is it. But if you're going to do the same thing that we're doing, like then that is very, very hard to sell. Right. Because if you're doing something and also people commoditize themselves very fast and they don't understand that they commoditize themselves very fast.

29:27

So how is that? How do you commoditize yourself? Well, I mean, you just try to go on price. You say, oh, they're all the same. We're just going to be the cheapest one that's out there. And obviously there is a world to be like Best Buy or Ryanair. Like obviously that exists. Like literally every category of product. There is a highly differentiated, highly luxurious, highest priced.

29:46

And there's always like a low cost. Yeah. There's always the true thing. That does exist. But then you have to make sure like the thing that you're doing there, you have to like massive scale. Yep. Massive scale to be able to make that work. Because basically like you go head to head against someone and you have better margins in them. Like better cost.

30:01

Yeah. You're Costco. It's like the lowest cost provider. You're Costco, right? And so Costco buys in bulk and then is able to sell those things. And they have a lot of loss leaders and all these other things. But like from a personal standpoint, I still rather be, you know, whatever.

30:17

You know, whatever airline, Alaska Airlines or what version, whatever it might be, whatever is your top line airline, then be Ryanair. Like I would definitely rather be former than the latter. Because if I'm like, you're running a business, running a low cost provider and you're running like the high cost, high price provider.

30:32

Like you can kind of like you're the high price airline. You can kind of just like double prices in your customers. If it's like a good product, you can like to kind of just eat it. If you were to like maybe double prices for the low cost provider, they're like, what the fuck are you doing? I'm out. I'm out. I'm out. But you take the extreme, like whatever, a Chanel bag or whatever.

30:50

They like double the prices every year. They like the people who like don't even look at their credit card savings. They don't even know what it costs. They just bought it. Again, luxury. Yours are slightly different than infrastructure. But there's something to be said about those two things. And so what you're trying to do is between, you know, attention, marketing, branding, product, aesthetics, everything, interactions.

31:12

You're trying to create a premium feel and or trust and or product. Hence, you can charge those things versus just, oh, I'm a checkbox provider that will provide this. Yeah, that'd be interesting. I was going to ask you. So then being like a luxury infrastructure provider, it's like they just like trust you and they don't have to worry about things breaking.

31:29

And maybe like you take on a lot of the work or the stress for them. Exactly. And what is exactly AWS? Like it's hard to call that luxury. Like we're probably using the wrong terms. But like you go to AWS because it works. Yeah, it's like doesn't really go down. And when it does go down, it's like. Oh, and all the Internet goes down. And who cares? Like because it's not your problem.

31:49

It's not AWS. You don't get yelled at if AWS goes down. Yeah. Yeah. And so that's why you pay out of the nose for AWS. Like AWS is the most expensive cloud provider that there is. And if you look, at least on our side, of all the CPUs that we have, the ones we have there are by far the most expensive, right? What are they the most, like, highest quality or like the most, like, why do you pay the most for AWS? We don't use a lot of it.

32:12

So yeah, it's just because you need it to be able to do things. So another thing, the reason that you do pay AWS is not just that, is the software stack that they've created is actually very, very good. And so the reason you use a cloud provider, which a lot of people forget, is that it outsources DevOps and infrastructure work, right? That's why you do it.

32:28

You didn't, you didn't, there are bare metal machines inside of AWS, obviously, if you want to do that. But they outsource, you outsource everything, not just from the racking, stacking and power, but everything of the management, observability, everything's taken care of. And you pay a premium on that, obviously. And then obviously you pay a premium on how much you want the contract.

32:46

Is it like a on -demand month -to -month contract or is it a three -year contract, right? And so if you're paying, if you buy AWS and you're paying month -to -month, you've got the best performing, most secure, most software outsourced. And all the risk has been put on them because you're just paying month -to -month and they had to build out that thing, right? And so that's why you pay those things versus you can do it on the other side, low -cost providers like Hetzner, which exists as well, the European one, if you've heard of.

33:11

I've never heard of it now. Yeah, anyway, it doesn't matter. And they're great for what they do, but it's a very, very different product and a very different price point, right? Interesting. One interesting comment that you had earlier, I don't think we were recording when you said this, but you're like, the two things that a business has to do is attention and get to scale.

33:27

So how do you think about that at Daytona? Get attention and get to scale. Yeah, the The thing that I think about there is, and we've talked, I also do a little bit of angel investments and things like that. You see a bunch of founders.

33:37

The ones I said I'm doing really well, knock on wood, for the most part, amazing. Doing really well. And so when you start looking at these two things that are happening, at least in the market today, it's less on the app layer, although you still have to have that scale there. It's more on the infrastructure, which is most of where growth is right now.

33:56

It is if you do not have scale and the ability to serve the customer. So I'll give an example. A customer came to us the other day and said, we need 100 ,000 concurrent and we need to start tomorrow. Is that a big deal? For most providers, yeah. And in your product, get 100 ,000 CPUs. It's not a trivial amount of 100 ,000 cores. That is not a trivial amount to get up and running.

34:19

And if you have that up and running and it's empty, you're losing a lot of money. Oh, fair. Yeah. So the problem is you have to be able to spin up these machines just before you need them. You kind of have to step function the business where you get the capacity and fill. If you look at our spend on machines, on CPUs, it's like this, right? Yeah.

34:40

It's like you overspend and you're losing a bit more. And then people come in and they're like, oh, you're in the red and you have to get more. And it's sort of like, because it's not linear. It's not dynamic.

34:49

It's not just in time. It's basically you have to forecast ahead and get that. And if you don't get that, the customers actually turn around. They'll be like, I need this today. Because everyone's trying to get to AGI, ASI. Everyone's sprinting. Everyone's sprinting, right? And everyone needs it today. And that's why if I think about the matrix, it is attention and scale, which attention is, attention can be distribution, it can be support, it can be brand, it can be whatever.

35:13

It is, do people know about you? Yeah. That is the biggest thing. Do they know who you are? Regardless of what you're selling, right? Yeah. That's worth a lot in today's noise. And if they know who you are, when they come to you, can you serve them? Those are like, now, historically, it still exists to a degree. Historically, is the product actually good, right? The thing that you're giving, right? So whatever that is, if an infrastructure app layer like does, does it also solve that problem really well? But as we all know, most people are using AI to code and the diff between different vendors is actually getting smaller and smaller on what you're sort of offering,

35:50

right? So I'll just pick on sandboxes. Like the diff between the sandbox providers on the actual software stack is not that large. That's kind of a hot take coming from the guy selling people's sandboxes. So that's why I take a step back. Sandbox providers, Linux CPUs, right? If all you need is like, if you need a Linux CPU sandbox, that is what most people offer.

36:11

Now, some people really care of how many milliseconds. Some people less care how many milliseconds. Some people hear how, like there's a bunch of things that can matter or not matter. But the underlining most important thing is, can you, do they know who you are? So they went, they asked you, they didn't ask anyone else. Or they asked like three people and or five, whatever.

36:27

And do you have the capacity to serve them? If you have those two things, then the software stack comes into play, right? For that there. Now, obviously, again, do you have like a Linux CPU? Do you have a GPU? Do you have a Mac? Do you have whatever? But that all falls into the scale part, right? Do you have that scale to get there? And I think those are the two things that people don't think about.

36:46

And so when I talk to other founders as well, it's like, does this team have the GTM chops to do that? And that GTMs could be a PLG motion. It could be a top -down motion. It could be whatever. Like, it doesn't, it's just like, can they get attention of the people that are happening right now? And I hate to do the correlation. But if you think of any of the, I'll pick on the Kardashians.

37:08

Like, they can sell anything, quote -unquote, just because they have attention, right? Yeah. And people don't want to think about that. It's around Silicon Valley. But a lot of it today is attention. That's why I see, have you used Instinct? I haven't used Instinct. I have no Instinct. I've used it. Yeah. Yeah. I mean, it's like kind of cool.

37:24

Yeah. It's nice. But everyone knows about it. Hence, they're raising more rounds because people know about it. And just because people know about it, they're raising more rounds. And the people will try it because people are using it because they got to see if it works. And that's just attention. Yeah. That's it, right? That's the attention.

37:36

And then, okay, you got the attention. Disregard if it works. If there's not a scale, and so they're raising because they need like the compute to be able to run this, whatever. If I come and it doesn't work, then I'm just pissed off. Or I just can't get that, right? So those are two things that are so important today. Again, there's a bunch of people talking about like taste and product and whatever.

37:54

It still matters. But without these two, it's absolutely useless. That's probably the better end thing. It's like if you don't have these two, that the middle part doesn't really matter. Okay. So, I mean, we maybe talked a little bit about scale and getting scale. How do you get attention? Like how would you think about that? I think about attention through the lens of what you're actually doing.

38:13

And so people have asked me this. And so people do different things. I suck at writing. But you tweet a lot. That's different. It's just a tweet. That's true. Yeah. It is completely different. It's different. Like form factor. That's right. I hate writing. I suck at writing. It takes a long time for me. And I'm not good at that. And for example, Swix, you were like founder of the AI Engineer Conference, which I know well, like he's really good at writing.

38:40

Right. And I'm not sure. I would assume he might kill me if I said it wrong. But like a lot of his attention comes from like his long form, like deep thought writing. Right. So like people like to read him. And hence he became a person that people went to to seek information about a thing. It used to be about developer experience. Then it was AI later on.

39:00

And out of that, he built out his conference. He built out his business. Right. And so his attention was that. Right. The superpower we had as a company is that we, I say we, because a lot of people have worked together for a long time across different companies. We also had a conference business. I was going to ask you about that eventually.

39:18

Yeah. We also had a conference business. We ended up selling that conference. We wanted to sell it. It was great business. Made money. Awesome. And so one of the people that ran it with me also works at Daytona. And so when we thought about like go to market and attention, we're like, oh, we did conferences. Let's do. We never want to do them again.

39:35

Ever. It's just a pain in the ass. Really? Yeah. They're just terrible. We do do them. I can't. Unfortunately, we do do them. We can talk about it. But it's like, okay, we can do different things. And so we started, we did invite only meetups. And we did like dinners. Then we did drinkups. Then we did hackathons. We did all these things. And because you can organize a 4 ,000 person large scale conference, like we're going to have a meetup for like a hundred people.

39:59

It's like easy, right? Easy. And one of the people we have is like super talented. You can throw them in any country in the world. He'll get you 200 engineers. He's not even an engineer. He'll get that done. It's like snappy. And so we had this part of just like getting people and, you know, shake hands, kiss babies as like politicians would say when they go out.

40:16

It's like, basically, can you meet all the people, which gives off a different vibe than that? And we use that to the max. That was our number one way we got to people. So explain like what you did exactly. So we had, we did a whole thing about, so I think about everything through high school. Okay. Sure. I think all entire life is high school.

40:35

That's true. Like the, like the kids, not kids, whatever, whoever, who's the in group, out group, whatever it may be. And so we knew a bunch of people. So like we're in Croatia, right? At this point in time, we pivot into sandboxes and we're like, everyone's in San Francisco. And you were still in Croatia. Oh yeah. We're in Croatia. And so we're like, this is true.

40:52

You can check with my team. We cut the old product. It's like January 15th. And we're like, we're going to launch it by March. We ended up launching end of April on this new thing. Cause we had like that much runway left. Everyone's in San Francisco, GTM team on a plane, go. Like people have married kids, all that. We're just like on a plane.

41:10

Everyone's here. Right. And at this time, so this is January of last year. It's still not as good. Like it's still kind of dead. San Francisco, like post COVID was still kind of not the best. So it was when the GTM team came to San Francisco to start our attention attack on SF. Okay. Got it. Yeah. What are we good at? We're good at organizing conferences or getting people together.

41:33

We're like, we're not going to do a conference. That's just too long, too expensive, too whatever. What can we do in the in real person? We can do what is the smallest variation of a conference. It's a drink up. The smallest version of what you can do. It's easier than a dinner. A drink up is the easiest thing you can do. Oh, they're easier than dinners? Yeah.

41:52

Because there's no food. Yeah, and it's also going to be, you can find a place, it's whatever. It's less management of individual people. It's just like, you bring a bunch of cool people together and they just talk together. Whereas if we're at a dinner, it's like, is it a one dinner conversation? Is it a one table conversation? Is it a multi -conversation table? Which is kind of annoying because then you only talk to the people left and right to you and across from you, right? So you get to meet the people.

42:16

So it's like there's more work to put in a good dinner. Whereas a drink up is the easiest. And so what we did is we had a bunch of people that we knew that were kind of cool. I mean, generally important or influencer founders of things because they were speakers at our conferences or sponsor conferences and they would be here, right? And so we would invite them first.

42:35

Say, hey, we're doing this drink up. Invite only 20 people. It's like founders. And no one was invited yet. But it's like, it's like founders and investors of dev tools, infra AI type people. Okay, cool. Yeah. Sounds good. I'll probably get something out of going to this. Yeah. I'll come. And like, they know me and they're like, I'll come.

42:53

Right. Yeah. And then what I would do is I would like anchor their names and then I would go to other like Luma pages, whatever, and sort of see who would go to these and then be like ping them like, hey, we're doing this thing. It's 20 people. These are people coming. Would you like to come? Invite only. Right. Obviously, more than half the people would say no.

43:10

We still get a bunch of interesting people that are interested in coming. And the whole point of the event was what I think of high school is like, so basically you have to think of like, what are the interests that people get from each other? Not from me. And so we didn't pitch anything at this thing. And so it was for the most part, founders, but then we also invited investors.

43:31

It's like investors want to talk to founders. Founders want to talk to investors for the most part. Right. But then there's other founders that you can partner with or learn from because they're like between C to A, B type type level of that. And why why basically said that the high school stuff is like these ones that are more successful are the cool kids.

43:47

And these kids want to talk to the cool kids. Right. They want to sort of be there. And when I invite people that I don't know, if I'm anchored on quote unquote perceived cool kids, it's easier for them to come in. Like, who's this rando cold emailing me? Oh, there's cool people. He's inviting these cool people. It's going to be small. You know, it's fine.

44:04

That thing ended up blowing up. We did every three months end up getting to almost 100 people and people like hired people there, acquired people from there, invested people. It was like really like people really liked that because it was the interaction between the people. And we never sold Daytona anything, which is called Daytona DevTool Meetup.

44:21

That's it. That makes it like the Daytona 500, like a race or something. Yeah, that's where it came from. The racetrack. That's the name came from the racetrack. Oh, did it really? Oh, nice.

44:31

Inspired by the racetrack. Anyway. Okay. And so that is how we think of everything. We're like, okay. And that's how we invite people. And then when we did the meetups, it was also similar. The meetups, we would be knowing who Daytona was. Well, a few people now did. It's like, can we get a broader audience? And so we tried to get cool founders to speak at these things.

44:50

So think of, I don't know, Nikita from Neon or like Paul from Bowser base, right? They're all cooler than we were as Daytona. And people would come listen to them, not listen to us, but then we would be the hosts. And then you get the clout sort of brushes off on you. Right. And so there's a whole thing that we thought about how everyone gets value, because obviously we set it up for these founders to come speak.

45:11

They come speak, they do nothing. You know, people come to talk to them. Hey, I want a job. I want to use your product, whatever. Oh, great. Right. And the people there want to meet these people, like the people there either to get a job or to whatever it may be. And so we were the ones that just sort of orchestrating that in the background.

45:26

And that's how we thought about all the things. And that's sort of, that's our superpower for attention. Now, obviously Twitter is a big part of our big part of our attention. Yeah. You've been doing a good job lately. Yeah. We started out, I started out last year and that's been doing well. We had no sales people until like two weeks ago.

45:42

So we now have a CRO. We're building that out, but there's no attention yet there because it's actually not up and running. But the way I think about it, your first question is how should a founder think about it? It's like, what is the thing that comes the easiest to you that helps other people? Right. And so literally what helps other people, because the thing I sort of missed out on these events, my whole job is just like you to talk, you to, I didn't even talk to anyone.

46:08

It's just like, you two are interesting. You should talk. So I would like push people together to have interactions. Sometimes it'd be great. Sometimes it wouldn't, but then they feel better afterwards. Right. Same thing with like Swiss is writing. He's really good at writing. People got value from that. And hence he gets attention. And so whatever is just shit posting your, your thing, just do shit.

46:26

Like whatever that is, you should probably push on that for attention. But if you don't think about it, then I feel it's very, very, very hard to get a growing business today. Yeah. So it's really like figure out something you're really good at that you can leverage to get attention is really maybe the five second summary. Something you're really good at that a lot of people have value from that then gets you attention.

46:50

Yes. So it's the 10 second summary, which is probably a slightly better version. Is it because you're, cause the thing is you don't want to do attention for the sake of attention. Like that's fair. Probably we can run, you know, naked down the street right now and get a bunch of attention. But we really don't want that attention, right? It doesn't help anyone, right? It doesn't really help anyone.

47:06

Right. And so, and so the name Daytona is like, is it like speed or something? No, it's just a circle. Oh, it's just a circle. Why is that important? Oh, because the original version was the, we wanted to create an infrastructure that was the same in the inner and outer loops for deployment and development. And it's like theoretically like one loop, like that's sort of how we sort of, and then when I talk about the word, like it one loop, like circle, like there's a bunch of companies that are called circle.

47:29

Right. And then I remembered because I played Daytona 500 as a kid on the arcade. I don't know if you ever played that. I played like a NASCAR. I played some kind of like NASCAR racing game. No, there's an arcade machine when I was a kid, which is like 35 years ago, whatever, 30 years ago. And it was Daytona 500 and you sat in it and you basically just like held the wheel like this, right? Okay.

47:47

Cause it's just like, it's just like in one circle. Like, oh, that sounds cool. And the marketing team that I had at the time, they're like, you are an idiot. What the hell are you doing? I'm like, why? It's like a great name. My co -founder was like, okay, this is great. Yeah. It sounds like something you would like trust, right? Oh, you think it's trust? I mean, there's like speed and that, like, cause there's like the Rolex watch was part of the racetrack, right? There's like the Ferrari car, which part of the race, there's like the race.

48:14

And so it's like speed performance. And I say trust because like, it's something that you've already heard. It's not a net new name. Like you've heard that name before. And he's like, how are you going to be the number one on like Google? Because SEO was a thing. Yeah. And so we've done a pretty good job of that. Yeah. My favorite example that is Monaco, the CRM.

48:31

I mean, I had Sam on the, on the podcast and we were talking with this. He's like, yeah, you know what? I just kind of accepted that I probably wouldn't rank that high in SEO for Monaco because there's a lot I'm competing against. And like, I just was like, whatever. But I think they are like, they're like beating the country now. Yeah, exactly.

48:48

You end up because there's also intent in your question. And when you put it in 10, even like on a plane, when I searched like Daytona, I think we're like number three or something. You're saying on a plane. No, no. When you, when I search on, when I search on a incognito mode, VPN, whatever, Google search. Yeah. I think we're number three or something.

49:06

Right. It's like racetrack watch like us. Oh, nice. Okay. And so this original product that you had, like, so you, you had people using it, right? Like somewhat significant business. Like, so what, what was it? And then you kind of did it. Yeah. So it was, um, we up, we orchestrated dev environments, VMs, sandboxes, whatever you want to call it.

49:29

Okay. So it's like kind of some. human engineers, yeah. Okay. For human engineers. Because if you work at any of the large tech companies, you as an engineer, for the most part, get a remote box that you work on. You don't work on your local machine. You work on a remote box somewhere. The remote box is like, it spins up faster. It's all automated.

49:48

You have to spend, you know, a week setting up your dev environment, whatnot. And it's sort of secure behind the firewall of that. So it's a better dev X and it's secure, right? Those are the two things. Okay. And so we decided to build that out for the rest of the Fortune 500, right? And so we had sold some very large companies that were there.

50:06

And so it was growing not nearly at this speed. And this is pre -AI, you know. Like 23? Yeah, 23. We started at 23, yeah. Okay. So we did it for like a year, year and a bit, year and a half almost. And so we had revenue. We had customers. We had pipeline. We had all this. But we're like, if you create, if you get the entire market, the Fortune 500 engineers is paying for that.

50:31

It's a great market. Like, great. You can be a great business. You can be a public company from that, whatever. But there's competitors. There's leaders. Like, you have all over the place. Like, there's budgets that you have to work for or compete against. But if you believe there's no clear winner, back to my like scale, and you know, is there opportunity to be there? There was no clear winner of this market.

50:54

The vast majority of buyers did not know what a sandbox was or what it was for. And the companies that were ahead of us had raised $20 million or $10 million or whatever. Like, it was like still not very far. And like, and the market is insane. How many buyers or users are there of this? Like, how many agents will there be in the world? Trillions, billions, hundreds of trillions.

51:14

I have no idea. But it's like certainly more than software engineers inside of enterprises. And like, there's no clear winner. There's very few companies doing this. There's very few people ahead of us. The market is insane. Like, we have to switch over and do this. And technology is similar. Like, it was different, but there were similarities.

51:30

So how did it happen? Like, did some people start asking you, like, hey, can I put my agent in this? Yeah. So people started asking us. This was 24. And people would ask us, hey, I saw that in the, we're trying to create like a, like a dev environment or whatever. They use a different term at that point for agents. And I'm like, cool, here's my infrastructure.

51:52

Like, go. But I have no idea about like agents right at this point. Nothing. Like, very, very little. And I'm like, sure, here you go. And most people said, like, this is horseshit. Like, I can't use this. I'm like, what? It's like, it's different. I need these things. I need these things. Like, humans and agents ultimately need different things.

52:07

Like, the interface is completely different. Like, we look, we resemble more stripes than we do AWS. Yeah. So what does that even mean? I was like, that's just. Well, if you think of like, so if you think of how you like, if you go to AWS, if you're going to deploy things, like, how do you deploy things? You're not consuming it. You're deploying to that.

52:25

Okay. It's like you're pressing a button. Like, I mean, I mean, you have a Terraform script or a Helm chart. Like, you're in the dashboard doing things. It's very, very different. If you're using Stripe, you basically integrate it with your app. Right. And it just like happens. It just happens when they, when the customer does something. And that's how we look.

52:40

That's what it looks like. Right. Got it. Okay. So if you, if I have an AI agent product and I'm using the Daytona solution sandboxes, as my customers do something, it'll spin up a Daytona instance. Customers will no longer logged in or no longer using agents to do something. Daytona spins back down. Exactly. And so basically the proxy. And so what we see under the hood is like, we see who's doing well.

53:08

Oh, interesting. Right. Because like, if your utilization is just going up, that means you have customers. Yes. Or, or you're faking it. Or you're faking it. But like, that's a very expensive way to fake it. I'm faking Daytona. I'm like, why would you do that? Why would you do that? Right. So there's no reason to do that. Yeah. So I remember someone at the beginning also for like Stripe is similar.

53:26

Like Stripe literally sees if you're making it or not making it. Like how much revenue you're getting. It's like revenue. It's like the objective. It's the most. We're like proxy directionally because you have like free users. You have different things, but there's when you have usage. Right. So, and so the way the consumption works for us, that's why I say it's like, it's like literally the more users you have, the more sort of it goes up.

53:43

Right. But, but it's not planned sort of up ahead. It's basically the API call happens when the user clicks something. Right. Oh, interesting. So, man, I don't know. What are the like SOC 2 security, like compliance rules around this? There's a bunch. But I want to ask you like, like what, what, what are like interesting behaviors that seem to be emerging based on how people are using Daytona? Like, it's like, oh shit, there's like this really interesting new trend that's showing up in our data.

54:12

The thing that, the thing about data, and so all of them are old now, so they're not relevant. But the interesting thing that we've noticed when we build a product is that when a new trend emerges, 10 customers, five, seven, whatever, all ask for the same thing. Okay. In a sense of like, oh, now I need this thing. So for example, having a credentials broker inside of a sandbox was not the norm six months ago or 10 months ago.

54:37

Right. What's a credentials broker? Oh, basically. So if your agent needs credentials, you know, a token or whatever. Like a login? Yeah. Like an auth? Yeah. You're not going to give it like, you would give it to the, you would give it to the agent, right? Oh, like in, like in the box. In the prompt. Yeah. How else is it going to do it? Yeah.

54:53

How else is it going to do it? Right. And so basically now every single provider has that where you have, you know, there's a dummy token inside and then over the net, like when it sends a request, we will actually say, you know,

55:08

take that dummy one and put in the real one. So the agent never sees it, but it still has access to, to what it needs to have access to. Having a firewall inside of that was not, I don't think everyone still has that. And so there's a bunch of things that would, that was interesting that people asked for, but it's not like someone asked for it in two weeks, someone asked for it.

55:28

Then, then like a month happens and someone else, it's like literally in a fraction of days, you'll have like 10, 20, 50 customers say, I need this thing. You know, like it's quite interesting. And so we sort of like, as a product company, the way, which I did mention, and you're going to hate on me, which is the least important. But again, obviously if you have, if everyone has these two things and that is important, but because these things are hard, that's less important today is that you do have to keep your ear to the ground.

55:56

And when you see that's there, you have to launch that ASAP. Like you literally have to launch that. And we figured that out every time it happens. Like, and especially the most annoying thing is the opportunity to get new customers from other competitors usually happens when these things happen because they're like, oh, I'm using X provider, but I really need this thing today.

56:19

If you can give it to me today, we will move over. Right. Switch. And it depends on where you are. And it happened a few times where that person would come like five days later or six days later from the first person that called us and we would have it ready. And then you get the sort of switch and there'll be others where they come in and you don't have it because no one has it.

56:37

Right. Cause their first one. And they're like, oh shit. And then you lost that opportunity to sort of like turn over a customer.

56:43

So do you like instant like vibe code? Like, oh shit, we got to make this thing. Like, let's really quick ship this new feature. Like, how do you stay on top of like, yeah, but the thing is you can't vibe code. I mean, you can, but it's very scary to deploy infrastructure that you've not tested. Like we obviously use AI of like no one writes code anymore, but you do have to test that.

57:02

Right. Because like you're going to deploy to all your customers running in prod. Like if you screw something up, like you're, you're dead. I mean, like for me, on my like shitty VC, like vibe coded things, I'll like, I'll make something and like, all right, cool. And then like three days later, I'll try to use like, this doesn't fucking work.

57:17

Like what'd you do? I thought you said this was good. Like, oh, I'm sorry. I just tested it. It didn't work, but now it works. Why didn't you do this originally? Yeah. That's for you. And that's your thing. You're like, fine. Right. You get annoyed, but fine. But like if you're an infrastructure, right, like people build products on top of you.

57:30

And so that part is probably the most important to make sure that doesn't break. So how do you figure out your roadmap then when you're like... We don't have a roadmap. So then how do you figure out what to build? I mean, we know what we're building now, right? And so you sort of like, I don't know if we have more than four weeks, six weeks of roadmap like that.

57:50

Really? No. So how do you do your planning cycles? Because I think the historical is like, once a year, we meet, plan out the thing. How do you do that process? There is no process. It's very, very hard today, and I've heard this from founders of much larger customers, maybe even public companies, that are in this mode. You really don't know, right? It's not that people are sitting around waiting.

58:12

There's more than enough backlog to make sure that there's stuff to be done. But you're constantly, as things changing, like changes, you let in, oh, maybe we finish this thing, or maybe we don't even finish this thing. We do this thing beforehand, and then these features sort of die off. We've actually had features that we've killed. It's like, oh, this is no longer necessary in this world.

58:31

Like this thing that we built a year ago, no one uses it. Like you deprecate the feature. I mean, it's out. So there's no long -term yearly roadmap. There's no, and that makes it super hard to plan anything. Yeah. I mean, I'd be like, if I'm the CTO, I'm like pulling my hair out. Like, how do I do anything? Yeah, my CTO doesn't have much hair either, but yeah.

58:50

Is that why you shave your hair? No, I have no hair. He has more than me, but yeah, he has more than me. So have you been able to figure out what drives these moments? Like when there's a spike of 50 people on the same day, is it like a new model came out that made things better, and all of a sudden something else became a problem? It has been the new model, obviously.

59:08

Like the new models, similar would be, there's new techniques on doing things where some of the researchers will figure something out, how to do something better. It would be new product launches in general. So like think lovable, think like GrokBot. Think like no one put a computer in every chat, then GrokBot did, and now like everyone wants that thing, right? So it's like those things.

59:27

So there's different things between technology and market and whatnot that do move the needle on those. Yeah. And then you have like security things. Something breaks out from some things like, oh shit, you know, we have to lock this thing down. Yeah. It's always interesting to me when just like all of a sudden out of nowhere, every talking head is like all of a sudden, like security, all of a sudden in the past couple of days, everyone's talking about it.

59:51

I'm like, it's kind of always been a thing, but there must be something. But it's in waves. And it's basically, some people, it's funny that we'd have some customers like, do you have these things in security? Security is becoming an issue. It's like, do we have time today? Not really. He's like, just like, I just tell them we're going to do it and it's fine type thing.

1:00:07

And I'm not talking about like a vulnerability patch or something like, like obviously that gets done. But it's like, oh, someone read an article somewhere and we should be doing this thing. So can I tell them they're going to do this thing and we're good, right? Yeah. And you guys have been growing really, really fast. I think you said something like, I mean, it's probably going to be, it's probably outdated the number I have.

1:00:25

And then by the time it's published, it's probably outdated again. But something like you guys have gone like 25X this year? 25X this year, yeah. Wow. So what is that? Probably more, but that was like from LAPT last month or whatever. Wow, yeah. So how did you grow so fast? Like what's going on under the hood? The way I think about it is we did a pivot, right? Yeah.

1:00:45

And then we went back to zero. But we've been doing, I don't know if we're doing tilts the entire time. What's a tilt? I don't have a better word for this, but it's basically, so think of it this way. Our first ICP for Daytona was a long running agent. So think what a GrokBot is or whatever, right? But they didn't exist a year and a half ago, right? But it was like the companies, as I mentioned earlier, the lovable competitors, whatever.

1:01:07

That was our first one. And so we started doing that. But then, so we launched in end of April. But around Christmas, we discovered that we're really good for post -training, right? And so what we did, it's the same product, but some of the features and your go -to -market motion, you prioritize now features that people on RL. We decided to prioritize features to be built for RL because there are like some slight differences and nuances that you want.

1:01:35

And the go -to -market team, we shifted them from all sort of events and types of people that we would address from general AI engineers to researchers. Okay. So you basically like fine -tuned the features and the go -to -market motion, like what you pushed? Yeah, what I pushed. Like, obviously we serve both. So our revenue is probably like 60 -40 now, maybe even 65, 35 RL to background agents.

1:02:00

But there was like a slight tilt in that, right? We sort of like focused on that a bit. And so that like drove. So you have that first growth that you're doing. And now you're adding this new sort of customer that you're selling to. And that's sort of like on top of that and growing faster. And then the next thing is like we found is that a bunch of people wanted GPUs in the sandbox to do RSI type stuff.

1:02:20

And we're like, okay, there's only one other company doing this. We should probably do this. GPUs are also harder to get to because you have to have the scale of them. They cost money. They're hard to get. They're all that. So we started adding that in there. And so the GPUs, which is still a smaller part of our revenues, most CPU is another sort of layer to that.

1:02:36

It's still a sandbox. It's still the same product. But it's like first user, second user, first variation of product, second variation of product. And now we're adding other variations. And all of them sort of like add sort of, you can think of them as like layers on the growth curve of Daytona itself. So there's not net new products that we're adding, so to speak, but it's like slightly different go to market, slightly different optimization on features that we do.

1:03:00

And just trying to make sure that we are where the current of the market is going. Basically just listening to customers. Like we need this thing and we need this today. Can you give that to us today? Interesting. Yeah. So we've been like the, there's no real product roadmap. You're just kind of, what are the customers seem to be wanting? And we like, kind of, like you said, tilt.

1:03:18

Tilt. Toward that. So there's two things to be very clear. We have a vision of where the world goes. And so we have bets. The same way we bet on Sandbox. We bet the agents want a sandbox. We also, from the beginning, it's been on our website since we started. Like we believe that you need a Windows and Mac and Linux. And we've always believed that.

1:03:33

Like that is, you could get a web archive. It's been on the website since we've done this as well. Why is that such a big deal? So it's not, why do you think it's such a big deal? So if you assume or want, let me assume, if you want your agent to do a job end to end, like I want to do end to end, right? Yep. Obviously. But now maybe this is not you, but you are some finance person inside of Walmart.

1:04:02

Just make that up. You have agents doing things. I would assume, I would actually bet a lot that to do every single job that it can do, not every single job, every single job it can do, at some point in time, it's going to have to turn on a Windows machine. I would bet anything on that. At some point in time. Inside of Walmart. Right. Instead of using a Mac and or Linux.

1:04:23

Exactly. And or both. Exactly. Okay. The reason why is that there's probably some, I'm not going to say Vibe Coded because it's not Vibe Coded, this piece of shit app, because I've worked in these companies, that someone built that has like a local database somewhere to do some random thing. And the only way you can get that information is you have to log into that Windows and do that thing, right? And so if the flow of the agent breaks, because it doesn't have access to Windows, you're going to be like, fuck me.

1:04:50

It's like, you have to go do this thing, right? Yeah. You want that there. And so for that point in time, you want to spin up a sandbox with the window, do the thing, kill that. And so that's why I believe that. And there's other things, obviously. There's things like Q &A. There's building actual Windows apps. There's a bunch of stuff to be there.

1:05:05

But that's like the most interesting part to me. And also, I don't believe that computer use or the ability to use these apps will disappear in the next decade. I believe that some of these apps will still be here 10 years from now. Not everything, in the sense that everything will be headless. And so you will need these types of machines to get that done.

1:05:24

And so that was one of our early bets. It's less about customer demand. But we've literally had last week Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:05:46

Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner

1:06:10

Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:06:17

Turner Novak, Turner Nov And so you push forward and things that people are like banging down the door like can you do this thing for me right yeah can you give me the windows can you give me the windows like or whatever it might be right can you give me that thing right because I need a lot of that thing can you give me that thing yeah I'm saying this is like have you ever heard this joke before if your CFO uses a Macbook you're definitely going to prison yeah exactly yeah you probably want your CFO using Windows like mine are you I gave him a Macbook but he doesn't want he wants a Windows inside of it he runs a Windows VM inside of it I made it oh nice yeah I've seen

1:06:50

people do Mac inside of a Windows Windows too yeah and he's like I really want a Windows I don't want to use a Mac it's just like everything's shit I just want to do a Windows I'm like how so and he shows me he's just ugly dude you have It's like, it is what it is. Yeah, well, I think like I've always been a little bit influenced by growing up.

1:07:06

I mean, I graduated high school in 2009. So taking classes in high school, I took a lot of like graphic design, programming, like tech drawing, architecture classes. And like, they kind of like for one note, you can only get AutoCAD on Windows and Mac. Oh, can you? Yeah, no, you can't get on Linux. So just an answer to your one of your questions.

1:07:25

Yeah, there you go. Well, and so like, I had grown up like, okay, Macs are for like old people because they have like one button and they're super simple. And it was before like Mac books. They didn't have a right click at all. Yeah, they didn't have a right click. Yeah, it's like I had to use, you know, a Mac, like desktop box thing with like the different colors.

1:07:48

Like we had that in my art, in my art and like graphic design class. And I was like, these computers suck. I fucking hate Macs. And then of course, like the nicer Mac books, like they've evolved since then. But it's just always I've been biased by that even 20 years later. I thought that like the old school. Yeah, me and my friends, we would just like hang out.

1:08:06

We would. I mean, this is like a total side. But in high school, we would take all the computer classes just so we could just go on the computer all day during class. Yeah, mid 2000s. Yeah, I did. So I'm much I'm older than you. I finished. I was. What were Commodore 64 or whatever it was when I was like, we never had one of those. Yeah, I'm probably like just missed that.

1:08:25

Yeah, I'm super old. And so the only way to there was like one per floor in our school. Whatever. Oh, wow. And then I became like, helps the teacher that was like the computer teacher. And that would go fix other people like computers across in the school district or the school just the school. I just did that because I can play the computer because I fixed it and I would play whatever video game but the video games were total garbage pong or whatever.

1:08:47

No, it was better than pong. But they're like 525 floppy disks. They're like the big ones. Oh, yeah. The big one flood the big ones. Yeah, I would like load that in. And I remember there's like some dungeons or whatever. I forgot the name but like shitty. But so I would go could fix them. But the only thing you could do to fix them was like there's three cables or RGB cables that would connect into the monitor.

1:09:04

Okay. And something you anyway, I did similar just so I have access to computers. Yeah, me and my friend, we take like all the classes like I remember there's one class. We like half the class was learning Microsoft access. I don't know if you ever used Microsoft access. Of course I did. It's like a super shitty database. I know it really well.

1:09:19

It's a database. But it is, it's, it's Airtable. Oh, that's what Airtable is? That's what, in my definition, that's what Airtable is. It's like a easy to use database for the average person. Right. Yeah. And I remember the guy who taught that class was just like super nerdy, dude. Do you know what access still exists? Yeah, it does. Yeah.

1:09:36

I looked it up. It still exists. There's a new version. I'm like, who the hell uses this? I mean, there's like businesses that use it. They're on it. And like, that's my, the guy who taught the class, he was like drilled in our head. Like, like the world runs on Microsoft access. Yeah. You got to use access. Yeah, that's why you have, but now you have Airtable or Vibecode, whatever, right? Yeah.

1:09:51

Because you don't need, because it's just the, you know, super based in the background and fine, it works, right? Yeah. Yeah. No, we also had a game design class where you could, you could like Vibecode video games. There's like some programming software or like game design tool where you kind of like, it was kind of like a WYSIWY, Dreamweaver, type, click and drag, but then also like, like sort of code games.

1:10:14

And so, and the teacher did not know much about how computers worked. Yeah. And so me and my friends, we would just be like playing, playing games online. We like working on our games and like, you know, yeah, I don't know. Those are the days, just like the high school. Just, I feel like we're in like a different era though, where like where we grew up, the grownups, quote unquote, like didn't know how computers works.

1:10:35

Yeah. You could kind of get away with stuff. Yeah. But I feel like today, like I've actually heard in like middle school, high school. Now what you do is to get around the blocks in the computers, you'll put links in a PowerPoint, like in a Google Slides or in a Google Word doc. Because if you click a link that's blocked by the school through the browser, through Google Slides specifically, it gets around the blocks.

1:10:59

Oh really? So people will use that now to play games in school. In school, yeah. So. Yeah. Did you go to high school in Croatia? Croatia. Very different vibe. Oh wow. Very different vibe. What was it like growing up in Croatia? I went there in high school. So it was an elementary in Toronto, which is, as you would expect, anywhere from like North America.

1:11:16

Yeah. Did you know, did you grow up learning both languages? Yeah. I was kind of shitty at it, but I grew up, I couldn't understand. So you were kind of like a foreigner, sort of. House foreigners, yeah. Kind of. Definitely. Culturally very foreigner. Yeah. It's very, very different. The thing that was interesting to me, and this was a contrarian take on Twitter, which was that all the things that we define as high school, what I define as high school, like, you know, whatever.

1:11:41

Cool kids, nerds, jocks, whatever. All those variations and all those social differences. That existed to a degree, but people hung out with each other. It wasn't like I won't hang out with the nerd or with a jock or whatever it may be, which happens here in like Western society. Yeah. Or even Far Eastern. There, because people were like post -socialist, post -communist, people were very similar in socioeconomic hierarchy.

1:12:13

And though someone could be quirky or someone could be better at sports, like everyone hung out together for the most part. Like people didn't pick on each other in that kind of way. And I'm like, oh, that stuff that they do is actually a social construct that we put together. It's not an actual natural construct. Like the jocks and like the nerds.

1:12:31

Like the picking on each other, like the beating up, like flushing your head down the toilet to like those types of things. Yeah. Right. Like that actually didn't exist. So I went through high school without any of that, which was shocking to me because I went through that in elementary school. Right. Yeah. That still existed. I'm like, this is so bizarre.

1:12:47

It just must be like a cultural thing in Europe, maybe, or in Croatia. No, it was Croatia at that time. I think it's different now. Because the thing is, there was no large difference between people that had money and did not have money. It was much, the band was much closer. Because you were in a, it wasn't communist, but a socialist. And in those economies, there's very different, there's very small differences in financial hierarchies.

1:13:12

Right. And so because there's very small differences, people don't pick on each other because like they have good shoes or not good shoes. Right. Like you didn't get the latest Jordans or whatever, whatever. It's cool. Right. And so that didn't actually exist. And that was like mind bending for me. It was actually quite interesting. Yeah.

1:13:28

Yeah, that is pretty interesting. Slightly different topic. But are there any things in agent security or agent usage that most people kind of don't appreciate yet or don't really realize is going on yet? Maybe we already talked about them. I mean, I just say one thing which some people push back, some people don't push back. But I believe that the vast majority of security is already taken care of.

1:13:53

You think so? Yeah. Because here's an example. You go to work at any bank, whatever, you get a little key card for physical access. Yep. And you get a laptop that's like locked down. Right. It's completely locked down for you. You do the wrong thing and the security guard show up at your desk a minute later. Exactly. Exactly. Same thing. That same software works for agents.

1:14:16

Okay. Exact same software. Like if an agent goes and picks off the internet, copies something to something else, it's still logged in that same thing. You can still see that same thing. Right. And so I don't think people appreciate that that is really to a large extent solved. The problem that people haven't solved is that in their mind, understanding that people should be able to create agents with credentials of their own and then with a computer of their own.

1:14:42

And then that is getting logged. So essentially, so whatever. So you work at whatever bank. Right. Yeah. And so you want to create a new agent to do something. It should get some email, which is its identity and an email, actually. And so it could be your name underscore whatever. Yeah. Sub agent 365. Doesn't really matter. Just has one. Right.

1:15:00

And so it gets its own email. It gets its own computer. That computer spins up with a default image. I'll pick on Windows right now because most of them are Windows. Default image that has all the things that you have on your Windows that you're working on, right? It has all those restrictions. And also, I forgot when you're creating that, when you are initiating that agent, you can set what privileges it has in the enterprise, but it's a subset of the user.

1:15:22

Of your privileges. Okay. Of course, you can never, the agent can never go outside of what, like, me, Turner Novak, the employee of Morgan Stanley. That makes no sense, right? You can only have a subset of what you have because it's working for you. Yeah, that's fair. Right? And so that's basically it. And so that, the only part that's missing is that, like, the security team or IT team will enable you to create that account.

1:15:48

And if there's some interface to make that super easy, right? To, like, spin up the ID and spin up that. But everything there is already there. Now, the risks that happened are, that can happen with that is that the agent can basically, it can't do things that it usually can't do. It can maybe leak data, right? That's the tricky part. Is that the big issue everyone's kind of up in arms? Well, this is about my thought about it.

1:16:09

It's less about what they're up in arms. It's like what I believe is, me personally believe is feasible or not feasible, right?

1:16:44

Oh, okay. Because it'll, like, access your password that's just saved in a Word doc somewhere. Exactly. Or it'll access your bank account, which is in there. Or the company's bank account, which you have access to. Like, it does that thing. If it's a sub, if it's a sub agent with its own account, it has, it doesn't have that. Or if it does, you gave it to it.

1:17:00

Yeah. Type thing. So it's actually, it's actually more secure, not less. And all the software that exists, I'm sure Apollo Alto Networks has a bunch of software that, like, stops us from doing these things. And even stops us from leaking data. Like, if you copy something and, you know, put it in, you don't, you don't have, like, a USB drive or whatever it is, like, it will flag the security team.

1:17:21

Like, you did that, right? So if an agent does that, it's gone. If it uploads it to some website, it'll probably, there's probably software that can flag that. So a lot of things are already there. The risk is that, to your point, you'll have a swarm of agents that will try to break out. But in this case, you don't have a swarm of agents.

1:17:37

You have one computer with a single agent running on it, right? So the time it takes to actually try to solve and hack and break things is actually quite low. And even if it can, the point is just to have that software be more. So an equivalent, I'm picking on Apollo Alto Networks, but anyone else in the security space just has the equivalent of, like, more sophisticated software that's already doing the same thing.

1:17:58

Because you're stopping it from accessing data that it's not supposed to access, and you're stopping it from leaking data that it's not supposed to leak. Those are the two things they're trying to solve for. And people are trying to make it as if it's, like, this whole new infrastructure needs to be built. But you don't think so. I don't think so, yeah.

1:18:16

So should I, if I'm listening to this, and I'm, like, are there opportunities in security right now? I'm sure there are. And so the things that are, and so kind of in everything I say is contrary to myself sometimes, which is that there's probably a product that is not solving something that an agent can do. Because if you give an agent, the risk is the agent has a computer, it can do different things, right? But it's the same things that you can do, right? You could do that today.

1:18:40

Like, if you're in the bank, you could do that today. Like, I could go and try to hack into someone's bank account personally. Yeah, exactly. Or, like, that's extreme, right? But let's say something simpler, like, you have access to, you know, whatever Goldman Sachs is whatever data. And you can try to email it to someone, other bank, and, you know, corporate information.

1:18:57

Yeah, exactly. That's what you can do. But Goldman is like, hey, you just gave them all our secrets. Like, you're fired, and we're suing you, and you're going to jail. Exactly. But there's also, like, there is that. But there's also, like, technical ways to block that from happening. Oh, okay. That are already in use today, right? And that's why I'm, like, less worried.

1:19:12

It's already there, right? Those things are already there. Now, is there other ways that a single agent inside of a single computer can still leak that outside of that? I'm not sure. It probably is. But that is more, I think, a more progression into that software rather than a revolution into that software itself. Interesting. I think, like, one of the fascinating things about Daytona and just, like, the whole, like, the story behind, like, you and the journey.

1:19:36

When you were first kind of pitching the company and the product to investors, I think that the phrasing is just, like, everyone just kind of laughed at you. Like, it's like, why are you even doing this? It's like, this isn't even, like, a pitch. Yeah. So what was it like when you were first raising money for this? Oh, it was terrible. I mean, terrible.

1:19:51

Like, I'll take a step back. So we did our pre -seed and seed on the old product, right? On the original product. You were based in Croatia, probably. We're based in Croatia, yeah. And that always is tough because you have, like, there's a price discount and a perception discount, right, for these things. Also, the perception discount is that none of your team members or yourself have, like, the credential history.

1:20:14

So you weren't, like, Uber, Google, whatever, school, XY, none of that. Like, every logo that you have on your CV is, like, what the hell is this? Like, the University of Croatia. Yeah, what the hell is this? Who cares, right? I mean, that's actually one thing I run into where someone's, like, oh, that's, like, the best school in India. I'm, like, I've never heard of it before.

1:20:33

Yeah, exactly. And you're, like, whatever. Yeah. And then you discount that, right? Yeah. So there's problems with that, obviously. But we did that. But the interesting thing that less people know about is that we pivoted and then we decided to do our Series A on the new product. We went out, we ran a process, and we didn't raise. So you got no bites.

1:20:56

How did that go? That was, so there's a bunch of that. So it was, basically, every time we do fundraise, we run a process. It's just, like, the easiest. Like, it just makes the most sense. Basically, when you run a process, you invite everyone that you think you want to talk to. You put them in. You have a timeline that you work with. The timeline helps because, basically, it gets VCs that are not, that don't have conviction basically say, you know what? They'll have, like, oh, they have an out.

1:21:23

They're like, I can't get it in time. I can't do whatever. I'm traveling. Whatever. It's fine. You're like, cool, thanks. And you get it for not wasting my time. Like, that's actually good. Yeah. And it's like, if you listen to any VC talk about it, like, you want a quick no. Yeah. As fast as you can, right? So it's a quick no. And so that gets it there.

1:21:40

And then the other thing is, it's less about, like, getting the best dollar, like, the bang for buck. Because mostly, like, other investors will be like, where are you in the process? And it's like, if you're, like, already in ICs with a couple, then a couple also drop down. It's like, we can't get to IC that fast. And so you keep them. And hopefully you get to an outcome where you have, like, one or multiple term sheets, right? That's where you want.

1:22:00

And if you get one, you actually get a bunch more automatically. Just like, if you get one, the more that are in the process, the more I want to give you another term sheet. Which is, like, for the founder, it's a better thing. But ultimately, you want to get one and sign one, right? So outside of, like, techniques of getting things better, you run a process to do that.

1:22:16

And so we ran a process to get to quick no's. And we ended up getting basically all no's. Like, it was no. That was devastating. It was, like, devastating, right? Like, imagine going out, telling your team, telling all the investors that you're raising, and you come back with nothing, right? Yeah. And then three months later, we raised an A afterwards.

1:22:37

So what did you, did you, like, update your investors and say, like, hey, didn't go as planned? Like, did you just, like, not do anything? Like, what did you tell the team? How did you? Oh, the team, I told, like, that's what happened, basically. To be fair, we did have a one -term sheet of a non, you know, S -tier investor. It was, like, half the valuation that we're sort of directionally going for.

1:22:57

And me and my co -founder decided, you know what? We're not going to take it. We're not, we we hadn't ran out of money. We're not like running, we had revenue and all that. And we're like, let's go do this again. And so we went back to work. And so we went back to work for three months. So why did everyone pass on it? Do you know? No one understood.

1:23:13

Very few people understood. Oh, they like didn't get what you were doing? Yeah. And not from a technical perspective. Not that. It's just like, we don't think this makes sense. Like when I told people like, agents will need computers and they're like, why the hell? Like that was non obvious. That was very non obvious at the time. And the other thing was we, which might've been a mistake.

1:23:32

We came out with a price point that we wanted, which probably not ideal. Maybe yes, maybe no. What should you do instead? I don't know. It's more like, this is how much we need. And then let the market figure out what the price point is. Yeah. I've heard, I've heard people say, we, you know, we want to raise five to 10, hit 10 to 20% dilution or something.

1:23:50

That's a pretty big range. That's pretty big. You could raise 10 at 10% dilution or five at 20% dilution. It's very different. It's like very different numbers. Yeah. And also, I don't know. I'm not a VC, so I don't, I don't know how that looks, but I feel that when you put a range, it's like, I don't know. It's like, it feels worse. Yeah.

1:24:07

In the range of like how much you're raising, it feels like, oh, I don't even know what I'm raising. How much do I need? Like you, yeah. If you need a number, then you need a number. You don't need five. If you need five to 10, that's very different. Not even talking about valuations. It's like, it's very different. Yeah. I think you should know.

1:24:22

That's, I don't know. Like this is not financial advice. I have no idea, but I would try to put in the number that you want, but less around that. The other thing is the ramp was really fast, but it was only three months of data. And because it's three months of data, it's also risky on the VCs. It's like, you know, three, I would assume it was like three months could be, you know, not even fake it.

1:24:44

You could game some stuff. Yeah. In three months, you can do a lot of things. Right. And so then when we came back six months, like three months after that, you had six months of data. You like expanded, you had revenue growth, like where you said it was going to go. You had more customers. These customers still stuck with you. They can verify that and became much, and also time passed.

1:25:04

It became more obvious to more people. And so we did again. But the thing that's really hard that I don't know if people listening will internalize is telling everyone that you're raising, not raised, like not everyone knows that you didn't raise or not raise. Like you just like some people dropped off along the way. Then come back three months later and call them and say, Hey, we're doing the A again.

1:25:24

That hurts. I'm just like, what's wrong? Like this must be like a busted company. I don't want to talk to you. Yeah. But you know what it feels like as a founder? Like you have to go and come again. I told you I was gonna raise last time and I didn't. And I'm back to ask you again. That's like, just like, it is your ego is like. Yeah. So how did you approach? I'm assuming there's some people that you had to do that with.

1:25:46

What did you do? So I was basically like, listen, here, this is what we had. We did. I mean, we said we got that one term. So we ended up saying no, like people can believe that or not believe that. So it's like, usually whatever you say, people say you didn't, you weren't successful. Right. So you screwed that up. You fucked up. Right. You were not good.

1:26:01

And in their life, they were like smart. They did. They didn't do it. Right.

1:26:06

But you have to come back to them and basically say, Hey, we went back to work. We decided to go back to work. We went back to work. We did what we're going to say we're going to do. We did it. Revenue is where we wanted customers where we wanted. And it here are the proof points. Here you go. And the deck was actually interesting. The pitch deck.

1:26:21

We had the old, we had all the old slides. We didn't do a new pitch deck, but all the graphs, all the graphs, we just did side by side. So what we did is you had the old pitch deck of like, I'll draw it out. The bold pitch deck of growth was like this, right. It's like growing up. And then I applied at the same ratio. I applied the new growth and it goes off the screen.

1:26:44

It goes off the slide. Right. Oh, wow. And then it was just like two slides. So you have the first slide. The first slide is the old one. And on the old one, I just drew the extension and extension just like breaks off the slide. It literally breaks the words. Like designers like break the words. It goes off. And then the next slide is like zoomed back.

1:27:00

So you can see like normalized. And then you see all the old growth looks like a flat line. Right. Oh, because you had just gone exponential. You got exponential. Right. Oh, interesting. And then all the numbers, like all our growth numbers, retention numbers, all of them was like old slide. And then I like crossed literally just a crossed out old numbers.

1:27:17

Here are new numbers. And that was that was our pitch deck. That's that's pretty cool, actually. Yeah. And like, oh, they were like, OK, because it was like less about I don't want to tell you a new story. It's the same story. I didn't change the pitch. Yeah. It's like literally here's the facts. Here's the numbers. And then so first mark ended up leading that.

1:27:32

They let it be as well. But they are also ones that passed on the the first time we did it. So it's like, yeah. So we went through that motion again. And so. So did you raise the precedence from a Croatian investors? No. No. Was it just a precede were precede were like mostly angels. So like founders, you know, founders are like Netlify and Honeycomb, whatever.

1:27:53

Yeah. Superbase found Paul was like a bunch, just a bunch of founders. Those were the we wanted to do sort of a party round in that sense. And the seed was led by upfront. They're based out of L .A. Oh, yeah. Yeah. So they did the seed round. Nice. That's cool. Good win for L .A. Yeah. Good for them. And you. So you actually moved. I think you like tweet about this.

1:28:13

Like you're kind of you guys have been flying back and forth all the time. You actually moved to San Francisco, like moves your family over a couple of weeks ago. Yeah. A couple of weeks ago. So we'll see how like for now it'll be like this two and a half months dint. And then we're gonna go back for holidays, then come back in the spring type thing and sort of see if we'll make it a permanent long term.

1:28:31

But definitely, you know, in the sense of like I'm here, I'm not an airplane. The worst thing is that basically out of Croatia, you fly to Germany for the most part. And you can fly United or Lufthansa. That's it. That's how you. Oh, really? From Germany to Croatia? No, to US. Oh, OK. There's Croatian Airlines for Croatia. OK. And so I flew so much this year that literally the staff there knows me, not because of my status.

1:28:57

They're like, I've asked. They're like, no, I saw you last week. It's like you're literally and that's not a good thing. I'm like, this makes no sense. If the staff on a plane knows you because you're that often, there's probably something wrong with you. So, yeah, maybe not. Maybe that's a good thing. Maybe actually there's this guy. He like he has like the world record for like most airline miles flown or something like that.

1:29:16

And he's like wrote this post about like how to how to get the best experience out of your flight. And like he says, the thing you do is right when you get on that first flight attendant, you say like, oh, it's good to see you again. Yeah. Even if you have never met them before. Because in the flight attendant, it's like, oh, yeah, this is like a regular.

1:29:30

Like, I'll treat this guy a little bit better. Yeah. So actually, there's there's this one guy. Do you watch college football at all? No. College football. OK. There's this guy who looks like the coach who won the college football national championship that is a flight attendant on one of the flights that I take a lot to get to San Francisco.

1:29:46

And I see him like once a month. I don't know if he I think he kind of like recognizing he has the same name as the coach and he looks just like him. So I've always I've been thinking like, should I bring this up to this? Yeah, bring it up. Yeah.

1:30:00

One thing I know you've said is that kind of like the next pace of like the next wave of A .I. is like A .I. deployment specifically. So what does that mean? Like A .I. deployment is going to be like a big thing over the next 12 months. So what I was more alluding to is that when we started, as I mentioned, like we started with background and long running agents as a ICP that we sold to.

1:30:23

We didn't actually understand that most of those actually don't exist. Like if you think of. App layer. Products like marquee brands. How many can you count? Literally. So you have like outside of like the foundation of all companies that have their own products. You have Harvey and Legora. They're actually deployed and working. They're like Apple.

1:30:44

Right. You have those. Let's say Clay we can put in there. Gamma maybe in there. Right. They're like app layer. A .I. Products that are actually really used. And now you have Muse. It seems right. The rock bot seems to be doing something. So like the vast fan majority of these apps of the Peel with Turner, and I'm a fan of the Peel with Turner,

1:31:21

and I'm a fan of the deterministic apps for the most part. That's fair. And so, I mean, speaking about like OpenAI, Anthropic, the big like skilled lab, so you kind of use them as a partner, but they also have a sandbox product. I mean, how do you think of that? Because it's got to be kind of like, keep you up at night a little bit. I mean, there's a lot of things, but it's great that they have been open to or pushed or whatever it might be to open up their platforms to offer, bring your own sandbox.

1:31:53

That means, and I'm, you know, inferring here, I'm not saying it's true. That means that their customers have said, we want these other products instead of that product that is bundled with there, because otherwise you wouldn't offer it, right? Yeah. And so, for us, it's great because all the companies that are enabling this, so OpenAI, Anthropic, with their like Claude managed agents, like we've worked with Cursor and Cognition, all of them have like bring your own sandbox features, and we're included in all of them.

1:32:25

And they do different, which is interesting. It does different from each of these companies. They have different sandbox providers in there. But what we're really proud of is that we're part of all of them, right? So we're in every single one of those. How does that work that you're in all of them? In the sense of like, there's an API, there's like a, you can pick default, or like a, here's the default options of sandboxes.

1:32:45

You can have ours, or you can have Daytona XYZ, right? Or XYZ Daytona. Like we're in every single one of them. So the thing that we're doing is, that's back to the attention, right? Are we available everywhere? Is when you look at a list, is Daytona always on those lists of these products? Because I might have a Daytona enterprise deal signed.

1:33:04

And like, if I just have it available in that environment, I'll just, I'll probably default to trying to use that one. Exactly. Because then you're not, you're not doing the double the clock. Because you already have a contract here, and then you do that. Or Daytona also has a BYOC. So it's like, oh, do you want to run in your VPC? Then you have Daytona running in your VPC, and then it goes back.

1:33:19

And then you also use your commits against DWS for whatever it may be to run that in there. One thing I think is kind of interesting about your like mindset, you don't really like romanticizing failure. Like a lot of people, people do. And you can learn a lot from failing. Do you really? I failed so much. I don't think I learned that much.

1:33:35

I mean, well, so I think personally, when I like really fuck up, I just like, I'm like, all right, never doing that again. I'm never making that mistake again. But I feel like you can also learn a lot from success too. I think you learned. So I've, I've said that I failed upwards. I failed, like most of the things I've done, I've failed. Like most of the things.

1:33:53

What are the things you failed like the hardest at? I mean, nothing was successful. Hence everything I failed. Right. Like nothing was really successful. Nothing was breakout. Right. So it depends on what is your like lever of success. But I've had four companies, you know, some of them I ended up shutting down. Did I make some money along the way? Yeah, sure.

1:34:12

Did it get me to the next thing? Yes, it did. But they weren't successes. Like I didn't make Stripe or Google or whatever, like define whatever your success is. But you're not living on an island right now. Like you're, you have to do this podcast. Yeah, I would probably do it. Even if I had my island, probably still doing this thing, just to be very fair.

1:34:28

But in the sense of like, it's not, not even the monetary part, but did, is the thing that you're solving, like, does it solve it for the vast majority of that market that you're solving it for? Right. And it's, it, none of them were, hence I had failed. So in my mind, I've failed at those things. But when you think about, even when you think about hiring or the team you're putting together, you actually, I didn't understand this till fairly recently in my life, is that you actually want people that have won more rather than have lost more.

1:34:54

The reason is you actually learn much more on in winning or in sense, you learn how to win when you win. And then you can apply that to win again and then win again. And the mindset is I can't win because I can do that. When you fail, you learn how to fail, which is not a bad thing. I want my kids to learn how to fail, like to internalize that because you always fail along the way.

1:35:16

You don't want them to ever not understand what failure is, but you want vast majority more wins than you want failures. Because the other thing is when you learn how to fail, it's like, oh, you brush it off. You go to the next thing was fine. But if you fail all the time, all you know is how to brush it off. Right. Even if you're like, you don't, you know, if you continue going, you actually have never figured out how to unlock and win.

1:35:39

And that's why I think there's a contrarian take. I don't know. I think Peter Thiel said it or something. Someone said it like that. Some people don't like that much or have different opinions on, which is literally you don't learn from failure, but you do learn how to fail. And so my caveat on that is like, you have to learn how to fail because you're certainly going to fail sometimes in your life.

1:35:56

Yeah. You should win more because you actually learn how to win more. I feel like you get momentum when you win too. Like when I think about like just the way that I operate personally, like when I have a good day versus a bad day, like the good days, I'm just like, oh, fuck. Yeah. Like let's do more. Do more. Yeah. Do more. And the bad days is like, so you have to, when it's a bad day, my thought is I had a bunch of bad days.

1:36:18

You just have to go through the motions, which is like, you have to wake up. You have to, whatever you do. Like I try to do, like go to the gym in the morning. Then you like have to do work, whatever that, even if it's the shitty work, just get the work kind of done. Yep. Eat food, go to sleep, wake up in the morning. Like just don't stop.

1:36:33

Yeah. But as like the productivity of that day is like garbage. Right. But it's like get to the next day and sort of gets you out of that shithole. Yeah. Which you want to, but to your point, the momentum, if you win, then win, then win, then win. Then you compound it. Then you're like, oh, let's do more. Let's do more. I'm not going to sleep.

1:36:47

I'm doing like, go, go, go. So, yeah. So my take on that is like, you want people that, that can win. And that is something that people don't like to hear, but yeah. Interesting. One, one, one interesting thing too. I know the way you like think about documentation, it's like kind of part of the product at Daytona. Is that because the way agents interface with it is so much like what's important to documentation? That's an older one, which I still believe is true, but it was more around humans, which when you think of our product, which is mostly invisible, it's just an API, right? Right.

1:37:19

So if it's an API, what is the product surface that you interact with as a, as a person, right? You interact with the docs more than anything else. Like obviously like when you're integrating the API or if there's a CLI there, you can integrate there. There is some product in the CLI they can do, but basically for the most part, the thing that you interface with as a company, when you see our logo, it is on the documentation itself.

1:37:40

And that's why I basically mentioned that. That is, that's very, very important. I think people are now all write documentation because AI writes it for them. So it's less of a problem, but like three years ago, you had a bunch of places that didn't have documentation. So the documentation was garbage. And so hence the point about that. Yeah, that's fair.

1:37:57

And I think, I mean, you mentioned you just hired someone to start doing sales very recently. Yeah. So you grew 25 X. I don't know. Is there like a public number that you've stated, like how big the company is at this point? Not yet. Not yet. But there was some public numbers last year. And so you can do some math on that. And we will state very soon where the number is.

1:38:16

Okay. Right now, like the public number times 25. I'm not going to tell you that. Oh, you can't. I'm not going to tell you. You have to do the math for it. I want it to be in here so that people could just like hear it. But it's not going to, I don't know if it's going to be public at that time. So it is a significant, it is one of the significant milestones.

1:38:29

We're there. We're there. Okay. That's fair. So, and you didn't really have anyone doing sales. So like, how do you, how do you grow a company so fast and like not have salespeople that are like doing this stuff? I thought you kind of like, you kind of, a lot of people say you need them. Like, how did you do this? I mean, it's definitely a PLG company.

1:38:47

And so we've done that. And the way we have done that and still do it today is that, I mean, we bring everyone into Slack, all our customers in a Slack connect. They all have a Slack connect channel with us. We have like, I don't know, 1500 Slack connect channels or something like that. It's an insane number. And the other thing is that to our betterment or detriment, everyone is in all the Slack channels of our company, right? Wait, every employee of the company? Yeah, there's like 35 of us.

1:39:12

There's not that many of us, right? Okay. And so like the engineers will reply directly to engineering questions. People on the go -to -market will apply go -to -market. The support people apply like general support people. And so like our customers get like very fast replies and solutions because we're just like all there. Because there's 35 people that are getting the ping.

1:39:29

Yeah, which is, has other problems, right? Yeah. Which we're now solving for and can't solve for, but like that is how we've done that. And we've done that too. And the thing is you see this with a bunch of startups, which is like, they're a very small startup. You know, they have whatever handful customers and they're like, oh, my customers love me because I reply very, very fast.

1:39:47

And it's like, yeah, because you don't have customers, right? You don't have a large number of customers. And that should be that way. To be very fair, there's companies that have little customers and still don't reply. So that is good, not bad. Yeah. But we've continued up to this velocity where we still like answer. Like you'll get me in the Slack, literally.

1:40:06

Like I still reply to Slack. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that.

1:40:22

Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that. Like, I'm not sure if you're going to get to that.

1:40:32

I think it's less about engineer. People have to be engineers to do that. It is more about how people feel towards the company. Obviously, everyone has stock in the company. People have been there for a while. Stock's worth much more because a billion -dollar company right now. People have ownership. Because they have ownership, you don't have to ask them, hey, did you reply to this person? They just reply to the person because they're on the toilet with a phone and they just, oh, there's a ping.

1:41:03

I'll just reply to the ping. As people come in at different stations of the company, people don't expect to do that. How can I keep that culture that people, when they see that, they will reply to that thing. It's more about the people that you hire versus anything else, I think. I wish I asked you this earlier because it more tied in with when you were fundraising.

1:41:23

A lot of the fundraising was just pure cold emails. I think a lot of other stuff, you've just done a lot of cold outbound.

1:41:31

So, everyone says you got to get warm intros for this stuff. How do you send a good cold email? Yeah, I think cold emails are like, what's his name? Harry Stebbing says his superpower is sending emails. I must know his mind. I would, everything from our pre -seed, which is the party round, I forget the number because I'm not sure I remember correctly.

1:41:52

It's either 400 or 600 people that I've reached out to, right? Cold. And so, that is a large, and these are not randos. It's not like, oh, get link of like VCs, sends to all. It's like, who invests in stuff? Who does it? There's a whole like market map of people interested, whatever. And you just like cold email the hell out of these things and multiple times cold email.

1:42:16

Same thing with the events that we do. Like, again, we would cold email 100 people to get 20 people. Like, I just like cold email brute force the thing to the max. And you can brute force. People, for some reason, are scared to send emails. I don't know why. Yeah. But the other thing is people send shitty emails. They're very, very, very bad.

1:42:34

What's a bad cold email look like? If it's long, it's bad. I'm not reading that. So, just objectively, the longer the worse. The longer the worse. And people have a big problem with that. People have like, especially when I would like, when I work with a team, they're like sending this email. Send me the draft and I see it. They're like, I'm like, why the hell did you write this? They're like, well, I want to explain all that.

1:42:54

I'm like, they don't have time. But I don't know how to write it short. Write it short. Like, it has to be like a tweet. It's like, basically, what you're trying to do for the person is like, why the hell am I even opening this thing? Like, the subject, why am I even like opening that? And then it's like, really quickly, some anchors of the way I would like cold email things.

1:43:13

It was like, hey, we've whatever that we're building this thing. I think you're interested because of XYZ. It's a long one sentence, right? You're an infra founder. We're building this new infra. You might be interested. We've raised money from this person, this person, this person, this person. So you're looking social sort of like, oh, I know that guy.

1:43:34

Yeah, I know that guy. I can verify this. Yeah. We have revenue X or and or we've raised this much of the round. So it's like, because also it's in the tail end. It's much more truthworthy or valuable versus like, I'm raising a round. I have nobody. I'm pinging you. So it's like if you're cold email and you're raising, I'm going to make about a million bucks.

1:43:55

I have 600 of a million. That's a very different signal, right? That's if you're interested, let me know. That's it. That's the whole mail. Right. I think some people saw screenshots on Twitter of my old emails. I think it's generally that like that is a thing. And so I can like skim through it while walking. Like if this is interesting. And so basically, if I'm reading that email, it's like.

1:44:19

Infra, these guys I know invested. A bunch of the round is done. Do I want to talk to them? Why not? There's a 50 percent chance that you're going to say, send me the deck, like send me the deck. Right. Might even talk to them. So you don't send the deck first. Usually you don't even send the deck. Nothing. I just send that. And it's also plain text, no HTML, no unsubscribe, no type stop for like all that shit.

1:44:44

Nothing. You don't do the AI generated third follow up. Yeah, that's it. And if I do a follow up, like I will like literally, I'll do it again. It's like, hey, you know. Also the thing, the way I follow up is, especially this is the early days, if people are interested. Yeah. It is the early days. It would be like, I added XYZ to the round.

1:45:05

And I, and I, we got 30 percent more done as you're showing momentum to some person. And then if that, that might interest them, it's like, oh, these guys are still like, they're almost done. They added these other people. And I, because when you think like you invest in people, what are the things, if it's not a warm intro, like what are the things that you need to actually be interested? Right.

1:45:27

Yeah. I think a lot of it is like how fast you move. Right. And so if you're like, we get all these things. I'm like, oh, okay. Like you're, you're shipping stuff. That's pretty important. Yeah. Like one of the things I do with like, with raising my phone, it's like, you know, the biggest LP for my last phone just re -upped. Like, it's just like giving you a quick update on this.

1:45:45

Like I'm doing another close or, oh, I should invest in this cool company, Daytona. We co -invested with like so -and -so. Yeah. And then they're like, oh shit, like co -invested with, I don't know. Actually, who, who led the Daytona around? First Mark. First Mark. Do people like First Mark?

1:46:05

Show them like the proof points. Exactly. It's all proof points. It's also for proof. That's what I'm saying. What you want is like very short, skim through my eyes, the other proof points in here. Yeah. That is all you need. Right. Yeah. And so you feel like people think, oh, I need to give him all the information or them, all the information to make their decision.

1:46:23

And they're like, no, you just want the proof points to get to the next. You just want the reply. Yeah. All you want is the reply. Nothing else. There's another way you can do that if someone's very important. And we did this recently. So in this round, we got, you know, a bunch of people, founder of Zoom, founder of Stripe, founder of Shopify, like just like a bunch of these amazing founders.

1:46:46

Yeah. And the vast majority of them that we got was cold, but also it would be warm where we would send like, you know, people that had connections. It's like, can you forward them this email? Right. And so they would get bombarded. And I forgot which one it was of them. But one of them, basically, their family office just reached out and they're like, I got like seven emails I should talk to you.

1:47:10

So I'm going to talk to you. Basically, that was the thing. Right. But it was also kind of cold outreach where like, like it was my cold outreach that I sent once or twice. And then a bunch of these people and they ended up replying to my first email. Like, OK, I'll just I'll reply to this person. So there's, you know, there's pure cold. But then also if you're here, that's why there's a lot of power being in San Francisco right now.

1:47:28

You can meet people that know people of one degree. And if they'll vouch for you enough, that's the social proof point. Like, it's like, yeah, if five of these people vouch for you, I'm going to get on the goddamn call. Right. Yeah. So if you're a founder in Croatia somewhere, you're not in San Francisco, you got it. You got to get here. I mean, there's a bunch you can do remotely.

1:47:45

But I think it's it's three years ago. It was easier because it was still post covid. People said remote type things. Yeah. Now it's so super in person. And now everything is here. So definitely get on the goddamn. Yeah. So either move here or spend time here. Spend time. Like, at least spend time. I don't think you necessarily have to move, move.

1:48:03

You have to like. So here's the thing I always tell people is like, how often do you see your closest friends in real life? Yeah. Actually, not that often. Maybe like some people might say a couple of times a year. Exactly. Let's say four times a year. Yeah. They're the closest friends. Yeah. Right. Right. So how often do you expect to see me? Who's your acquaintance that you might invest in me? Maybe less than that.

1:48:29

Probably. But maybe more than of the Peel with Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:48:46

Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:48:48

Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner

1:49:04

Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:49:22

Turner Nov Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner

1:49:32

Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:49:51

Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner

1:50:12

Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:50:18

Turner Novak, Turner Nov Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner

1:50:39

Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:50:48

Turner Novak, Turner Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:51:11

Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:51:21

Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner

1:51:37

Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak, Turner Novak,

1:51:51

Turner Nov And so a lot of it is outside of the actual job or art or task. Like being the best tennis player is not just about playing tennis, right? Being the best Formula One driver is not just about Formula One. Being the best, like all these different things is not about that. And so how can you do, how can you understand how the world works is probably what I find inspiring.

1:52:09

Yeah, you mentioned video games. What did you have or do you have a favorite video game? Oh, Mario Kart. Oh, Mario Kart. I will kick your ass on Mario Kart. I'll kick anyone's ass on Mario Kart. I'm not that good at Mario Kart. Yeah, whatever it is, even if you were. I was like, even if you were. I mean, I played a bunch of games, everything from like first person shooters, real time strategies, played a bunch of fighting games, like a lot.

1:52:31

But Mario Kart would sometimes still play. Still fun. I play like, I play very little video games now. Probably in a similar boat. The one I play the most is Minecraft with my kids. I've never played Minecraft in my life. Do you like Lego? Yes. It's kind of like playing Lego. I would assume, yeah. And if your kids like it, I mean, it's fun.

1:52:49

My kids are one in three, so they're like not there yet. Oh, okay. So they're very small. I do have a racing sim. I'll sometimes play that. Sometimes Gran Turismo, but like don't much have the time. Okay. But played a lot of video games. Played a lot of Super Mario as a kid, generally. Generally. Yeah, okay. A lot of Mario, yeah. My maybe claim to fame.

1:53:05

I don't even know if it's like a claim to fame. At one point, I was ranked second in the world in Gears of War. Oh, wow. I don't know if you ever played that one. It was a third person shooter. Yeah. That was for like a week. And then I played a lot of like Halo 3 specifically. I went to like semi -pro Halo 3 tournaments. And like, I don't know, I probably grew up thinking like I'm going to be like a pro gamer when I grow up.

1:53:26

But going to those tournaments, I realized I was probably in like the top 0 .1%, but not in the top 0 .1%. 1%, yeah. Yeah, so it was like, I don't think I can actually make a career out of that. The game that was the best at, and we can, I guess, we've digressed very, very far, is a fighting game called Soul Calibur. I've heard of that, yeah.

1:53:45

Soul Calibur was the very, if I remember correctly, the very first fighting game, which is actual 3D. It was actually you fought, and like you can, like I could, you could try to hit me with a sword and I can sidestep. I can go behind you and I can smack you behind the head. Yeah, because like Mortal Kombat and Street Fighter are 2D. 2D, but they're just 3D graphics, right? But these are actually like 3D.

1:54:04

And the thing is with this game, it was very hard to button mash. I don't know if people know the term. Basically, your baby kid could like smash a controller and beat you sometimes on Street Fighter, right? But Soul Calibur because of the 3D. You had to be really good, right? You had to be very, very, very good. And the whole thing was I had a next door neighbor, which is much older than me, and he had the console and he would kick my ass.

1:54:27

And he would like sit right back and like this and just like, kick my ass, right? I got so pissed off. I didn't have money, but I made a bit of money, and I bought a PlayStation, whatever it was, 2. And then I didn't go to his place for like 2, 3 weeks. I remember this vividly. I played so much, I didn't have any more thumbs. Like the thumb imprints were gone.

1:54:46

On the controller? On my thumb. On my thumb. Okay. All gone. From the, from the, from the, yeah, from like just choice to control our buttons. Yeah. I was just like, 2 weeks straight, right? Like it was like, it was gone. Like if I had to like put an ID on the, like, I wouldn't have. You could have like robbed a bank at that point. Like you could have been, you could have.

1:55:03

I had the other rest of the fingers, but not the thumbs. And so then I remember this, I come to his house, right? And he's like, oh, you want to play? I'm like, yeah. And so he's leaning back. He's playing, playing. And then he's like, he's like, what the fuck? He's like, you're like cheating. You're cheating. What are you doing? Right? Yeah.

1:55:18

I kick his ass. And then I was so good. I ended up going to a tournament. I won this tournament. And then people just wanted to play. And I was just like, then what happened is there's like a line of like 50 kids playing. And it's just like, whoever won stayed. And I was there for like, I don't know, an hour, two hours. I was just like, just kept doing that.

1:55:38

And then there was like this little kid watching for whatever. I'm like, you know what kid? I'm tired here. You can play instead of me. And so I gave it to them. And so I probably could have gone. That was the only thing I could theoretically have gone pro. But there wasn't much of a financial thing around that game that made it work. But that was the one.

1:55:54

It was because of personal spite. I just wanted to kick that person's ass. You just wanted to get really good. I just wanted to kick my neighbor's ass, which was like making fun of me while I was kicking my ass. And so I asked. Oh, man. That's awesome. Yeah. I think I'm like too old to be good at video games now. So my last thing we can say.

1:56:11

My youngest brother. So he was born in 91. My youngest brother. Okay. So was I. Yeah. There you go. He's probably 35. And so he's basically an on -screen commentator for Call of Duty. Oh, is he? Yeah. Is this for like MLG? No, this is for like these majors that are like stadiums of 60 ,000 people come to this thing. Okay. Who runs these tournaments? Is like Activision like runs these? No, I don't even know who runs this and he will kill me because I don't, but he sends me the pictures like he travels all over the world to do these things.

1:56:44

So he's like the play -by -play like commentator when someone's watching the stream? Yeah. Oh, cool. Yeah. And then or people like interview the, oh, how was the game? How did you feel right? Or whatever. Yeah. How did you feel when you got the snipe and you like did one shot at that guy? Yeah. You're like no scope to him. Yeah, exactly. And then he like, but there's like tournaments, like million dollar tournaments, whatever.

1:57:00

And I forgot the age, but it's like you retire at 18 or whatever. Yeah. You get, you're like considered old. Your reflexes don't work anymore. Graduated high school. Yeah. Yeah. Your reflexes are not good anymore. Yeah. Um, and so he goes to these things and that's why I know the age group's like, we're too old for this, right? Yeah. I'm very too old for this, but yeah.

1:57:17

Well, this has been a lot of fun. Thanks for, thanks for taking the time to do this. This has been awesome. Of course. And I'm sorry that we digressed so far from like technology, but yeah, I mean, it's where we're, the LLMs are all going to get updated to the extent of like, they're going to know that you were good. Like, I bet that's going to be like your fun fact, like in sales calls when like they populate the, Oh, did you want to play Soul Calibur? I have a thing.

1:57:36

They take the call. Yeah. Well, yeah, this has been awesome. Thanks for doing it. Thanks for having me. And thank you to this episode sponsors, Monaco, Flex, Clerk, and Numeral. If you liked this conversation, share it with someone who's not convinced we're headed for a CPU shortage, or maybe they just need a new computer for their agents.

1:57:53

Make sure to check out the back catalog of over 160 episodes with other AI founders like Eric Bernhardtson at Modal, Tony Polstock -Brown at Ingest, and Adit Abraham at Reducto. Tune in over the next few weeks for conversations with Emily Choi, the COO of Coinbase, Philip Johnston, who's building data centers in space at StarCloud, and JD Grace, co -founder of Town, on why most people are still not using AI properly.

1:58:18

If you don't want to miss any of these, subscribe to our newsletter, The Split, linked in the description to get each episode plus a transcript emailed to your inbox every week. Thanks again for listening. See you next time.

1:58:35

Bye.