Want to join the conversation? Connect with us on Tribe.
Find “Founders in Arms” on Apple Podcasts, Spotify, and YouTube.
Transcript of our conversation with Ian Tien:
[00:00.3]
Every government, is investing in sovereign AI, right? They understand it's the future and they're building platforms and they really want to be sort of independent of influence and be able to build things the right way. Is sort of a national, you know, part of national resilience, is just to have your own sort of like, you know, to control your supply chain.
[00:18.0]
Right? So how do I make sure I've got AI there? And then part of that for our business is like, okay, if I've got sovereign AI, I need sovereign collaboration. I need an environment that I can turn it multi user, that I can have it sort of do workflows and actually do actual human machine teaming. So whether it's on sort of intelligence or defense or, you know, government, sort of analysis, that's, that's kind of our markets.
[00:39.5]
We're flying around, you know, having these conversations and it's, it's super interesting how every different sort of nation thinks about their national resilience as a part of their national security.
[00:53.2]
Hey everyone. Welcome to Founders in Arms podcast with me, Immad Akan, co founder and CEO of Mercury. And I'm Raj Sooey, co founder of Lima and Tribe. And today we have Ian Ting, founder CEO of Mattermost and good friend of the podcast. Welcome, Ian.
[01:09.1]
Yeah, thank you for having me. So excited to have a conversation with y'. All. And so I was just on Ian's podcast a few weeks ago. Ian, tell us about your podcast. First, my podcast is called Tools for Tech Leaders and I basically invite folks that, you know, are leaders and I just basically try to learn from them and hit the record button.
[01:27.0]
So one thing I got from Raj is, you know what Raj does is he, he has the system where you have director level and above send out weekly updates. And you know, that becomes sort of like the backbone of just keeping everyone, you know, aligned and accountable and making sure everything's working.
[01:42.8]
So, that was amazing. And we actually had a clip of that and I actually do that now at my company director levels and above. And I actually have the clip to Raj on explaining why it's important and how it works. So that was just an easy button for tools like Tools protecting this. That's what it's about. It's an easy button to go and take something and have it run.
[02:00.7]
How many directors and above do you have? Like, how many reports do you get? Oh, gosh, I don't know, like, 15ish, probably. Like, this includes like VPs and directors, and these are like, couple of bullet points. Is that the idea? So I do it as going well, not going well.
[02:18.1]
What is the most important thing you need to do next week? And then, what's something that the whole company or all the directors love above, need to would. It'd be good for them to know. And then I actually have one which is like, what's the use of AI this week? And the questions will rotate and they'll change.
[02:35.1]
But that's how I do it. It's a little different from how Raj does it. Yeah. I feel like. Isn't there something similar about Elon Musk, where he does, like, what did you get done this week? I thought that was, like, a similar kind of idea that he uses at his companies. Yeah, it's really important. That's why I have Raj's clip there.
[02:50.3]
Right. Because, like, Raj has been doing this for many years and it's a little different. And I'm not doing exactly the way that Raj. We had every, leader, very clearly write down, updated metrics from their department and highlights, lowlights, any kind of important documents, they would attach to the email.
[03:13.0]
Basically, it's a way to actually manage the company by a written update, by weekly written update. The weekly cadence is very important. And the other interesting benefit of it is that it's a complete transparency to the whole leadership team over what's going on in every department.
[03:29.0]
The two things I like about it is like a. It creates this, like, transparency, and it makes it like everyone knows what everyone's up to, even if it's not their department. And I think the other thing that seems cool about it is you share with your own team what you're doing. And like it, like, it allows these leaders to, like, have, like, a way of communicating.
[03:47.9]
Like, this is what's most important how I think about things. So that. That is cool. Ian, maybe tell us a little bit about Mattermost. And I guess we can go into more CEO learnings after that. But I looked at the website. It looks cool, but it's also kind of hard to understand.
[04:05.0]
So maybe talk a little bit about you're not the customer in my. Yeah, I know I'm not the customer. So Mattermost has sort of evolved over the time. We started off as sort of like an open source Slack alternative and then moved, to sort of DevSecOps. And then what? The market's really kind of reacting to that and saying, hey, this concept of sovereign collaboration is really important.
[04:23.6]
So what's happening in the world Is you know, geopolitical change that's happening really rapidly and they're like okay, what if I need to communicate without any external dependency? All I've got is like a computer and a network. So mattermost is self hosted installs whether it's on on prem or it's in a private cloud or an enclave that's air gapped.
[04:40.7]
But is the experience still like a Slack like app experience? Is that the kind of end result but it's like hosted differently? Yeah, we have, we have five products right now and one of them is that messaging collaboration. We also have a workflow platform. We have audio and visual. Then we've got a sort of like KANB board product and then we have like agents or multi user, multi back end agents.
[05:00.8]
So it's sort of like a suite of which one is. We're most known for that sort of like chat op style, communication. So yeah it's it and it's all. And it comes from an open source platform. It's kind of like Get Lab. It's got the paid version and the open source version. It's kind of cool that you started with this open source Slack thing and you found this like market.
[05:21.5]
Did the market just pull you towards it? Or is this like what was your kind of discovery from like hey, this is because I think you started with a game company that built this open source messaging platform, right? Oh yeah. If you want to have the like you know, coming out of left field, sort of like the journey.
[05:36.9]
It's like we started at Y combinator with an HTML5 like game platform. Right. So that was in the early days when you're doing like okay, multi platform and we built it and it was really good but we found out no one wanted it. It's like all client side JavaScript, right. They're not going to pay us for that. So like okay, well I guess we got some prototype games on it and the games were making money.
[05:54.9]
So I guess we're a game studio and we ran that game student for a few years and we started because when you run this it's like if that game ever went down because we were essentially self funded, right. Like the company would lose revenue, like we would physically lose revenue. So we were actually using a different not Slack but another sort of like chat app to kind of coordinate everything and it got bought by a big company.
[06:16.3]
It started crashing and having bugs and errors and like we tried to leave and yeah was it didn't want to say it. Yeah, we have friends at Hip. There's some Waterloo folks that were involved with that that we really love. So we're not going to like, that's fine. Right? It's. Yes, that's the product, man. Hipchat really like squandered, that prize.
[06:34.4]
It was such a. I also used it. It was a great app. I had the same thing. I loved hiptast. I met the founders and it started crashing and we had to switch. Yeah. So this is kind of funny that Slack also came from a game company. So is it this about game companies that like, you really can't go down and you need to collaborate in real time that like leads to chat apps?
[06:52.9]
Is. Is that the throughput through way? Yeah, I think there was a time where yes. It's basically like the chat app is really like a community. It's really like who's in the room. Right. So there's a lot of things that come from games and online communities that translate really quick.
[07:09.3]
It's very intuitive. So people who build corporate apps are very different than people who build sort of like, you know, online entertainment and engagement. So if you notice, not only Slack as it used to be a game company, but Discord as well. Right. So Jason Citron, in the early days, like actually I was in grad school, I invited, you know, a bunch of gaming people and Jason Citron was one of the speakers and he had this, he was doing like a MOBA at the time or he's doing like his own online game.
[07:31.7]
And he actually pivoted sort of after, you know, after a little bit. So yeah, that's three of us all. All started with games. And then how did you get from open source, Slack alternative to like helping sovereign defense, you know, sovereign, what did you call it, like sovereign independence kind of thing?
[07:52.3]
Yeah, sovereign collaboration. It's really just talking to users. Right. And we really start off in DevSecOps and people were building things and we started having these customers. We're like, they don't have any developers in there. Why are they using us? Oh, it's like, oh, that's a corner case, right. Like focus on DevSecOps. And then a lot of the people using us, they wouldn't tell us what they actually do.
[08:09.5]
And it turns out that a lot of them are national security. Right. Because they need, So there's this concept of a skiff, right. Which is if I'm working on classified information, I have to have like no windows, it's a special room only information. Is there and we were. And their it because they don't have any networking, their IT there is very limited.
[08:26.3]
They're working with terrible tools honestly. So having something that is, you know, we're in golang and react and very sort of like modern, easy to use, standing, we just kind of stood up. We just found ourselves all over the place. Like people are just running mattermost as part of their core kit, their open source version and as they get to a certain scale they'll start to buy the enterprise version with advanced security and features and scale.
[08:45.8]
So that's kind of how it, how it kind of just went bottoms up into all these places that we never predicted. Wow, I've never heard of a PLG motion to this kind of customer. I guess open source was the real enabler there, right? Yeah, 100% and community being transparent saying this is what we're building, this is where we're going with the roadmap.
[09:06.1]
And then we got a lot of people engaged and we actually ended up winning a number of sort of government contracts to add capability. So there's SBIR programs for defense innovation. So we've won enough of those. That's public information. So that was a, a great path for us. What are useful metrics that are public about your company?
[09:24.4]
I don't know. I assume revenue isn't a number of employees or other things that people can understand the scale. Yeah. So we have 4,000 contributors to our open source platform, over 30,000 code contributions. We employ about 120 people and it's profitable and growing.
[09:41.2]
So that kind of lets you sort of give the shape of things. Nice. Congrats. When did you get profitable? Well we were profitable before we took VC money and then we took VC money and by definition you're not profitable. And about three years ago we went back to profitability.
[09:58.0]
What do you mean by definition you're not profitable. VC money? Yeah. So you take on VC money in order to burn it? Well, not necessarily. Immad doesn't agree with that. He likes to take VC money just to, just to keep it in the bag.
[10:14.1]
Keep it in the bag. That's what we should have done. Yeah. Makes the mercury KPIs go up. Higher interest. Yeah, that is totally what we should have done. So you took VC money on that? Tell us the story. So why did you take VC money and what did you do with it?
[10:31.1]
And then. Yeah. Do you regret it? It sounds like maybe you have mixed feelings. No, no. We shouldn't have spent it. Right. We should have just Taken it and spent less. I think, you know, just, I, I really love the, our board and our investors.
[10:47.0]
And so it was really, you know, being engaged for like over a year. We're growing, we're profitable, and it's like, you know, go to the next level when you, when you, for us, taking VC money, just let us attract a different level of talent. Right. And it's, it's a sort of like a signal.
[11:02.6]
And I think that was one of the benefits, also just having, you know, the resources of a VC fund and like, you know, executive recruiting and just people who've seen it and been there. So I very happy to do it. You know, at the same time, we had, we really appreciated yc. Like, YC was just, you know, so much such, an amazing community, so many people can go to for like, you know, input advice.
[11:23.1]
So we were, we're seeing that and just like doing the sort of, the more advanced version of, of that sort of investor help. But you feel like you shouldn't have spent it. We should have spent a lot less. We did silly things, right? Like there was a, there was one. We spent at one point on internal swag. Like $270,000 in a year.
[11:39.6]
Right. Just on internal swag. Right. And I was going and like a lot of stuff down in my parents house and I was just checking it out and it's like, I got these towels, I got like, Uggs. I've got like just ridiculous, you know, amounts of, of, of things. Yeah. So that's, that was a little overboard.
[11:57.0]
Amazing how that happens, right? When the money gets in, you're like, it's hard to resist. Yeah. And it's, yeah, it's so great just, you know, being profitable and having that, you know, thoughtful conversation on, on what's important. You can't do all the things. So it actually provides this wonderful, sense of focus.
[12:15.5]
Maybe they should just dribble out the VC money over like three years or something. They'll get this much a month. No, I mean, the incentive structure is VCS want you to spend it, right? Not in a nefarious way. They just want you to grow faster and spend as much of it as possible.
[12:32.7]
They'd rather you just maximize your chance of being a 10x return or 100x return. So, that's just all they care about. That's what the optimization function is, which isn't necessarily aligned from does this company survive or does, you know, does the founder get a great outcome?
[12:48.4]
But they also want you to like, be dependent on them to some degree. Right? Like, you know, so like, even if the burn doesn't quite result in like a 10x outcome, like, you can buy more ownership if you believe there will be a good outcome. I don't necessarily think that's how they think.
[13:03.7]
I don't think it's like a, like, oh, keep them hungry for some more capital. I don't, I don't necessarily think it's like that. I mean, there is like some incentive there, but I think that's, I really think most VCs are just like, hey, how does this become the biggest outcome possible? And that's just the only thing they care about. And to some extent they think money can accelerate that.
[13:21.0]
Did you really think money can't accelerate that yet? Like, you don't feel like more money gives big outcome, for your company? I think it totally can. But also we were not ready to deploy the capital because we didn't understand our customers and we didn't understand our business. And even now we're going through this, this sort of like, expansion of our teams because being in national security, not just US national, but international, is, it's, it requires like domain expertise and all these things that we really didn't have.
[13:48.2]
So we're putting money in, but we were, we weren't putting them in the right places. We kept putting money into developers and, and sort of bringing the wrong sorts of leads to the top of funnel. So I, think yes, capital can absolutely accelerate, but you need to have the right sort of domain expertise and experience, especially in our space, which is very different.
[14:06.0]
Like, we don't come from central casting. This isn't a, you know, there's not like 10 of us, right? There's like a very few, folks that are in like the space that we're in. Whatever you had to learn as a CEO in this space, like, what are things that like, you know, you've had to. Yeah, it's like enterprise sales, government kind of tech.
[14:23.4]
Like, what have you had to learn or unlearn as a CEO? Oh, gosh, like basically everything. My background is like, I got out, I did a whole bunch. I was at Waterloo, just like Raj had a bunch of, you know, co ops. I was at Trilogy, which is wonderful. And then I was at Microsoft for sort of five years. And then, you know, going to.
[14:40.0]
And then I ran a video game company. And then everything here, there's like, you know, it's national security, it's critical infrastructure, it's how to sell to the government, how to connect with open source communities. Like all that is new hadn't worked in open source before. Mattermost. Yeah, and it's, it's and it's actually you know, really wonderful to say like, okay, I'm rapidly building this understanding and skills like that's almost like addictive sometimes just to have that, you know, that really growth and expansion.
[15:05.5]
So I had to learn basically everything about the business. There was very little to, to take from my previous experiences. Can you give me some examples of things that were like unexpected or like these kind of big aha, moments for you when you were kind of you know, learning to sell into this space or thinking about these customers?
[15:22.0]
I think it's like we actually had this this offset recently talked about being sort of like bilingual or multilingual and I think people who speak multiple languages kind of like they can you know, you know, adapt to that, that world. Like my French is terrible, but I am Canadian. But now I to speak, you know, U.S. military and I have to talk about, you know, DOD.
[15:38.8]
I have to talk about, you know, command and control is C2, right. And then you've got like ATO authority to operate. You have you know, IL4 56 which is impact levels, which is like, you know, areas of sort of different levels of security requirements. And I'll say this is that I like sometimes we'll go out and we'll have like, you know, our team, but then two military people will get together and have you seen that meme where it's like the AIs like are talking to each other, they realize each other are AIs and they just start going their AI language, that's kind of like DOD, right?
[16:09.0]
Yeah. I went to D.C. for a trip and I'm like these people obsessed with acronyms. The level of acronyming is like next level. It's just so extreme. But it is a way, I guess it's more efficient if you know all the acronyms. Right. And there is like a common language.
[16:27.7]
I'm sure the DoD is like any big company, but the federal government is like the biggest company of all. So they probably have that level of acronym. Right? Yeah. There's this concept of multi domain and a domain can be originally land seed, air, but the domain can be like army.
[16:49.3]
There's this concept of you have to have these sort of ontologies like dictionaries in each domain. So US DoD is a domain. It also has dialects Right. Army, Navy, Air Force, different acronyms, different meanings. So you have to be able to move across. And there's even things like if you're in the army and you're targeting, right, you're targeting coordinates, they're spherical, right?
[17:09.4]
That makes sense. You're in Air Force Oval targeting coordinates. Right. So like even you know that that's an example of like being, working across domains. What's the best way to get up to speed on this stuff? Is it just, you just have to hire some people that know, know what they're talking about when they speak to these people or like, do you just kind of throw yourself into these meetings and figure out I think it's being in person in conferences like military conferences, like fca, Potomac Officers Club, ndia.
[17:40.5]
There's these organizations and you just, you're there and it's like, it's just non stop, like one day full of presentations, full of these acronyms and if you just kind of go multiple times, eventually absorb it. It's similar to like cyber. Like go to cyber is another one of our capabilities. So we go to like RSA and it's like all that cyber, you know, lingo.
[17:56.3]
It's like you have to absorb that as well. So I'm learning cyber. I've already known DevSecOps, then I had to learn cyber. I also had to learn like mission ops and DoD. So that's just like the multilingualness that our business requires. Yeah, that's funny, the same thing happened when I started Mercury because I was like, you know, I don't know anything about banking and fintech.
[18:15.1]
So I started talking to people and there's so many acronyms and for the first month I was like, I have no idea what anyone's talking about. But then, yeah, you just keep exposing yourself and every time you do it you just ask stupid questions of like, what does this mean? And then after the, their 50th meeting, you kind of know at least what the acronyms mean.
[18:35.5]
It's interesting, right, because entrepreneurs have to learn these multiple languages. Almost all of our careers have done multiple industries, right? So we've had to get deep into different industries and like, I guess the startup ecosystem has it too, right? MVP and Series A.
[18:51.3]
Yeah, yeah. All the funding terminology for sure. The investor in us, Mark Leslie, who is the founding CEO of Veritas and he was talking about like the life of a CEO and it's like the best job in the world. It's the Worst job in the world. It's the loneliest job in the world. And. But it's also, like, very addictive, to Raj's point.
[19:09.0]
You have to, like, you're solving these, like, you get the hardest problems the company, like, the ones that no one else can solve. Like, that's your job. Like, you just work on the hardest problems of the company and priority order until it's time to go home. And that's how you sort of describe the job to me. I don't know if that relates to, if that, if that resonates with y'.
[19:24.4]
All. Yeah, I try to phrase as like, either the, either the hardest problem or the biggest opportunity, because it isn't always downside. Right. Sometimes you're just like, this isn't a problem, but if we just crack this, we could be like 10x bigger. And it is extremely addictive, but also stressful.
[19:44.1]
I definitely relate to the loneliness thing. We actually have this, is a champagne problem to some extent. But we had this all hands thing. And it's weird because a thousand people work at Mercury, but everyone has their own team. But the CEO doesn't really have a team. You know, it's like even the exec team has their own team.
[20:02.1]
It is like the strange thing where, yeah, in theory, everyone's your team, but, like, you also, you know, with a large enough company, basically no one's your team. Which is like a hard, role to have for sure. So Immad, one thing I did at the latter stages of presto is I had an office of the CEO team.
[20:21.9]
So, like, there would be like, my personal assistant, like, personal executive assistant, but also a chief of staff. So that actually starts to feel a bit more like your own personal team. And they're executing on your initiatives personally. Right? Just the three of you together? Yeah, just like that. And you can have more.
[20:37.6]
You can build out that team. But yeah, having a good chief of staff and a good executive assistant, but also you can have a corp dev or whatever you want to bring under that. That actually is also useful, I think. One thing I did, and one thing I found about, like, being CEO for a while is like, it got less lonely over time.
[20:57.9]
Like, I got better at sharing stress and problems with, with my team. To be honest. Like, you know, they're bringing me a hard problem. I'll be like, what do you guys think? Like, you know, how do we solve this? Do you go and think about it? I'll think about it too. But you, it's not just my thing. Like we're all going to share this together, right? So yeah, I think CEOs that take on every problem as their own problem without like sharing it definitely have it way worse.
[21:21.7]
I've always shared it. And I also have two co founders. So I think having co founders also helps a lot in this. Yeah, co founders are the best. Like me and Corey Hulen that started mattermost. Like it's great because we're very like it's very Venn diagram. Like there's very little confusion about like is it a, is it a Corey working on it or Ian working on it and the stuff that's sort of in between.
[21:42.1]
Like sometimes we have to both work on something, but it's just working with a co founder is just like one of the, I think the best. You know, just having someone with you to build the company. Yeah, you're right. So the CEO doesn't have to be sort of just totally alone if you're, if you're founders. Well, I do think actually, you know, my first two companies had a lot of co founder issues and I think this point that you made about the Venn diagram is such an important point that I think is like probably the cause of most relationship problems in general.
[22:06.8]
Whether it's co founders or colleagues or whatever. It's like you don't have like well defined, like you own this, I own this. If it's your thing, it's your problem and I'm not going to interfere kind of thing. Having that like definition be well defined and everyone be happy with that definition, as in everyone feels happy in their lane makes such a huge difference.
[22:28.9]
Did you write that out or is just like kind of implicit? Oh no, it just was implicit. Just sort of like natural. We've known each other like 20 years, worked at Microsoft together and you know, and Microsoft like so like I'm semi technical, like I've got a dozen software patents and you know, have like the degrees and everything.
[22:44.4]
But like I was CEO and he was CTO and then you know, as we're building the company I'm talking to a friend of mine and I'm just, I'm complaining a little bit. It's like I gotta do the insurance and I have to get the payroll right and have to do, you know, these things. And then my friend is like, yes, but isn't that the job of the non technical co founder?
[23:03.1]
And you were like, that's me. Or you were like no, never. No, I think that that's the job of like that's the job of the early, early stages. You have the cto, you have the technical people, all they got to build. Right. And like the, the CEO or whoever it is is just doing everything else. This is like you days.
[23:18.7]
Right. But it's like but it's also interesting perspective what engineering thinks about like the business side. I mean the role of the CEO in general. The way I think about it is just to do everything else what's not covered in the company that needs to be done and like that I think over time as CEO, you have to be good at so many different areas or like decent at so many different areas.
[23:37.6]
Like one thing I was surprised about, like how, how much I had to learn like legal stuff, you know, you know, just to like, you know, operate a company for as long as I did and eventually go public. Like, like there's just so much legal stuff to learn. And we had like at some point seven or eight different outside lawyers, like outside law firms we were working with.
[23:54.7]
And I know Immad Paul, you learned a ton about, you know, regulatory stuff. And I think the tricky thing is the thing that you have to learn a lot changes every six months. Right. Like there was definitely a time where, you know, I had to understand legal stuff a lot. But now we have a great GC and I don't actually have to think that much about legal. I mean obviously like I need to know high level, but he needs to understand legal very well.
[24:12.3]
But I think that is probably the trickiest thing you have to know about a bunch of things. Have to sometimes go very deep in some things. But the thing that you're going very deep in changes every six months. Yeah. And I think you mentioned the point about the gc, but I think that's really important too is your executives are almost like mini co founders to some degree in some ways.
[24:33.2]
And then when you're picking executives, you have to be really kind of almost selecting for that because you're really in it together. So if I have a finance issue, like, you know, I'm going to be deep with my CFO on it and we have to be really like, you know, joined at the hip about it. Right. Like so, yeah, that's how I thought.
[24:49.4]
I have this like, I think somewhat contrarian view on executives, but I'd be interested to hear your take. I once had this conversation with a pretty well known company CEO and he, I was like, oh, you know, what do you do? And he had a much bigger company at the time, he was just like, oh yeah, I'm just recruiting execs all the time.
[25:05.7]
And I was like, oh, wow, how many execs do you have? He's like, oh yeah, I have like 10. But they leave every two years. So every two years he has to replace basically three execs. So I said every year he has to replace three execs because he's got this continuous in and out kind of process.
[25:22.0]
I was like, this sounds awful. To Raj's point, if they're supposed to be your co founders, but you're mini co founders or whatever, but you're replacing them every two years, how can you even build a co founder, like a resilient culture company? So I really went out of my way to not have that.
[25:39.4]
Like I try to hire execs that can either. Maybe they can't, they haven't done the scale, but like, I feel confident they can learn. But I just spent a really long time hiring execs. Like we're still hiring for one role that's been open for more than a year, but I'd much rather like spend way longer and get the right person and have that be a more long term thing.
[26:02.0]
And obviously we make mistakes still, or sometimes it doesn't work out. But at least that's my intention going in that these people are long term and ideally they'll be at the company for 10 years. And that's a very different mindset. 100% agree. It is so painful when you don't have that.
[26:17.3]
It's so painful when you're sort of on that search and trying to find the person and for our space. And then for a lot of people it's a unicorn you're looking for. But we're in the National Security DevSecOps. Open source is really hard to triangulate. So I'm totally with you. And then, you know what is a great exec?
[26:33.0]
I think there's a couple patterns that I'm seeing which is I always have like the Fords, like your experiences are done something like been in a big company, been an early stage company, had a tremendous success and had a catastrophic failure. Right. If you have kind of all those four, it's like, okay, great, you're not going to be like freaking out at any point.
[26:49.8]
You've kind of been there, done that. There's a great sort of like, there's a great sort of, you know, foundation. You know, I asked one of, one of our, one of our Advisors. I just asked, okay, like you know, you've done a lot, you've been a public company, you know, how do you find like the right exec? And he said had like one phrase, he had scar tissue, that was it.
[27:09.9]
We end up often doing this, although this isn't like a requirement for us, but we often hire an exec that like started at a company, you know, very early, not in the executive position, and just stayed there for like 10 years. And like that company became much more successful and they grew into an exec at their company.
[27:29.0]
Like I, almost half our exec team like fits that kind of model. Either they became an exec or they were like second in command. And I think it is probably like inadvertently it also like captures this kind of scar tissue idea. But it's also this like they've seen the scale, they've seen like, you know sometimes you have an exec they're just at Google or something like that, but they don't know how Google all the steps.
[27:52.1]
They just know the end result of a large marketing team or whatever it is, but they don't know what the steps are. And they try to replicate the end result but actually the steps matter. And I found those people who've experienced that 0 to 1 in their function are just like, I'm sure that it's like scar tissue combined with, they have this kind of first principle idea of how you get that kind of thing.
[28:16.7]
Yeah, I think getting the stage experience right is really important. Right. So they have the experience at the same stage company or near about like sometimes I would look for people with like adjacent stages like a little bit earlier and a little bit later.
[28:31.8]
Like if they have both of that, if they have earlier and later, it's like almost as good as someone who's like you know, been at the same stage. Because they've seen both. What's bad is if you only have one of them, right? If they only have early stage or if they only have late stage, you know like a Google or if they're only been an early stage company, it often won't fit in like a mid stage growth stage company.
[28:54.9]
And yeah so I, and you know actually you know over time like my, I got the relationships I built with my execs were some of the most rewarding like you know, kind of work, relationships I built. You know they're very collaborative relationships and in many cases now I've invested in their companies and you know, hiring their Their kids, you know, to work in my new companies.
[29:16.8]
And, and like, so it's, it's. It's been, you know, that's one thing I think I got better at over time is like, building really good relationships with execs and, and not having this antagonistic relationship, which a lot of people do actually. You know, Elon sometimes asks, like, with his execs, but other people also have this material mentality of replacing execs.
[29:36.9]
What was your key to building a strong relationship with execs? Just like, time spent or is there something more? I think what you hit the nail on the head is just filtering correctly in the beginning and just setting it up for a trusting relationship. So they know that I trust them and they also trust me.
[29:52.3]
And we feel like co founders a little bit of that function. Right. And in the interview process, that's what I would say to them. I would say I'm looking for a partner on this, you know, in this area. Right. And I'm looking, and that's the way I treat it. I'm not here to, like, you know, you know, bust, your balls or something. Like, I'm here to, like, make this work together.
[30:09.7]
Right. And setting, up that relationship in that way where they feel like they can trust you and you're not there to like, kind of like throw them under the bus in front of the board or do something like that. Like, they know that you got your back, you know, that you've got their back. I think it was really important. And that's something I didn't have that in the early stages of my career.
[30:27.4]
I would say I was in the mindset of like, you do this or you're gone. You know, like, that was a very different mentality. Let's pivot a little bit and talk about Ian more. Ian, you mentioned there's, you know, you've got this AI agents. Are you also thinking about the same way? Like, these are still, like, open source, like, Edge deployed, AI things.
[30:47.6]
And yeah, tell me about, like, that product and how you're thinking about it. Yeah, absolutely. I think all around the world, like, I'm traveling next, next six weeks, I'm traveling five of them, and I'm going all international, like, around the world. And what's happening is every government is investing in sovereign AI, right?
[31:06.8]
They understand that it's the future and they're building platforms and they really want to be sort of independent of influence and be able to build things the right way. As sort of a national, you know, part of national resilience is just to have your own sort of like you know, to control your supply chain. Right. So how do I make sure I've got AI there?
[31:23.6]
And then part of that for our business is like okay, if I've got sovereign AI, I need sovereign collaboration. I need an environment that I can turn it multi user that I can have it sort of do workflows and actually do actual human machine teaming. So whether it's on sort of intelligence or defense or you know, government, sort of analysis, that's kind of our markets.
[31:42.7]
We're flying around, you know, having these conversations and it's it's super interesting how every different sort of nation thinks about their national resilience as a part of their national security. And that's happening with sort of global supply chain right now. It's really been a little exacerbated.
[31:57.8]
But it's amazing to see the transformation in the last less than a year actually. So if you have a Chinese open source AI deployed, if you're not China, in your country, is that sovereign because it's open source or do they want to actually have their own foundation models?
[32:19.8]
It's I'll say it this way, is that like how do you, how do you penetration, how do you pen test an AI model? Right. So like the traditional ways to do security like that we would on an application are different. And one of the new ways to do it in governments is to actually do it like a human being, like a background check.
[32:37.7]
Where were you trained? Who do you talk to? Like where do you connect to? So I think that's the, that's the kind of scrutiny people have. So it's almost like if you take something that is, that is foreign trained, it's almost like taking a human being from that world. But you would say like hey, actually EU does want to have a locally trained model and most of the bigger kind of country blocs will want to have a local sovereign model in some way.
[33:04.1]
Yeah, there's a lot of variables. I can go into detail. There's going to be national models. So Minstrel, like in Europe and in France, right? That's their like everyone's got their champion cohere in Canada, right? Like us, you've got maybe llama op and like the governments are funding these things because I thought they're mostly private, right?
[33:21.1]
Or like is, is like France funding Minstrel. I actually don't know that but like it's, it's basically having closer relationships with those, with those AI companies that are sort of building the future. So. And it's still cross, right? Still like Minstrel is used internationally and we certainly see it outside of France and the eu.
[33:39.2]
My impression is that China is kind of winning the open source AI models race. Is that what you see as well? Is that not nuanced enough? Don't really know that much. We're really not because of our sort of, we don't really do PRC stuff very much.
[33:59.1]
We're a little bit more sort of European and US and sort of like US allied countries. And they would never work with the Chinese open source model, is that what you're saying? Well, I think that US would not I think us maybe for experimental purposes, people like parts like that.
[34:17.3]
And I think it really sort of depends on the different nations out there on what their policies are. But it's really just a matter of sort of, you know, how do you think about the lineage of what are you bringing into your environment in terms of like your platforms? And I'll say it higher, higher level.
[34:32.3]
It's about supply chain, right? And every open source library, right, has a supply chain and an open source model is just part of that supp supply chain. So how do you make sure you can secure like what are the, what are the governance law? What's the governance that you have around your supply chain, right, the, the proprietary and the open source component.
[34:51.3]
So a model is no different, right? So you have to go and look to see, you know, who's controlling it, you know, what are the and what are some externalities. So I'll just say this is that there are nation state adversaries to the US that you know, will target the open source supply chain and they will target upstream libraries, right, that are a little bit more fragile, maybe overlooked and then they'll inject vulnerabilities there so it gets picked up over time and sort of like less detectable.
[35:16.0]
So those are, those are real sort of like challenges that you have in the world today. What has been like the biggest attack like that where like an open source library has been targeted by a nation state actor. I don't know, that those are just sort of examples in the briefings that are, that are out there.
[35:31.7]
So I don't know if that's, that's public. So that that information is public that open source libraries can get targeted. And that is an issue for both open source and Closed source. I don't have very, I don't have specifics. I mean the ones, the 2023, like Microsoft master key thing, that's pretty public and that's pretty open.
[35:50.7]
Salt Typhoon in the US if you're familiar with those, those are all very open. PRC basically was able to, through social engineering, intercept a master key from one of the major sort of cloud providers. And that became a big issue in a lot of defense and critical infrastructure organizations.
[36:08.4]
So that's just an example of, of nation state actors. And you can just Google it, right? Salt Typhoon, and that was more recent on US networks around sort of telecom and PRC actions. So these things are pretty well documented, but maybe they don't make mainstream just because they're a little bit more technical.
[36:27.7]
Do you think these kind of major US foundation models should be run more like national security projects? Right. Like I feel like it's probably not that hard for nation state actors to have access to what's going on in OpenAI anthropic and things like that.
[36:45.3]
Do you think it should change and there should be kind of closer security kind of supervision of this kind of stuff? I think it's really important to have, so it's really important to have supply chain intelligence, understanding where things come from and then you can make the decision, right?
[37:00.3]
It might be okay to use what is okay to use and what is sort of. And where is the risk, right? So every supply chain is going to be different and AI is just another example of that supply chain. So it really depends on what does it have access to, you know, what is the risk level. But like if I think who can work at like Lockheed Martin is very different from who can work at OpenAI, right?
[37:22.5]
Like the, the rules around these things are pretty different. I'm just saying, like should those rules actually be like similar. They could be, it really depends on. So like I'll, I'll say this one, way I'll just kind of go in sort of cyber security world, right? Like what is what, how do you set your policy and the, and you can spend a ton on cyber, right?
[37:42.4]
And you know like what's the right level to spend on uncertain defense? And it really depends on your, your adversary, right? So think about adversarial risk management, right? What, what are the threats like? Just like you do a red red team or a pen test, right? How am I attacking your system and then how do I set up the appropriate defenses?
[37:58.3]
Because the other thing you don't want to do artificial, like, you know, inhibit, like slow yourself down too much. Right? So you've got regulations on AI. Different nations have different regulations in AI, right? And the more regulations the slower things go. And what is, what is the, what is the right sort of place to set that dial?
[38:16.8]
Is, is a question every sort of nation or national security organization needs to, needs to set. So it's it's. So there's going to be absolutely places where you want sovereign models, right. Where you know, 100% it's safe but they're potentially not as performant as the ones that are broadbase in general.
[38:33.3]
So you have to make a sort of like as always, you know, performance versus cost and risk trade off. You started, you were working on this way before defense tech got hot, right? Like defense tech is like a big thing now and like, but you were probably one of the pioneers. Do you get asked for advice by new startups a lot on this, working with DC and doing, you know, this type of area?
[38:53.2]
Oh yeah, there's lots of YC startups that are doing defense tech and I'm happy to you know, spend time with them and walk through like they're all kind of like on that journey and some of them a little further than others. But like, like defense is, there's like two places that are growing in budget in 2025. One is AI and one is defense. Right. NATO is going from 2% to 5% of GDP.
[39:11.0]
Right. That is, that is incre. Enormous amount of, of investment. And you know, people are looking for sovereignty. People are looking for so both sovereignty as well as, you know, what's that future of AI. So yeah it's, and, and I was funny, I was, I was actually talking to a bunch of like Stanford MBAs who are like ex military, like veterans.
[39:28.0]
And they're talking about like oh yeah, but isn't that thing where you know, whatever the MBAs pile into that, like that's next, that's the next sort of like crash. Right? So is it like crypto or like social mobile? But I think you know there's going to be a level of defense spending. What are the MBA piling into?
[39:43.0]
Are they piling into defense tech? Oh no. Sell now. Well, right now Palantir is What a Rule 95, Rule 94 company. They're like 48% year over year growth, 46% margin. Right. And people are like, you know, what is that?
[39:59.4]
You know, what does that look like as a, as a business? And there's a lot of YC folks that are ex Palantir. Right. So and there's also this other thing of just sort of like the calling. Right. Which is, you know, there's, if you look at the history of the world and conflict and people wanting to serve. Right. Defense tech is also very sort of mission oriented.
[40:15.3]
Like you're helping these people who are like absolutely dedicated to the mission. And you know, even though you're, you know, maybe not the kind of person who would be able to help, like on the very front lines, you can actually sort of do your part for national security and resilience. So I think that calling Defense tech you talk about double bottom line or triple bottom line.
[40:33.4]
Is Balenteer a competitor to you? Sorry, I think like everyone, I am confused about exactly what Balentier does, but it feels like they sound a little similar to some of the things you do. They're basically solving, they have a one product that does secure collaboration inside the Palantir, model.
[40:51.9]
So in some ways a little bit, but mostly they about really data and solving really hard problems. We're really about sovereign collaboration as a platform. So I don't really see a, ton of overlap, but would love to hear what, I don't know everything that Palantir does.
[41:07.1]
So apart from learning these acronyms, what else do you advise? Defense tech startups, what are things that you think a software mobile gaming entrepreneur doesn't know about defense tech? That they need to like, get in the head quickly?
[41:23.2]
Yeah, just like get, go, like just get out of the building. Right? Get out of your building. You got to go meet your customers in person. You have to hear their problems. I think there's so much like theoretical things, but you know, going to conferences, going to fca, ndia, Potomac, Officer Club, and they're not, they're not super expensive conferences to go into.
[41:39.8]
Right. And they're hungry, they're hungry for innovation, they're hungry for startups. You know, they're kind of locked into that, you know, time and materials world and like the innovation that can come out of like the Silicon Valleys and the, and the sort of just like the startup world is like, it's really powerful if you can get this stuff.
[41:55.8]
There used to be a vibe in Silicon Valley that defense tech is too hard because it will take two, three years just to make your first sale. Is that not true that it's just really slow? Has that changed recently or is that still the case and people are just making it work? It's like when it works it really works and it can be a slog if you're sort of outside versus inside.
[42:18.9]
So I'd say that, I'd say enterprise sales is not that different. If you don't know how to do enterprise sales, it's going to take forever. Right. Like MSAs and contracts and blah, blah and compliance. So I think it's the same. But I think that the. But the difference is it's just further apart. Right. Like the Venn diagram is sort of people that understand defense and understand good technology is just a little further apart than enterprise and startups.
[42:41.5]
But it's very doable. Sounds like selling to restaurant chains. They were pretty far apart there too. But probably not the same compliance issues. One thing, about Palantir. I had a funny story about them or an interesting story.
[42:57.7]
At presto time when we went public, we were going through this SPAC motion, right? And Palantir had this business where they were investing in de SPAC companies. They were actually putting dollars to work and they were providing their product in kind so that these SPAC companies can recognize their in kind product donation as like an investment on their pipe, which would help them.
[43:20.9]
You know, their pipe is like their amount you have to raise before you go public. So it actually increased the stock value of the company. And as part of that they were looking at our company and they made us an offer. I was looking at their product and my investor would be like, you should take this deal.
[43:36.3]
It's going to help the stock. As you know, it's, you know, Palantir is going to, you know, do this thing. I'm like, I looked at their product and their product was terrible. And I was like, there's no way we can use this product. You know, it's just like just like a big database, you know, that had no value to us, but I don't think would have any value to anyone in Silicon Valley.
[43:53.3]
You know, it was a very overly. That's a harsh review. No value to anyone in Silicon Valley. Yeah. How many million dollars was this in kind, database? It was 10 to 15 million or something. There was some cash involved too.
[44:11.0]
And then there was in kind. I, but, but that's how like Palantir was like making money doing the de spack. Just like how different it was. It was like pumping. Because what they would do is they would, they would count this as revenue to them so they would get an equity stake in the D spy company and they would get the revenue that they recognize you know on their books as showing their growth.
[44:29.2]
Right. So the whole thing was so shady. I was like, this is like this company is stupid like that. That was my conclusion coming out of this. Well, presumably I guess the non nefarious argument might be they were trying to basically pre sell your product and once you're using it properly you'll basically never get off them and eventually they'll actually get real revenue.
[44:51.0]
Right? Yeah, I think, look, they probably custom built their product for government which had no value to startups and they were forcing it down people's throats like trying to get some case studies of Silicon Valley companies using their product. My guess was that. Right. But anyways, it's very different from that. Kudos to them.
[45:07.7]
I mean they're like the hottest company around right now. Yeah. But you think this exact product that you thought was really lame and useless is useful to government, situations? I want to know. Actually that's what I was hoping you would tell me. I don't know what government buys, but it's not something I would buy.
[45:24.7]
But the government has different needs. Was there a lot of things like this where Wall street was playing these kind of weird, totally bad. It's all of us. I do worry a little bit about that with AI. AI is this weird circular game where Nvidia sells chips to the cloud people.
[45:42.1]
Cloud people are selling the chips as AI to startups. Startups are selling it. And then someone said this, I don't think it's necessarily true, but if enough of those people are making negative margins then they actually kind of compounds, where everyone's.
[45:59.1]
The amount of money in the system is growing. But like no one's making profit all throughout it. But I mean I think Nvidia is not making negative margins, but maybe some of the other kind of players are. Yeah, I mean this is 2021 so like, you know, this is a crazy time.
