Avery Pennarun (00:00:00):
I mean, the joke is in Silicon Valley, if you go asking for money, they'll give you advice. If you go asking for advice, they'll give you money. We actually went asking for advice. And most of the advice we got was like, hey, you're kind of ready for a seed investment. You should just raise some money. So we've been more than 100% doubling revenue every year since we started. Similarly with user count. And almost all of that can be traced to, like, word of mouth. I guess the most important thing is just. You actually have to listen when you talk to people about the problems they're trying to solve and what your product might be able to do. Especially in the early days, like the thing you currently built almost certainly does not solve the problem. Right. And the most important thing is, listen really carefully. They can describe their problem in a lot of detail and figure out how you can help them solve their problem better.

Previous Guests (00:00:47):
That's product market fit. Product market fit. Product market fit. I called it the product market fit question. Product market fit. Product market fit. Product market fit. Product market fit. I mean, the name of the show is product market fit.

Pablo Strugo (00:00:59):
Do you think the product market fit show has product market fit? Because if you do, then there's something you just have to do. You have to take out your phone. You have to leave the show five stars. It lets us reach more founders, and it lets us get better guests. Thank you. Avery, welcome to the show.

Avery Pennarun (00:01:14):
Hi, great to be here.

Pablo Strugo (00:01:15):
So you had a massive fundraiser last month. I mean, you know, 2020, 21, maybe 22, we're seeing insane fundraisers all the time. And now with AI, we're still seeing some crazy stuff. But it's not as common, right? You raised $160 million last month. And so, like, it means more. Is, I guess, how I would put it. So congrats on that. And then, you know, I guess, but you only started it like in 2019, right? Just six years ago. Maybe walk me through a little bit. What Tailscale is, you know, how it started in the first place.

Avery Pennarun (00:01:48):
Sure. So Tailscale as a product, the one-line description is it's a tool that lets you connect any device to any other device or any number of other devices. Regardless of whether they're behind firewalls or have static IP addresses or any sort of networking skills or anything like that. So the simplest way to get started is. You download it from an app store. You log in with, like, a Google account. And that, you can do it. For example, on your phone and your laptop. Right? And then if you log in with the same account on both of those, those two things are connected across time and space, even if they're no longer on the same physical network. And it's pretty neat. You can send photos directly from one to the other, but it doesn't do it by sending up to the cloud and back. It goes point-to-point direct in the Internet. And so it's fun to use for free. We have a generous free plan, but it's also people end up bringing it to work. To solve all the connectivity problems at work. And that's where our business model is. The way we got there was definitely not as simple as that. So we started off. We actually didn't know, me and our three co-founders, we didn't know what exactly we were going to build. The name came before anything else, Tailscale. It's the opposite of Internetscale. By which I mean, a lot of people at the time were tending to use really, really scalable stuff that's designed for serving a billion users, even to solve simple problems like, I have an internal dashboard I want to share with my 20-person team. And you end up, when you use these overcomplicated solutions, most of your work ends up being setting up and rolling out these overcomplicated solutions, and a tiny fraction of your work ends up being doing the actual thing you set out to do in the first place. And so Tailscale is like, hey, all of these billion-user problems have lots of people working on them. Weirdly, nobody's working on how to make simple things simple. So let's make simple things simple. And then we took that sort of starting point, and we started talking to our friends, contacts, businesses, and potential customers. It's, like, just. We threw it out there. It's like, what are you doing? That seems like it's more complicated than it should be.

Pablo Strugo (00:03:41):
And for context, you were. Where, at the time? Did you have another startup, a different job? And how did you end up here?

Avery Pennarun (00:03:48):
I left my job at Google about six months before that. Played video games for about six months.

Pablo Strugo (00:03:53):
Nice. Which game?

Avery Pennarun (00:03:55):
I played a lot of Grand Theft Auto V. I found it very relaxing.

Pablo Strugo (00:03:59):
I'm a FIFA guy myself, but I remember GTA is solid, yeah.

Avery Pennarun (00:04:02):
GTA V is amazing. I highly recommend it. Although I guess it's like 10 years old now. I heard GTA VI is coming out like next year.

Pablo Strugo (00:04:10):
Oh, wow.

Avery Pennarun (00:04:11):
So I'm going to have to buy a new computer to make it work, probably. Anyway, so yeah, I had just left Google. One of my co-founders, David Crawshaw, had also left Google a little bit before me. And then a third co-founder, David Carney, I had worked with at my first startup. He was actually in my class in university a long, long time ago. And he had just left his startup that he had been doing a few months before that. And it was ready for his next thing. So timing was good. But we're all in three different cities. It was Toronto, Montreal, and New York at the time. David Crawshaw subsequently relocated to Berkeley, California.

Pablo Strugo (00:04:44):
And you just kind of came together, and all of you were just. Right place, right time. Decided you wanted to build something and decided that this was an interesting space to kind of explore.

Avery Pennarun (00:04:53):
Yeah. I mean, I started with the name Tailscale. I'm like, somebody's got to do things that make stuff easier. And I guess the initial version was just me going around to my friends saying, like, "Hey, I'm going to start a startup doing this. What do you think?" And some people are like, "Uh-huh, uh-huh. Yeah, good for you. You do that." And the two Davids were like, "Oh, this sounds really interesting. Can I join you?" Right. And so once we had the three of us, that's when we went off searching for, like, what are we going to build?

Pablo Strugo (00:05:18):
Yeah. Dive deeper into that phase of it, like finding, you know, a lot of the startup stuff is around what you do when you have, after you have, like, the idea. And in your case, you had a high-level idea, but you didn't have the specifics of what that might truly look like. Dive deep into how you, you know, those first three weeks or months, whatever long it took once you had those co-founders, what sort of steps you took to figure out what the kind of first problem you might solve could be.

Avery Pennarun (00:05:44):
So I guess a lot of people want to start a startup based on, like, I think we should build X. And I guess that works for them. It's never been my way of doing it. And I've done a few startups in the past with varying levels of success. Tailscale is certainly the most successful one so far. I like to start with a problem or a group of people that have a problem. Like, OK, we are going to do something that works. Thoroughly solves this problem that I can tell people have, or I'm going to do something that makes life better for, like, this group of people trying to do things right. And when you do that, you have to maintain a little bit of flexibility in your head about, like, okay, how are we going to solve this problem? Right, it's not always obvious. And so in our case, we started with the problem of, like, stuff is too complicated. The whole computer industry is making things hard for no good reason. Like 20 years ago, we weren't doing that, or like it hadn't set in as much as it had. Like, you could have a two-person team building stuff and build, like, a full-on product, or you could have. My favorite example is like back in the 1990s, Microsoft Access was really popular. And I was in high school and doing some consulting work for a local computer store. And I went by places where, like, the front receptionist had built the inventory tracking system for their business 10 years ago. The receptionist was long gone, but people were just, like, poking at this inventory system.

Pablo Strugo (00:07:06):
What was Microsoft Access? I'm not familiar.

Speaker 02 (00:07:09):
Oh, yeah, I guess you see this. I'm showing my age, right? Microsoft Access is a visual database editor. Right. So it's if you can think of it sort of like Excel, except that instead of just building sheets, you like put together tables, and then you can put together little forms that are like drag and drop that like let people edit the forms and tie these tables together. It was, really, really elegant system in that super non-technical people can build real database apps. This was before the web days. The web basically killed access. Once upon a time, when we all used to go into offices, the internet was not that popular. If you wanted to do the database, it was not surprising to have to go into a machine at the office and use an app that you installed on your computer. Over time, that model went away. But like, weirdly, nobody built the Microsoft Access of the future, right? It still doesn't exist today. Everything is harder now. Like everything, if you want to build an app, you need a team of people. You need to, like, start a startup.

Pablo Strugo (00:08:06):
It's like everything is, like, built for scale. I guess that was your original point.

Avery Pennarun (00:08:09):
Everything's built for scale, right? It's so hard to build an app that now you have to amortize the cost over, like, a million users in order to make it worthwhile to build a thing.

Pablo Strugo (00:08:16):
Right.

Avery Pennarun (00:08:17):
Versus Microsoft built Access and Visual Basic around the same time. And then one receptionist in their spare time could build an entire inventory app for your business. That doesn't exist anymore because technology makes things too hard to do. Or we've decided culturally to do it the hard way.

Pablo Strugo (00:08:37):
Tangent. But does AI change this, like with Lovable and all this kind of stuff that's happening that lets people just build apps, you know, with language, basically? (context: Lovable AI is a Swedish startup that enables users to create full-stack web applications by simply describing their ideas in natural language, leveraging advanced AI to handle coding, deployment, and integrations like authentication and payment processing.)

Avery Pennarun (00:08:44):
I think the dream is yes. I think the reality is no. And so what I think we can get there. We can get to a world where that can be that good again. But the thing holding everybody back, and I guess this ties back to the Tailscale story, is like, well, what is the reason that it's hard to build stuff now, right? And the reason it's hard to build stuff is the internet, right? If you don't build something good and yet you publish it on the internet, somebody is going to break into it and ruin everything, right? It's like stomp. On your sandcastle. And there's no way to stop that happening without, like, building a big team that. Who's going to spend all their time building production systems and doing security and doing, like, intrusion detection, and what about backups? And blah, blah, blah, blah, blah. And tell, like, 90% of your work isn't building your app anymore. It's just boilerplate. So Tailscale says, you know, what if you could just build apps and not put them on the public internet? What if it was more like the world of the 1990s, where there was no internet and this Microsoft Access app, which had lots of features but was certainly not secure? But it didn't matter. It wasn't secure because only the employees of your business were allowed to touch it. What if you just created a virtual platform, private network, which is what VPN stands for, where you could host your own stuff and not have all these intruders come in. If you could do that, then, you know, there's no modern Microsoft Access, but LLMs can do that part, right? Like LLM, please throw me together a crappy app. And we know that there's nobody looking at security, and it's probably full of holes, but like, who cares? It does its job, right? Slap this thing into my private network. It doesn't matter that it's full of holes, right? So I can see that world coming true. The default world where people, like, ask an LLM to slap something together and then they publish it on the public internet, it doesn't

Pablo Strugo (00:10:24):
Further away.

Avery Pennarun (00:10:25):
Seem like a very good idea to me.  

Pablo Strugo (00:10:27):
Right? 

Avery Pennarun (00:10:28):
Right, like yes, you can do that. But you're going to get, like, your stuff broken into in the next 10 seconds, and that has happened multiple times, right? Where people are like, "Hey, I just started a cool startup that did this, and it turns out that you'd like, paste it, in a public key automatically from the LLM that belonged to somebody else for some tool. Because it just happened to know the key to access this database, and everything was working until the person is, like, "Hey, why is my key in your app?" Right? And shuts it down. Like crazy things like that that are just not acceptable in the real world.

Pablo Strugo (00:10:56):
So I took you on this tangent, but back to the storyline, I guess. So you know you want to do something small, and you were talking about kind of two different ways. You know, finding problems, either a set of people that have a problem or a specific problem that you can solve.

Avery Pennarun (00:11:09):
So the problem, the problem we started with, of course, is like, look, I want to do simple things. It's hard to do the simple things. And we started just like talking to various people we knew in various businesses and saying, like, what do you think about this problem? I bet you have this problem because everybody has this problem. Tell us about the problems you have. Right? And it rapidly turned out that. The most common. What I would call a category of problems they have or what I call connectivity and security. Connectivity is like, how do I get the people on my team to be able to even connect to the thing I built? Right? And security is, how do I get people who are not on my team to not be able to connect to the thing that I built? And if you're deploying a dashboard, people who have done this at a company. They know that, like, 90% of the work is somehow those two things. And then the remaining 10% is actually making the dashboard. And then we just, this just turned out over and over and over again, like authentication, logins. What happens if I forgot my password? Where do I put the login database? How do I open the firewall port? How do I get the DNS name? How do I get my IP address? How do I, like, where should I host it? All these things are not building the dashboard.

Pablo Strugo (00:12:08):
And is this like talking about a dashboard? For example, like, this is for people that want to build their own versus, let's say, use a Tableau or Clipfolio or whatever, or even for those people?

Avery Pennarun (00:12:16):
Well, even if you're using Tableau, then you have the problem of, like, OK, it's in a SaaS. So presumably they handle all the security stuff. How do I get the data into it? Right. Well, now it's hard. Right. So now I've got to build a whole internal system that imports data into Tableau or And I don't think Tableau can do this, but some vetters can do this. I can have Tableau call into my network and query my database directly. But now where's my database? Is my database on the public-facing internet? That's not a very good idea, right? So now I'm spending all my time securing where I put my database and how I secure it, right? So it's all the same problems. Like, how do I connect thing A to thing B so that I can do thing C?

Pablo Strugo (00:12:52):
You know, the challenge, as I think through it, for me, I'm like, it's such a broad problem. There's so many different kinds of instantiations of this problem. That so many things that people could want to build that they would want this kind of connectivity to. What's the next step after people tell you, Yeah, sure, I'd love to build, like, little simple apps, but it's too hard. I don't do it. What are you trying to kind of get out of those conversations?

Avery Pennarun (00:13:14):
I think we learned a lot about the structure of the problem by having those conversations. The one that finally put it all into focus was the company that turned into our first customer. It's a bank in Canada. It happened to know the CEO of this bank from, like, just previous rovings in the business world. And he, when I talked to him. He said, "Okay, Avery, I have a problem. We have a Windows client-server banking app. and it does not support two-factor authentication." And so my problem is this banking app is probably never going to support two-factor authentication because it's not a web app. It's a native Windows client app. It's really hard to tie that into modern 2FA systems. But I need to because otherwise, like, our core banking system is not protected. So my options are. Replace the core banking system, which will cost millions of dollars, or do something creative. Avery, do you have something creative? And I said, "Well, what you could do is you could move the servers for this banking software onto a separate network." Even if that separate network is technically like in your back office, right? Your front office and back office could be on separate networks. And then to get into that network, you have to use a VPN that ties into your identity system and requires you to do two-factor authentication. Then, just to connect to the server, you had the two-factor authentication so that the server itself doesn't have to do that stuff. And so, you know, when I propose this, I'm like, "Oh, so I will help you out. I'll just go find a VPN that meets these characteristics and plop that into your system." We'll move on. You know, I did you a favor. You pay me some money, and I'll go.

Pablo Strugo (00:14:40):
Oh, this wasn't like a Tailscale thing. This was almost like an on-the-side.

Avery Pennarun (00:14:43):
It started from a Tailscale conversation. But I'm like, well, this doesn't sound like a product I would build. It sounds like something I will just go find for you, and we'll buy it off the shelf and help you deploy, and off we go.

Pablo Strugo (00:14:53):
And in their case, this was an app that their employees would use or the public face?

Avery Pennarun (00:14:56):
Yeah, it was. So, yeah, it is a banking app for their employees.

Pablo Strugo (00:14:59):
Okay, yeah, internal. Makes sense.

Avery Pennarun (00:15:01):
Bank tellers and stuff, right? So every bank, if you go in there, they've got computers doing things. Same idea, but the system needed to be proctor authentication. So the problem is I went and I looked for this VPN that matched these characteristics. So it can talk to, like, Microsoft Azure login. It can do two-factor authentication. You can leave it on all the time because it has to be on all the time. You're happy to run it even when you're in the office, which is not like a normal VPN. Normally you'll use it when you're remote, but it needs to run in the office, not degrade performance. And also not interfere with traffic that's not to your banking app. And, like, how hard can this be? This is, like, basic networking in my head, right? It turned out there was nothing that works like that.

Pablo Strugo (00:15:37):
I mean, all the VPNs I know of are just to, like, get access to Netflix. Not Netflix anymore, but, you know, Disney Plus or. 

Speaker 02 (00:15:43):
Yeah, as a sidebar, there's two kinds of VPNs. There's what we call, people call it, it's a funny name, privacy VPNs or consumer VPNs, which is exactly for that. They route your traffic somewhere else so you can get access to Netflix in other countries or whatever. And then there's the original meaning of virtual private network, which is your stuff is on a private network, even when you're not physically there, right? Which is what corporations use. But in any case, neither of those tools existed to solve the problem for my banker friend. And so we ended up like, OK, well, I could build you one of these. There's this new thing called WireGuard. It just came out. It's a super reliable, super simple VPN. So it'll be perfectly fine to leave on all the time. We can run it only to your banking app. And then we just need to build a little key generator for it that ties into two-factor authentication. So we threw that together over a few days, and we built it for them. And they really loved it. It's like, hey, this solves their problem. The auditors are pleased. The security is really good. The user experience is really good.

Pablo Strugo (00:16:41):
You literally saved them millions of dollars.

Avery Pennarun (00:16:43):
Exactly.

Pablo Strugo (00:16:45):
Crazy ROI. Crazy ROI. In a couple days. Not bad.

Avery Pennarun (00:16:47):
Yeah, that's great. And that's when, as founders, we're like, hey, maybe we're onto something. And then a few months later, COVID hit. And what was fun was that the bank was able to just tell all their employees, like, Look, this desktop machine, our computer under your desk, take it home, set it up at home, and you'll have to do all your work from here for a few months. And it turned out a little longer than a few months, as we all know. Right. But what was neat is that Tailscale, the prototype product, was on these machines. So, because it was already a VPN behind the scenes to add the security feature. It turned out it didn't matter if they were in the office. So they just brought their computers home, and everything kept working exactly like it was.

Pablo Strugo (00:17:24):
That's wild.

Avery Pennarun (00:17:25):
And that was the other big. Is, like, whoa, guys, what did we build here? This is really neat.

Pablo Strugo (00:17:29):
I mean, you enabled remote work from home. Remote and still access, like, you know, very sensitive information. Without having to be behind a firewall or whatever, but without being, like, actually in the place.

Avery Pennarun (00:17:40):
Yeah, exactly. And so, yeah, so we realized we were kind of onto something. And so we started talking to people about, here's the thing that we built for this bank, which was still in the prototype stage, right? It was this was still 2019, or well, I guess by then 2020 COVID times.

Pablo Strugo (00:17:53):
Were you funded, by the way, or you were just doing this kind of bootstraps at that point?

Avery Pennarun (00:17:56):
By the time we hit COVID, we had raised a little seed round from Heavybit and Uncork.

Pablo Strugo (00:18:01):
How much did you raise?

Avery Pennarun (00:18:03):
That was $3 million USD. But that was basically on the basis of at that time it had rolled out to the bank. The bank had not yet sent all their employees home because COVID hadn't hit yet. But it seemed like we were onto something, and we were already talking to people about it. And we had adapted the same technology to be used for, like, a toy at home.

Pablo Strugo (00:18:23):
And by the way, maybe walking through this, it may be a stupid question, but like you talk about the two types of VPNs. The one to access, like the consumer one, is obvious. We all understand that. And the other type of VPN that companies use, what was its main use case? Why didn't it not support this use case? Like, what's different about what Tailscale is supporting than them?

Avery Pennarun (00:18:42):
So it's at this point. It's really hard to remember because all our brains are like post-COVID now. But if you remember a world before we all knew how to use Zoom, which was only like five years ago. Right. I mean, I knew how. I'd been doing video calls because I worked at Google and stuff. But, a lot of people, it was like they had to learn how to do video conferencing for the first time because they suddenly got sent home from work. Similarly, like VPNs at that time, five years ago, the distant past, right? Were really like a backup plan for when you're not in the office. And so you had it on your computer, but it didn't need to work all that well because you only used it sometimes. And even if it failed while you were out of the office, it's like, eh, not a big deal.

Pablo Strugo (00:19:22):
It's your fault for being out of the office. Get back to it.

Avery Pennarun (00:19:24):
Yeah, exactly. Like in a few days, I'll be back in the office. I'll just pick up where I left off. And so all of this stuff was like the quality expectations were low. The performance expectations were low. And when it didn't work, you just kind of dealt with it. Right. And so the user experience was just awful. And so when I tried to apply one of these things to, like, oh, this is going to be on all the time, even while you're sitting in the office, it's unacceptable. Right. Like people would demand to turn off the VPN unless it's absolutely needed right now. So making an app that you use eight hours a day in the office require this VPN made everything else intolerable. And that was just the way the world was. And then everybody had to go home for COVID. And as another sidebar, all of the VPN servers that these VPNs would connect to were suddenly overloaded because they provisioned them all for a maximum of 20% of people being out of the office at a time. When it's suddenly 100%, you hear current stories about these companies where they actually had to. Like, okay, your time to get to use the VPN is between 2 and 3 p.m. And your turn is going to be between 3 and 4 p.m., just to try to spread the load out because the system was overloaded. Right. Like all of this stuff. That's the way the world was five years ago.

Pablo Strugo (00:20:30):
So, but. Funny enough, you know, remote seems to be a huge tailwind, but this all started with 2FA. So did they? Was that a specific problem for that bank, or was that also this multi-factor authentication? Was that also related to the way VPNs worked?

Avery Pennarun (00:20:46):
So multi-factor authentication, I mean, it was really compelling for this bank at the time. And while VPNs have not evolved very much, ignoring again consumer VPNs, corporate VPNs have not really changed very much in, like, 25 years. Right? And so traditionally you had to deploy, like, special keys for your VPN that are completely separate from your login process. Because again, it's hard to believe, but like SSO and like single, you know, login with Google, login with Facebook, login with Apple, like didn't exist. 20 years ago. And so you had to do all kinds of weird stuff just to manage usernames and passwords. It was pretty uncommon for multiple apps to share the same login. You'd have to create usernames and passwords on every single website, which many people still do today. But you know, whenever you, by the way, whenever you have the chance to click the login with Google or login with Apple button, do that instead. It's so much more secure it'll blow your mind.

Pablo Strugo (00:21:36):
I do it anyways because it's easier, and I forget passwords.

Avery Pennarun (00:21:39):
It's easier.

Pablo Strugo (00:21:39):
It's good.

Avery Pennarun (00:21:39):
It's easier, and it's more secure. And people are really resistant because it's like too good to be true. It's like Google must be trying to screw me because I click login with Google. It's like, no, they're not. Just click the button, guys. Anyway, all this stuff didn't exist. And so VPNs that existed at the time required this entire IT provisioning process where every new computer you might want to run the VPN on, you had to get a key from IT. You had to get it signed from IT. There was a whole security, like, certificate authority thing. Like all of this was a mess. And so two-factor authentication, again, a relatively recent thing, didn't work into any of this. And to do the two-factor authentication, what are you going to do? You've got this machine key. The IT team's already deployed to your machine. How do you do two-factor with that? Like, what are you two-factoring against? It was a whole big problem, right? So a VPN that just fundamentally only uses your corporate SSO login and your corporate SSO login is set up to use two-factor authentication, was almost unheard of at the time.

Pablo Strugo (00:22:32):
So that gets you going, and then COVID happens. And you realize you're in a pretty interesting position because you can effectively power this work from home. What happens then? Do you start getting pull? Do you start going out to other enterprises or banks and trying to sell what you've built? Or is it ready for that?

Avery Pennarun (00:22:51):
Well, one of the realizations we as co-founders had is that we didn't want to be a pure enterprise software company just because of, like, that doesn't sound that fun, which, you know, is embarrassing maybe to say this. But just like, you know, I don't know. It's like we wanted to build something that was like cool networking technology and for people like us. And we're not. We were not a bank. We were a three-person startup.

Pablo Strugo (00:23:14):
Well, you have to, you know, this is, it's a funny point, actually. You do have to, I find. This is the thing about startups. It's not, I mean. It's, you can't just purely rely on emotion, but you also can't purely rely on, like, logic and just be purely analytical. You do have to find this Venn diagram between what the world needs and what you're well suited to build. So, like the, you know, it's kind of laughable, like, oh, we don't want to build that kind of company, but it actually is maybe more important than it might seem at the outset. Like when you're in it, you realize. You really want to. You need to want to do the thing you're doing because it's not going to be easy no matter what.

Avery Pennarun (00:23:44):
Exactly. And so we had that in mind. We had also written a blog post essentially about how this works because we had the banking customer. And we're just like, you know, I like to write blog posts and get your name out there and see what happens as part of searching for product market fit. Right? So we just described what our thing did. And then we put a little button on our website, like a waiting list. Like enter your email address to join the Tailscale waiting list. And somehow this blog post got to the front page of Hacker News. And like hundreds and hundreds of people like filled their email addresses into the waiting list. And so the secret of the waiting list was. There was not actually a waiting list. It just sent me an email every time somebody put their email address in this forum. And so I've got an email box now with like hundreds of. 

Pablo Strugo (00:24:23):
That's awesome.

Avery Pennarun (00:24:24):
And I just, I stayed up for, like, 48 hours straight just answering these emails. And the way to add an account to Tailscale at the time. We didn't have an account system or anything. I had to. I authorized you by, like, literally modifying the source code and adding you to a table, which we did over and over and over again. And then redeployed the server.

Pablo Strugo (00:24:43):
But you were at least ready to actually deploy it with them. I mean, it was messy and manual, but you did have a product for them.

Avery Pennarun (00:24:48):
Yeah. It was nominally a product. You can download the client. And it would talk to our server with the hard-coded user IDs in it. And yeah, it did some stuff. And like, you know, more than half the time it didn't work, right? But on the upside, as a CEO. I now had, like, an email personal relationship with each one of these people. Right? And, you know, they were Hacker News users. Nobody expected it to work right the first time. Right? Because there was a blog post about this experiment that we were doing. Right? And so they enjoyed debugging it with us. Right? And so we immediately started getting a lot of one-to-one relationships with, like, nerds who wanted to play with stuff.

Pablo Strugo (00:25:21):
This is all post-COVID?

Avery Pennarun (00:25:23):
This was around the time of COVID starting, basically. So this was before the bank actually brought all the devices home. We had sort of started this, like, I don't want to do too many more banks. Let's see what else we can do with this. Let's write a blog post to see what happens. And so all of this excitement was there, and we realized we were onto something. And that is when we ended up going down to California, and we started talking to investors about raising our seed round. And the lead investor in our seed round, Heavybit, were the ones who actually explained to us how this works. It's like, Avery, you can have the best of both worlds. You don't have to say no to banks, and you don't have to say no to people who want to play with it like you do. What you do is you build a product that is really fun to use for free. And also they can bring it to work if they really like it. And when they bring it to work, you have to have a process eventually for being able to sell it to executives and roll it out across the whole company like you did at the bank. But you don't have to worry about that part right now. You can build the part where they use it for fun, and then they can bring it to work and do small problems. You can worry about moving upmarket later.

Pablo Strugo (00:26:28):
This is a classic like PLG motion.

Avery Pennarun (00:26:32):
Right. And so, yeah, exactly. And I think, again, it's hard to remember, but like five or six years ago, the word PLG was not that common. It was only starting to catch up.

Pablo Strugo (00:26:40):
That's true.

Avery Pennarun (00:26:40):
We hadn't even started calling it that yet. And so what I called it at the time was a hybrid bottom-up top-down. And I think even today, I would say PLG doesn't really capture this hybrid nature of it. People imagine that PLG is always going to be like a free plan, self-serve, and then we're done, which for a lot of products is the case. For Tailscale, we have the extra step of, like, okay, once you start getting popular at this company or once people start thinking bigger, you do need to talk to executives and do a much more normal top-down process. It doesn't seem PLG, but it was initiated by PLG. And when you're doing this top-down sale, it's only for people who contacted you in the first place. So they're very warm leads. And there's somebody in the company who is what we call an internal champion. Like, they really want you to win, which is super different from, like, your sales team going out and calling people and trying to beg them for a meeting, right? But nevertheless, it's still a top-down sale.

Pablo Strugo (00:27:31):
I'm going to ask you for a small favor, a tiny little favor. In fact, it's not even now that I think about it. It's not even really a favor for me. I'm actually trying to help you do a favor for you. Just hit the follow button. You won't miss out on the next episode. You'll see everything that we release. If you don't want to listen to an episode, you just skip it, but at least you don't miss out. But I think PLG has evolved, at least in most cases that I know. If you're doing B2B PLG. Most people have realized that, you know, that's the best of both worlds. Like, why wouldn't

Pablo Strugo (00:28:01):
Exactly.

Pablo Strugo (00:28:02):
You add enterprise sales once the product is working. You like the usage and retention. You know, the driver of that to get an entire, I mean, you basically need it if the customers you're selling to are big enough, like you do need to go through their process and run that enterprise motion at the same time as you actually have, like, usage from bottom-up.

Avery Pennarun (00:28:24):
Exactly. And all of this was sort of evolving over the last few years again, right? It's all pretty new. I like the term PLS, product-led sales. 

Pablo Strugo (00:28:31):
Sure, yes.

Avery Pennarun (00:28:33):
For describing the sort of, like, advanced version. But yeah, modern PLG, a lot of people are following the same thing. It's still pretty rare in the security and network infrastructure space. Because very rarely do you get a good PLG motion working in, like, security for the simple reason that, like, no person wants security voluntarily. Most of the time, security is something that's imposed on you that slows down your work. And so you're not going to bring a security product to work most of the time.

Pablo Strugo (00:28:55):
Yeah, that's so true. In this case, it's because they're building apps. So they're like, I want to build this app. And then you need Tailscale to do it. Is that what's driving that PLG part of it?

Avery Pennarun (00:29:03):
Yeah. So the magic part of Tailscale is because we combine connectivity and security, which are not two things that normally go together. You use it because it makes connectivity super easy. Like, it's easier to build your dashboard because you had Tailscale. And coincidentally, it's also more secure, right? It goes back to the same thing, like login with Google. It's like, use it because it's easier. Why do I need to manage usernames and passwords? I'm just going to click this button. And then it turns out that it's actually way more secure than managing usernames and passwords.

Pablo Strugo (00:29:32):
What, by the way, are some other things, especially in those early days, that you saw people building with Tailscale?

Avery Pennarun (00:29:37):
It can be hard to answer that question. Like Tailscale is such a, it's like a low-level module you can use to do the thing that you were trying to do. How do you explain it? It's like, well, one of our customers, one of my favorite quotes, I asked them to describe what Tailscale does. And they said it's like, "Tailscale makes the Internet work the way you thought the Internet worked before you found out how the Internet worked."

Pablo Strugo (00:29:59):
That's good.

Avery Pennarun (00:30:01):
Right? And what that means is normally if I've got a phone and a laptop and I want them to be able to talk to each other. Right? Even if they're on the same Wi-Fi, it's actually kind of annoying to make that work. Right? What's the IP address of my laptop? I have to go look it up. Right? And if they're not on the same Wi-Fi, like, forget it. Right? And that's actually really weird because the original internet. The whole point of it was I can connect any device to any other device. Right. And yet the way evolution has happened with NATs and firewalls and insecurity and logins and passwords and, you know, connect with Google and OAuth and everything else is like, now. Almost impossible to connect one thing to one other thing. Right. You have to always go to the cloud and then back from the cloud, and you're paying rent every time you do that. So Tailscale just makes that go away. Now my laptop has a name. Right. It's in DNS automatically. My phone has a name. It's in DNS automatically. I can write a really simple web app that just like. Accept an API call, or let's me navigate to a web page. For my phone, and do whatever I want. And it'll work.

Avery Pennarun (00:30:57):
But it literally doesn't go up. Like, it just goes direct to. Device to device.

Avery Pennarun (00:31:02):
Yeah, that's the magical-like feature of this. It's partly WireGuard. It's partly some Tailscale magic sprinkled on top. But because of that, you can build anything you want. And it's hard to say you built it with Tailscale. Because maybe you didn't even have to think about Tailscale. You're just, like, expecting the network to work correctly. And then you build something under the assumption that the network worked correctly and the attackers can't get it. And therefore, you know, 

Pablo Strugo (00:31:22):
For simple-minded people like me, can you, you know? Bluetooth is obviously device-to-device, but this is at a higher rate. Like bigger, bigger stuff that you can move through, or what's the, what's, you know what I mean?

Avery Pennarun (00:31:34):
Like Bluetooth works if your two devices are like three feet from each other.

Pablo Strugo (00:31:38):
Yes, right.

Avery Pennarun (00:31:39):
It's not very fast, but it works in those conditions. You can also do, like, you know, if you've got Apple-only devices and they're right next to each other, you can use something like AirDrop, right?

Pablo Strugo (00:31:48):
Yes.

Avery Pennarun (00:31:48):
Which behind the scenes is like generating a fake Wi-Fi network, and then two things join the same Wi-Fi network and send a file and then disconnect. But Tailscale works even when they're not right next to each other. Right? That's kind of the magical thing is, like, you no longer have to care where the devices are or what they're connected to. It'll use your Wi-Fi. It'll use. It'll switch between Wi-Fi networks. It'll use your cellular connection. It'll use whatever in order to find the other device magically elsewhere on the Internet.

Pablo Strugo (00:32:11):
And it'll get there, but it'll never actually go through a cloud in between. Got it.

Avery Pennarun (00:32:14):
OK, exactly. So it'll go through the Internet, but it doesn't have to go to Amazon and back. And if the two things are side by side, it won't even go out to the internet. They'll actually find each other side by side also. So you get the best connections possible.

Avery Pennarun (00:32:26):
How did it go when you went down to California to raise that seed round? Was that relatively easy? Was there any friction? Did people just kind of get it?

Avery Pennarun (00:32:34):
I think, I mean, the joke is in Silicon Valley, if you go asking for money, they'll give you advice. If you go asking for advice, they'll give you money. We actually went asking for advice.

Pablo Strugo (00:32:44):
And you got money.

Avery Pennarun (00:32:44):
Most of the advice we got was like, "Hey, you're kind of ready for a seed investment. You should just raise some money." But yeah, I was looking for advice because we were torn between, like, hey, it looks like we have something that would be great for enterprise. I don't really want to do enterprise. or not enterprise only, but we also have this thing that's kind of cool for just developers to play with. That seems fun, but I have no idea how to monetize that, right? And Heavybit was the one who explained to us sort of this nascent PLG concept. And they said they were really excited about it because they'd never seen someone do this in the infrastructure security space. But conceptually, it was the same. And we already had. We had demonstrated that we had the ingredients of both because people loved it. And it worked at a bank to solve real problems that were worth money.

Pablo Strugo (00:33:23):
How many users, more or less, did you have at that point? Do you remember, like, 100 users or something like that?

Speaker 02 (00:33:27):
It was, yeah, it was probably in the hundreds-ish.

Pablo Strugo (00:33:31):
And how much did you end up raising?

Avery Pennarun (00:33:33):
We raised $3 million in the seed round.

Pablo Strugo (00:33:35):
Okay.

Avery Pennarun (00:33:35):
And I think our revenue was, like, you know, less than $100K. And it was from, like, one. I think one and a half customers, because they had gotten a second customer using something similar by that time.

Pablo Strugo (00:33:44):
But that, sorry, that $3 million, was that the same $3 million you mentioned before?

Avery Pennarun (00:33:47):
Yeah, that was the same.

Pablo Strugo (00:33:48):
Okay, I see. It was your first money. And what was the plan with that?

Avery Pennarun (00:33:52):
The plan with that was just like, hey, I think we're onto something. Let's make the product actually reliable and see what happens if it's actually reliable. It was reliable in the specific banking scenario because they had a really, really well-defined situation. And we just made sure to fix all the bugs that related to that one situation. But as soon as you got the hundreds of people trying it with whatever, more than half of those people were failing to actually make the thing work for them.

Pablo Strugo (00:34:15):
So it was really just kind of product development. You figured there was a clear market here. There was a clear market need. You just need the product to work instead of 50% of the time. Like closer to 100% of the time.

Avery Pennarun (00:34:25):
Yeah, exactly. So in the, putting it in the product market fit context, I would say we had like idea market fit at that point, right? It's like clearly people wanted this to work. And it just, you know, just mostly didn't.

Pablo Strugo (00:34:38):
And what about, I assume, how long did it take between there and you had the product that you were confident in?

Avery Pennarun (00:34:46):
Product is still not perfect. Six years later, I would still 

Pablo Strugo (00:34:49):
But confident, confident, not perfect but good confidence. Like, okay, this should work. Yeah. 

Avery Pennarun (00:34:54):
I'm a person with a lot of confidence. So even that is not quite the right. The right metric necessarily. 

Pablo Strugo (00:35:01):
What would be, what would be the right? You know. What I'm getting. What would be the right? 

Avery Pennarun (00:35:06):
So yeah, yeah. From the very beginning, Tailscale was something where if you understood where it works and where it doesn't work, you could deploy it successfully, right? The point we wanted to get past is the point where most people who just downloaded Tailscale and started to play with it would have a successful experience. That took until we hired our first two employees in i think February 2020 after our seed round and the first thing they did well the first things they did was they helped us open source the client software which we'd already always been intending to do and just hadn't gotten around to it and we also added this function called Derp which is like the fallback network so when you can't get a direct connection it would go through Derp and negotiate a connection like us through a slower path, which greatly increased the probability you're going to get something from like maybe 80% before because we could get a direct connection about 80% of the time and it just drops dead the other 20% to like basically 100% now because that 20% would be covered by the fallback path. And that changed the behavior from, like, some people just had a terrible experience every time they tried it to probably everyone had a pretty good experience every time they tried it. And that's where you could say, like, okay, if you have problem X, you should really try Tailscale and see if Tailscale solves that problem for you. So if that's the definition of confidence, then I guess maybe, maybe that was a good way of thinking about it, but it's just.

Pablo Strugo (00:36:24):
And that was when more or less.

Avery Pennarun (00:36:25):
That was, so we released Derp in April 2020. There were some bugs to fix. So I would say by, like, May or June 2020, things were really rolling in the sense of, like, if you tried it, you would be successful. And then we had the problem of, like, a lot of people who tried it. They're like, I have no idea what this thing does or what it's for, right? But that's a bit of marketing and documentation, positioning, etc.

Pablo Strugo (00:36:46):
How did you get people? Like, you had the Hacker News post, which I'm sure brought a bunch of people at the beginning. But how did you market this in those early days?

Avery Pennarun (00:36:55):
In those early days, and for a long time, we did almost no marketing. This was something where the people who got it just became rabidly excited about it. And there was, whole aspects of just like, look, we paid attention to the people who emailed us and actually made sure they got answers. From the very first day when I was manually signing people up, the founders were answering tech support problems and working really hard to make sure that they had tech support answers, making the product like, you know, the so-called time to wow. I don't know if you've heard that.  

Pablo Strugo (00:37:23):
Of course.

Avery Pennarun (00:37:24):
Before. But, like, make sure, you know, how many minutes is it from deciding you want to use Tailscale to, like, oh, hey, this thing actually works. Like get that down to the smallest number of minutes possible and then give people a way to share it with their friends. Right. And that was the number one way Tailscale spread. Is just through word of mouth. And then, yes, we kept on writing blog posts. We, for whatever reason, the early people at Tailscale are popular blog post writers that seem to like we would write something. It would often get to the front page of Hacker News. I still can't explain why that happens. It's really not a voting block, even though we're accused of that sometimes. It's just like, I don't know why they like our content so much, but they do. And that helps a lot because Hacker News is like a lot of people who have a lot of friends that, even if they don't, those friends don't read Hacker News. And so it spread pretty fast.

Pablo Strugo (00:38:09):
And your ICP was all like developers back then?

Avery Pennarun (00:38:13):
Yeah.

Pablo Strugo (00:38:14):
And they would just start using it, build something, share with their friends, and tell them, "Hey, you can use this thing." And then it was really all just word-of-mouth that was kind of taking care of the early growth.

Avery Pennarun (00:38:23):
Exactly. And so, and what was neat is some fraction of those people just started bringing it to work on their own. And it was not a high fraction, but it didn't really matter because the way Tailscale works, this is like, again, getting into the weeds a little bit, because Tailscale makes a direct connection between your devices. We don't have to pay for the bytes, right? It's kind of, like, obvious in retrospect, but because we don't have to pay for the bytes, all we're paying for is just a little service that helps your devices coordinate with each other. So we can offer Tailscale to a very large number of customers for free. Without it bankrupting us. And then as long as some percentage of those people end up bringing it to work eventually and wanting to pay us, we've got a good business.

Pablo Strugo (00:39:00):
And do you have visibility into that? Like people who bring it to work?

Avery Pennarun (00:39:03):
Well, some visibility because they usually. They login with, like, their Gmail account when they're using it at home. They'll log into their corporate account when they're using it at work. So we don't 100% see exactly which person brought it to work at what time. But you can kind of draw some estimated conclusions.

Pablo Strugo (00:39:17):
When did you start going back to the enterprise, like adding that enterprise motion again, kind of getting the second bank to do an enterprise-wide deal?

Avery Pennarun (00:39:28):
We still, even to this day, almost exclusively pursue, like, deals that are like incoming. So people who were, like, wanting to use Tailscale. Because they heard about Tailscale. It's kind of funny sometimes, like one bank that we talked to, the CTO sent us an email saying, like, "Hey, can I talk to your sales team about deploying Tailscale at my bank?" And we got on the call, and it's like, okay, how'd you hear about Tailscale? Common question we ask everybody. And the answer is like, my friend runs a network of Airbnbs. He uses Tailscale to manage that network. I was telling him about one of our networking problems. He said, "Oh, you should think about Tailscale for your bank." And so here we are. Right? And so it's a really interesting process how that works. But if people really love your product, that kind of thing can happen.

Pablo Strugo (00:40:08):
But it's incredible to get, like, inbound. Almost fully inbound enterprise deals is a wild place to be in.

Avery Pennarun (00:40:15):
It is definitely a wild place to be in. But that's, you know. That is the sort of magic of Tailscale, is we put a lot of work into that flow and making sure everybody has a really great experience. And as many people as possible are exposed to the fact, because we ran into this problem for a while. It's like, oh, is Tailscale a toy? Because I can use it at home. Is it really enterprise-ready? And it's like, look, our first customer was a bank.

Pablo Strugo (00:40:34):
Right.

Avery Pennarun (00:40:35):
But not everybody understands that. You just have to make sure that they're aware of all this enterprise-y stuff that's in there. That you don't necessarily need to play with it at home. We didn't start the enterprise motion explicitly. What happened was this inbound interest. And so for the longest time, it was just the founders. Would answer these enterprise-sounding inquiries and turn that into a sales call. And then we expanded to, like, hiring one or two salespeople whose job it was to answer the incoming inquiries and try to find the pattern and try to optimize that. And eventually, only at the beginning of 2023, four years into the life of the company, did we finally hire a head of sales.

Pablo Strugo (00:41:10):
How big were these? Like, what's the average deal size back then?

Avery Pennarun (00:41:15):
It's so hard to talk about the average deal size at Tailscale because you get people paying like companies all the way down to like $10 a month for two seats, all the way up to like tens of thousands of seats. And so, you can talk about, like, what's the most effective deal size to steer toward the sales team versus try to steer them to continuing to do self-serve. But even that doesn't totally answer the question because sometimes you want to steer them towards self-serve, but they really want to have a half-hour conversation just to understand that we're really humans before they spend a bunch of money. So it's, yeah, it's very hard to answer that question.

Pablo Strugo (00:41:51):
Even in that first year, second year. Like, you were getting $100 deals and $100,000 deals and, like, half a million dollar deals. It was just all over the place.

Avery Pennarun (00:41:58):
I would say the first year we didn't have any half-a-million-dollar deals. It was more in the tens of thousands of dollars for the most part, like smaller companies. Even today, we're only now getting into, like, the first million-dollar deals. Because we're intentionally, like, we start small. We're moving towards enterprise. We're scaling up over time. Right? But most of the deals are, I mean, this will always be the case because most of the deals will always be tiny, and smaller and smaller numbers of them will be larger. Kind of for obvious reasons, right? The power law.

Pablo Strugo (00:42:28):
And then, like, late 2020. Once the product you feel is working pretty well for most cases. What's the main KPI you're thinking about? Is it revenue? Is it usage, retention, like referrals? What's the thing that you're obsessing about?

Avery Pennarun (00:42:46):
I guess a combination of usage plus retention are the most important things. Like Tailscale, is a very high retention rate generally. People who use it almost always stick with it. I would like the retention rate to be even higher. Usually when we dig into why people didn't stick around. It's often like, oh, I, you know, I was trying to do X, and Tailscale doesn't do X. And we're like, wait a minute. Yes, it does. You didn't find the information about how Tailscale does X, right? So that's like our fault for not talking to you sooner or whatever, right? So we can get better at that. Sometimes those bugs, we need to get rid of the bugs so that people don't have a frustrating experience. There's always the long tail of, like, just networking is networking. So there's all kinds of weird things you can run into, right? And we're just trying to eliminate all of those. And then on top of that, if they have a good experience, if they're using the product, if they're retained, then it's just a matter of, like, okay, are you so excited about Tailscale that you tell your friends, right? And the more they tell their friends, we have this thing we call the word-of-mouth multiplier, right? The higher that multiplier is, the faster Tailscale can grow. And to do that, you just have, like, super high customer satisfaction, right? It's kind of like obvious.

Pablo Strugo (00:43:45):
What is that? Like, how do you calculate that? The word-of-mouth multiplier?

Avery Pennarun (00:43:49):
Let's see. This is something you can actually find websites about this. But roughly speaking, you want to know how many customers. How do you think of it? A good, I guess. A good way to do this is, like, how many people visit your website that are not already customers? That are coming in through organic sources as a fraction of how many people come to your website that are already customers. There's a few different ways to do it based on that. But if you're doing web advertising, it's kind of cheating. If you're doing SEO, that's a different path. But if they're just showing up, as often happens, most of the people visiting our website show up because they Googled the word Tailscale. Like, well, how does that happen?

Pablo Strugo (00:44:28):
Wow.

Avery Pennarun (00:44:29):
Right? It's because someone told them about Tailscale.

Pablo Strugo (00:44:31):
Were you doing any SEO or not much?

Avery Pennarun (00:44:33):
I mean, we were. I mean, you know, our blog posts count as SEO. Documentation counts as SEO. We weren't doing, we were not at the time certainly doing any kind of like consistent.

Pablo Strugo (00:44:42):
It wasn't like a strategy, like what long-tail words, you know, draft this and that.

Avery Pennarun (00:44:45):
It was one of those things we knew we should do. And we occasionally inserted a few words into our blog posts and stuff, but we weren't doing it very, like, very reliably. Yeah.

Pablo Strugo (00:44:54):
And how fast did things grow, like through, let's say, 2021?

Avery Pennarun (00:44:57):
I mean, pretty fast. So we've been more than 100% doubling revenue every year since we started. Similarly with user count, similarly with number of active nodes. And so it's growing really, really fast, and it's continuing to grow really, really fast. And almost all of that can be traced to, like, word-of-mouth.

Pablo Strugo (00:45:17):
How fast did you hit, like, a million in ARR, for example?

Avery Pennarun (00:45:20):
We hit a million in ARR during 2021, I believe.

Pablo Strugo (00:45:23):
So it's about a year after the product was fully fleshed out.

Avery Pennarun (00:45:26):
Yeah, we raised our Series B like not too long after hitting that milestone. So it was a relatively small number of millions in revenue at that time.

Pablo Strugo (00:45:33):
When did you raise your, what was your Series A?

Avery Pennarun (00:45:35):
Series A was in 2020. So it was actually a fairly fast fall. It was like September 2020 or so.

Pablo Strugo (00:45:40):
Oh, wow. Okay. It's like six months later.

Avery Pennarun (00:45:42):
Maybe nine months.

Pablo Strugo (00:45:43):
Nine months. Okay. Six to nine months. So Seed was like $3 million. Then nine months later, how much was the Series A?

Avery Pennarun (00:45:49):
That one we raised $12 billion.

Pablo Strugo (00:45:51):
Twelve? Okay.

Pablo Strugo (00:45:51):
And then you hit a few million ARR, and then the Series B came?

Avery Pennarun (00:45:54):
Yeah. And that was at the height of the market. And we had sort of been warned by our investors to expect a decline in the market. So the word was like, Avery, if you're planning to raise any money in the next two or three years, you should probably do it right now. And I took that advice seriously. It turned out to be really good advice.

Pablo Strugo (00:46:10):
But the business kept kind of going up. Into the right after. So it's not, I mean, a lot of people actually suffered. It doesn't seem like you did.

Avery Pennarun (00:46:16):
Yeah. I mean, investors have told me it's like, you know, sometimes their favorite segment is security products because security seems to be recession-proof. You know, cross your fingers, never know what kind of recession is going to get caused next. But people can't just, like, oops, I guess I don't need security right now and cut their costs.

Pablo Strugo (00:46:34):
When did you feel like, personally, when did you feel like you had found true product market fit?

Avery Pennarun (00:46:39):
True product market fit, I think. Like I said, you know, early on, we had what I would call idea-market fit. Product market fit for me is when people are showing up and just buying the thing almost without you having to do anything and demanding it. And I would say, 2021-ish time period was when we started seeing that. Our investors have since then started and made a distinction between, like, product market fit and maybe enterprise product market fit. It's like, okay, do you know how big are these? And again, it's not actually like a threshold between enterprises and non-enterprises. It's actually like going upmarket. And I, you know. We keep finding new product market fit with bigger and bigger customers, if that makes sense. Each scale size of customer you reach has different requirements of the product. And we're just like slowly working our way into those higher and higher levels. So over time. You know, I would say we didn't really get to enterprise product market fit until maybe late 2023-ish or mid-2024, at least for, like, really big deals.

Pablo Strugo (00:47:40):
Did you ever feel, this is the opposite question? Like, things actually just maybe wouldn't work out, like they were just going to fail?

Avery Pennarun (00:47:47):
Hmm. No, I don't. I think with Tailscale, we actually never really had that experience. I know it's funny to say, and I know almost every company. They wish they were that lucky. Right. But we just bumbled into the idea market fit was so strong. It's like all we need to do is just build this thing, and people love it. Right. So there's got to be something here. And it's just like if, you know, the curve just kept on going. At this point, the only question is, OK, well, how fast is it going to keep going up? Right. Are we going to saturate or something? What does that saturation look like? You know, do we need to keep adding? Layering on more kinds of sales and marketing and product features to make sure this curve keeps growing? But it's like it's already kind of a success. So the question is like. How much more success?

Pablo Strugo (00:48:27):
How big of a success. So last question, you've built a few startups. I'm sure you talk to early-stage founders all the time. If you had one piece of advice for early-stage founders, what might that be?

Avery Pennarun (00:48:39):
I guess, most important thing is just. You actually have to listen when you talk to people about startups. the problems they're trying to solve and what your product might be able to do because, especially in the early days. The thing you currently built almost certainly does not solve the problem, right? The most important thing is listen really carefully. They can describe their problem in a lot of detail and figure out how you can help them solve their problem better, right? And then you have to choose, like, okay. Does this person need anything like my product? No, well, that's too much of a distraction. I can't take them, or, oh, that's almost right. I should adapt my product to match them. And in the early days, you really have to adapt a lot. And you really just have to talk to people. And the connections to individuals is so much more valuable than, like, measuring a lot of stuff. Right? I love numbers. I love charts. I love data. But like when you're small, like anecdotes are absolutely the king. And the only way to get good anecdotes is like a real one-to-one conversation, either over email or something or, ideally, a call. Right. But it makes such a huge difference. Just like if you're building something for somebody, that means you've got product market fit for one person. Right. And that's way better than product market fit for zero. Right. And you just like go from one to two to four to eight to 16. And now you're off to the races.

Pablo Strugo (00:49:54):
Perfect. We'll stop it there. Avery, thanks so much for sharing your story with us. It's been great.

Avery Pennarun (00:49:58):
You're welcome. It's been fun.

Pablo Strugo (00:50:00):
You remember, like, the first person who told you about Bitcoin? The first person who told you about Uber? You want to be that person because being first is cool. So be a cool person and tell your founder friends. Send it to them on WhatsApp. Put it in a WhatsApp group. Put it on a Slack channel. Let people know about the show. Let people know about this episode. Don't let somebody else beat you to the punch and share it with your founder friends first. Remember what Ricky Bobby said. If you ain't first, you're last.