High Agency

Mo Dhaliwal 00:00
This is High Agency at Web Summit. We're meeting founders, builders, and thinkers. I'm Mo Dhaliwal, and today I'm joined with Will Wilson, the CEO of Antithesis. We're live on the floor of Web Summit. Welcome, Will.

Will Wilson 00:15
Thank you so much. Thanks for having me.

Mo Dhaliwal 00:16
It's obviously a very busy three days, so thanks for taking the time to chat with me, it's good. So tell me a little bit about Antithesis. It's described as, I think you describe it as a platform for the impossible. You got some major appliance, like Palantir, Mongo, Ethereum. How did Antithesis come about?

Will Wilson 00:34
Yeah, so actually the roots of Antithesis come about in our previous start of FoundationDB, which we sold to Apple back in 2015. So that company was making a distributed database, which is a hard kind of software to get right. Because when you have a distributed database, anything going wrong on any one of those computers could cause some kind of whippling problem that makes everything break. And it's really hard to think ahead of time of all the possible things that could go wrong.  And so in order to get this piece of software right, we built a really sophisticated autonomous testing system, which is basically just a fancy way of saying a system that goes and looks for all the bugs in your software, including the unexpected ones, the unknown unknowns, before they like happen in the old world. And the cool thing about the system was it didn't just find the bugs, but when I found them, we were able to perfectly indeterministically reproduce them. And that was like a huge accelerator to our activity and made it so we could build a really cool product. So we got bought by Apple. I later left Apple. I went to Google. A bunch of my colleagues went to Facebook and Snowflake. We dispersed all through Silicon Valley. And we would occasionally get back together and chat. And one day we're all chatting. We're all at these different companies. We're like, why is it that nobody in Silicon Valley has this technology, which we had at this tiny startup? And we all saw teams that would really benefit from using this. And it just felt crazy that this didn't exist. And so we decided to come back together and make it real.

Mo Dhaliwal 02:09
Is it that it didn't exist because it's really hard to do? Yeah .

Will Wilson 02:13
It didn't exist because it's really hard to do. And because if you were trying to do it without technology like Antithesis, you would basically have to plan to do it from the very, very start of your company, right? It's like, in its conventional form, it's a very hard capability to tack on midway through a project. But like, hardly anybody starts hanging from scratch. And even if you're starting something from scratch, you're probably using dependencies, right? You're probably using a bunch of software written by other people. You didn't write that from scratch, no. And so, you know, that all made it very hard to adopt this kind of technique.  The magic of Antithesis is that because we've gone down to the very bottom of how computers work and we've written a hypervisor, which you can take random software and like move it into and then start testing it in this way, it means that it eliminates that barrier to adoption. Now, any team can do it so long as their software runs on Linux.

Mo Dhaliwal 03:08
Yeah, I mean, I like hearing stories where like the main thing you were building on the main problem was something else, but there was some feature, there was some tooling there that had a ton of value, if you pay attention to it.

Will Wilson 03:20
It's really funny, right? Because when we were doing that previous startup, we were like, oh, yeah, we're going to compete with Oracle. We're making a distributed database. But there's a million distributed databases now.  Everybody's made one, but nobody's made anything like Antithesis. And so it feels like, yeah, like you said, the really valuable thing that we invented was this side project along the way.

Mo Dhaliwal 03:40
And personally, I mean, uh, I think your story is kind of interesting. You've, uh, you know, described yourself as a mathematician, but what you wanted in the software, cause you thought it was easier.  And you've also said that computers are the enemy. Also seems a multiple, you know, antithetical. So yeah, what are you talking about?

Will Wilson 03:58
Right. So have you seen this wonderful, wonderful movie called Terminator? No. Okay.  So no, I'm like, I mean, those statements are probably like a little bit trolly, but I think what I'm getting at when I say the computers are the enemy is, you know, computers do so much right now in the world. They're running everything. They're running on airplanes. Like think about that when you fly, like they're running your hospital, they're running, you know, your bank. And unfortunately, they're still very stupid. And, you know, they are going to do exactly what you tell them to do, not what you meant for them to do. And if there's any situation that they run into, that's something that you didn't anticipate, they're gonna fail and break down in all kinds of gory and horrible ways. And as Gary was saying last night, it's not gonna say the addition of AI like makes this even higher stakes and also even more fraught like AI is like stupid in a whole new way.

Mo Dhaliwal 04:50
Yeah, we like exponential levels of idiocy that we're now capable of.

Will Wilson 04:54
Yeah, right. And so basically, I think people often don't reckon with this or like haven't really thought through what the implications are better.  And, you know, if we can help in some tiny way to make it so that the software that's running all of our lives and like, literally, like literally our lives sometimes is like a little bit less stupid, like, I think that would be

Mo Dhaliwal 05:14
that would be pretty cool. Yeah, absolutely. Okay. So it's not that you have any sort of animosity towards computers or the industry, but it's just being alive to what the weaknesses are.  Yes. So that you can fully realize a company that can actually accommodate them and mitigate them. Yes. Yes. That's right. Yeah. So the founding of antithesis, you know, it sounds like Silicon Valley, yet a number of old colleagues that were sitting at all these various other companies. And I don't know, you got somebody's house over pizza at a bar somewhere and you're like, wait a second, we're going to start this thing. But a ton of people from FoundationDB actually came across to antithesis. Yeah. Were they all the foundation or where were they?

Will Wilson 05:52
And a whole lot of them moved back to DC as well, moved across the country in many cases, which I think is pretty incredible actually. People left Silicon Valley, left places where they had settled to come and join our tiny company in the middle of nowhere. It's pretty neat.

Mo Dhaliwal 06:08
And is that where you're based now? You're in DC? Yeah, we're in DC. Not sure. And why Washington, DC as a...

Will Wilson 06:14
Well, basically, we've discovered that it's actually an amazing place to start a tech company, especially a really deep tech company that's trying to do something really hard, because there's a ton of really, really talented people there who work for the government or used to work for the government now, or, you know, work for contractors or, you know, Facebook and Google, they have engineering offices there. But the thing is, there's nothing very interesting happening at DC, from a tech point of view.  Like we're like, you should check out Fox News. Yeah, right. Exactly. So what that means is just like, there's all these really smart people sitting in and doing really boring tasks. And even like the Facebook and Google engineering offices, like, no offense to anybody who works there, but like the projects they get are the most born ones, those companies. And so it just makes it very easy for us to hire and very easy for us to retain talent. Like we basically have lost maybe one or two people who we like, we're sad left over the entire seven year history of the company. It's an incredible retention rate. Like, right. Like, show me any startup in the Bay Area that could do that. Like it's not just because we're a great place to work. It's also because like the market is really working to our favor.

Mo Dhaliwal 07:28
Yeah, I mean that that's an incredible retention rate and no, I don't think we could easily show you another company that was able to hold on to that number, that amount of talent and leadership over a seven-year period. But I should make a few to my next question, which was actually pretty fascinating about how Antithesis came up.  You guys were in stealth mode for five years, some really long time. Yeah, I mean like especially in what is, you know, sometimes perhaps too much, but in the startup technology hype machine, right, where the amount of frankly, storytelling, pitching, talking, networking required to actually move your company, advance things, recruit talent, even find investors, that's kind of baked in from day one for a lot of startups that say, you know, they're doing a lot of story coming, day one. So how are you able to exist in what is, you know, essentially silence go by yours and then come out?

Will Wilson 08:24
Yeah. Well, first of all, we got really lucky, right? So we were really lucky in that we had these previous successes. And even before FoundationDB, my co-founder, Dave, had another startup that was also successful called Visual Sciences. So what all that meant was that we had a lot of people who wanted to invest. And then we also had previous coworkers and colleagues from FoundationDB and from that company, Visual Sciences as well, who wanted to work with us again. And so that's two of the main reasons startups talk is to get investors and to get employees, and we kind of had those covered.  So then the third one is to get customers. But you can totally do sales while you're in stealth. It's just much harder. But actually, it being much harder is good for you when you're small. When you're a tiny company and you barely got a product, nobody's judging you based on what your current revenue is. And so having your sales machine learning to sell with no website and no market reputation, it's like how ancient Roman gladiators would train with wooden swords that were heavier than the real swords. So when they actually went to fight, it was easier, right? It's the same kind of thing. And we were able to get, as you said, some pretty cool customers, some good logos while we were totally secret. It's a completely doable thing.

Mo Dhaliwal 09:43
Yeah, but it does require like extreme levels of product market fit. You know, probably extreme levels of validation. Yeah.

Will Wilson 09:51
I think, I think that's, that's somewhat true. And I think, look, I think there were real costs to be in stealth for so long. Like I think if I had to do it all over again, I wouldn't have done it that way. I would have come out a self faster.  Like in four years. Yeah, maybe. Yeah, I think four years would be about right. But look, the other thing that was a little bit different about this company is the thing that we're making, there were no competitors. There was no, there was like nothing out there. And so it's not like we were watching somebody else gobbling up the market while we were in stealth. And more to the point, it's not just that we had no competitors. It's like nobody even thought that this was a thing you could do. And so by not being in stealth paradoxically, we would probably be like encouraging others to start competing companies and encouraging people to invest in other companies doing this. Yeah.

Mo Dhaliwal 10:40
There's some validation that comes when you're kind of creating a category and the industry starts seeing investment.

Will Wilson 10:44
Yeah, that's right, that's right. And so we're basically like, look, let's figure out how long we can, how long we can pull this off for. And you know, eventually it became too much.  And honestly, one of the things that was most interesting to me, I wrote a whole blog post about this, actually, about which we can put in the links or something. Was basically that there's all these costs to being in stealth that have nothing to do with sales and marketing and hiring, like a really, really big one was just internal alignment. When the company got to a certain size. I mean, a challenge for everybody when your company grows to a certain size is keeping everybody marching in the same direction. But a thing that I didn't realize was that being public and having a bunch of public marketing and messaging that's for your customers actually helps to align people internally as well, because they all read it, right? Like they're all looking at your marketing, like, you know, like, and so it's like, it doesn't matter.

Mo Dhaliwal 11:35
Double duty.

Will Wilson 11:54
You're saying who you are. And like, that's important for outside stakeholders, but it's also important for internal stakeholders. Yeah.

Mo Dhaliwal 12:00
Um, so that explained why you've never paid for optimizing up till now.

Will Wilson 12:06
We have we had to start a baby rabbit. I'm saying you got started. Yeah

Mo Dhaliwal 12:09
A little bit. All right. Because when I read that about your background, I was like, you know, again, as an owner of a little agency on a very upsetting.

Will Wilson 12:17
Well, we got we got really far with like, or maybe like content marketing, like we were very good in the early days at getting onto the front page of Hacker News. Like, you know, that kind of stuff built up a lot of buzz.

Mo Dhaliwal 12:30
And for deep tech and engineering talent, like you're right in the heart of it then.

Will Wilson 12:33
Yeah, that's right. Now, nowadays, we are trying to move beyond that sweet spot and get to like a wider circle of people.  And so, you know, part of that's coming to things like Web Summit, which are a little bit less like strictly deep tech. But part of that also is starting to pay for some ads and we're doing experiments with that.

Mo Dhaliwal 12:49
What do you think people get wrong about like doing deep technology development that you guys really got right?

Will Wilson 12:56
Oh, that's a, that's a great question. Um, well, the first thing I should say is I'm not convinced that we've got it right. Like we hadn't proven that we got it right.  Right. Like, you know, the company is going pretty well right now.

Mo Dhaliwal 13:09
If it's selling, everything's going good.

Will Wilson 13:12
Yeah, I don't know. I have high standards for myself. We've not taken over the world yet, so we don't know if it's the right decision.  I think that one of the keys to our success was just, and I credit my co-founder for basically all of this, a certain level of technological fearlessness. People had tried doing stuff like what we've done before and all failed in various ways. And I think the reason that we succeeded was that we just tried doing it in a way that was like 10 times crazier than anybody else. And when you explain to people how we're doing it, they're all kind of like, oh yeah, that's obviously the right way to do it, but you're crazy that you tried that. It's like, well, if it's obviously the right way to do it, why are we crazy? Because nobody does that. But you have to get over that reaction and be like, just because nobody would ever do this, just because it seems crazy, doesn't mean it's the wrong decision. And David is somebody who has a real fearlessness about doing things the way that he wants to do them. And I think that that permeates our technology approach as why we've been so successful there.

Mo Dhaliwal 14:36
That's incredible, and also refreshing to hear, because I think we go through kind of waves of change and waves of what's cool, frankly, like, you know, Web Summit 70 years ago, I'm sure was filled with blockchain companies sent also as part of the things, super refreshing to get somebody to get a pure engineering approach, but also brought a lot of creativity and humanity to it.  Are there other tech companies that you follow? Like, is there somebody out in the industry that's also doing either some deep tech or some sort of innovation with technology that you're super excited or?

Will Wilson 15:07
Uh, I will give a pitch to somebody who's actually here in Vancouver. Um, there's a tiny startup much smaller than us, uh, called Loophole Labs. And they are another company that has done something where when you hear it, they're like, no way that's impossible. You're not doing that. But then the more you learn about it, you're like, Oh my God, they're crazy, but it totally worked and it's totally brilliant.  So basically what they're doing, you can think of it as live VM migrations. Like, you know, you have a compute instance, it's running an EC2 and you wish that it was like running in GCP now. And it just now it does this. Wow. And like, but it doesn't just do it. It does it while preserving all of your TCP connections, preserving like all of your, like literally everything. So like nothing that's talking to that computer from the outside can even notice that it's now somewhere different. It's like, it's nuts. And it actually works and it's very impressive stuff.

Mo Dhaliwal 16:05
Antithesis, you know, seven years, five years of stills, and now the trajectory that you're on, if you were to maintain this trajectory and live out into the future, what does a fully realized version of antithesis look like? Cause I'm sensing some passion from you.  Your world domination is definitely an aspiration. So if you look out into the future, what does the fully realized company look like and as a result, what is the impact you've made in the industry? What is software development? What does technology look like?

Will Wilson 16:34
Yeah, so I think the best way to answer this question is to start by just describing what it was that we had a foundation to be because the product that we've built now is not quite even up back to like what we were able to do for ourselves there. So really? Well, yeah, but we're trying to do it for everybody for a single program, which is much harder. So the key thing that we had there was we basically had like a magic genie or an oracle. But you would write code and it would just tell you if the code was correct or not. It's kind of like what people actually think they're going to get from AI, except it's not what they're going to get from AI at all. Right? AI is just going to always tell you your thing is corrected and what's wrong. But this was a genie that actually told you the truth. And when you have that genie, everything about your life changes.  Suddenly when you're writing code, you can just like be creative and take risks and just fly because you have this like magical, incredible safety net. It's like it's like you're playing a computer game and you can save your game whenever you want and go back. And so if you die, it's like you don't care, right? Imagine if you had that for your life, that would be pretty nice. Well, we can't give you that, but we can give you that for your software project, like, and it just like completely changes the nature of what an individual can do and what a team can do together. And basically what we want to do is expand beyond what we're currently doing, which is fault tolerance testing in a relatively high latency way and get to this like always on always their magical companion, which is like, you write some code and it tells you if it's doing the right thing or not. And I believe this is like very achievable the next few years. And I think it's going to be pretty significant.

Mo Dhaliwal 18:24
Yeah, I mean, I started the conversation thinking that you guys are building, uh, auto testing technology, but what you're really building is an environment where developers, people creating software can actually do their work with an extreme level of confidence that they probably don't have today. And of course that would create lots of room to then be creative, innovate, uh, and perhaps try crazy things.

Will Wilson 18:46
That's right. Sometimes I, sometimes I'm on a sales call with somebody and they're like, why should I be talking to you? I already have tests. And I'm like, that's great. You don't need to talk to me.  Uh, but just answer one question for me. Uh, when you write some new code and your tests all pass, are you 100% confident when you deployed the crash? Like 10% and they're office, like, no, oh my God. Are you kidding? And I was like, right. So that's why you're talking to me is like, let's get you there. And they're like, wait, that's not possible. And I'm like, well, it's not going to happen overnight, but no, we can't actually get you there. We've been there. We have customers who are there right now. Like let's, let's do this.

Mo Dhaliwal 19:22
Are there any um sort of immediate milestones coming up Antithesis, any uh like, is there a next release, is there some next big push that you guys are doing, right?

Will Wilson 19:32
Yeah, we're going to actually be announcing a huge new feature towards the end of June. I can't tell you what that is right now.

Mo Dhaliwal 19:41
It's gonna be really cool. Took out some of that stealth DNA in here. Yay!

Will Wilson 19:43
So that's going to be neat. We are also, towards the end of the year, going to be basically expanding the types of software that our offering covers.  Right now we're mostly useful for sort of large distributed backend systems that are doing very stateful things. And we're trying to make this applicable for many, many work on software. And we're doing the first such big new push, probably the end of this year.

Mo Dhaliwal 20:13
Amazing. Uh, closing question, uh, advise to young tech start-up leaders, uh, in, in leadership entirely, uh, what advice would you give them?

Will Wilson 20:24
Ah, so I think the most important, so leadership is a lot like marketing, and but I'm not going to us in that, that the substance is way more important than the like branding and polish, right? Like the most important thing about marketing is not like seemingly cool.  It's actually having something really cool substance so that you can then talk about the most important thing about leadership is not like rhetoric or managing people in a particular way or, you know, trying to motivate them in some particular way. It's like actually caring about people. Like people aren't stupid. They can tell if you are only enough for yourself and they can tell if you actually care about their wellbeing and like they can figure that out very fast, like, and my company's too big is like not an excuse. You know, a cool fact I learned recently is Napoleon would, who had a very large army, like had this incredible memory for his soldiers and like microscopic details of their lives. And he would remember like some random artillery sergeant, like, Oh yeah, you got a seven year old daughter. Like, you know, like, you know, just like he would also obsess about tiny details, like whether they had the right size shoes, because obviously if your shoes don't fit and you're marching across the snowy Alps, like you get gangrene. Even if you don't, you have a horrible time. And so just like caring and knowing and, and really being clear that you care, I think is like so much of it. Wow. That's fascinating.

Mo Dhaliwal 21:53
And I think that's an aspect of Napoleon Bonaparte's leadership that most people probably weren't aware of.

Will Wilson 21:58
Right, well, I mean Napoleon was in a different way very uncaring right he like spent his soldiers lives like nothing just which I would not recommend it but but I think he did get one day, right.

Mo Dhaliwal 22:09
All right.  So we like Napoleon, but in that way. Yeah. All right. Well, thanks for your time. If you want to learn more about what you're doing and what's going on at Antithesis, where should they go?

Will Wilson 22:18
Our website, antithesis.com. There's a lot of good stuff there. We also post on X and on LinkedIn.

Mo Dhaliwal 22:24
So check us out there too. Amazing, all right. Thanks for your time. Appreciate you.

‍