Podcast Transcript
Welcome to the RSAC Cyber at the Top podcast, where security leaders across industries share the strategies, hard lessons, and real experiences shaping modern cybersecurity and AI. Insights are grounded in community and built for every level of profession.
Hey, everybody. Welcome to the entrepreneur's journey, which is a part of our RSAC cyber podcast universe. We're trying something a little bit new now with founders and entrepreneurs and innovators, and I'm super excited to be able to welcome Varun Badhwar, who is the CEO of Endor Labs. We're gonna jump into what Endor Labs is all about and Varun's own journey. But first, I just really wanna welcome you to the show.
Jen, thanks so much for having me. I can't believe for all these years we've been in cyber and haven't got a chance to sit down and chat one on one. So I'm excited and really looking forward to this.
Yeah. I'm super motivated about it because it is kind of strange that we've never been able to to chat before because so much of what you've been working on really, like, sings to my heart in terms of all of the focus on, you know, how you innovate security. So I'm really excited to be able to to jump into that.
So we have a series of questions that we ask all of our, founders, and you've obviously been on our Innovation Sandbox stage, I think, three times, which may be a winner winner chicken dinner for, for founders in the ISB universe, so another great reason to spend time. But I've chosen a couple of our sort of what I call the top 20, and then we can just make it a little bit of a riff on the conversation. But I'm really, really excited to learn more about your journey. So let let me just start out and ask, were you always gonna be a founder?
You know, if you ask my parents, starting at the age of seven or eight, I was showing entrepreneurial traits. Their friends would come over. I'd be like, great. I'm gonna help you serve dinner and help with everything. And at the end of the night, I would give them a bill, you know, for service fees at home.
Did you get any money out of it?
I did. I would get a lot of pocket money out of it. Granted it was Indian rupees, I would get 20 rupees, which is, you know, 20¢. That was fine. That was a lot of money then. And then over the summer, I would... Every time I'd, like, buy a bicycle, I'd be trading my bicycles. I'd be just doing stuff constantly.
And and so yeah, and it's funny because if you think about my parents, my dad was in the Indian armed forces, you know, not a business family, but it just...
I don't know. Sometimes, I guess, people say you're born with certain traits, and that's how I was. And then even when I came into undergrad at USC, obviously, being, you know, Indian, you either...
You go computer science or doctor. I went down the computer science route. But by sophomore year, I was acing my computer science courses, but it was not something...
I wasn't passionate about being a coder or a software engineer. I always kind of try to gravitate towards where's the business side of technology. So I was doing more consulting like program management. Like, whatever got me to the business side of technology were the kind of gigs I was doing.
And then luckily for me, and this is back in 2005, like, wasn't taught in most universities, but USC had just started an elective course in ethical hacking, and I took that as an elective because I was trying to pursue other ways outside of software development in my career. And I was blown away with the fact that you could break into people's networks and systems and get paid to do it and not go to jail. And so that kinda took me down this security journey.
And my first gig out of college was in consulting at KPMG, and, you know, if I pleaded for them to put me into security and they did, and, you know, the rest is history.
Wow. I'm really interested in the, in in what your parents would say about you as you were trying to trying to charge their guests for, making part of the meal. But that's fantastic. So you're saying you weren't passionate about computer science, kind of the heart of coding, but became passionate with the security journey, which very much resonates with me. But let me ask you, what problem are you obsessed with right now?
You know, I think right now the problem is very evident, which is the entirety of software development is transforming far beyond anything we've seen in our lifetime.
And more people are able to produce more code faster than ever before, Jen. Now my mom and dad can probably think they can write applications and software today. Scary thought. But it...
It's true. Right? And at the same time, you have more vulnerabilities and security issues being uncovered, not less, with AI generated code. And we're all obsessed with, look.
These frontier models are so incredible, and they are, at discovering security problems. Right? The last twenty years of software problems, you've been talking about this, that security is fundamentally a software quality problem, and obviously, all of these issues are bubbling up now with these models. The thing we're not spending enough time on, Jen, right now is talking about, well, great.
These models are great at adversarial testing, offensive security, finding bugs, but right now, discovery is outpacing remediation and fixing 16 acts.
And I go back to the root of this, great, if these models are so incredible, are they actually producing far more secure code than we ever have before? And unfortunately, I hate to report the bad news, but even with mythos class models, 70% of the lines of code being produced, and we do this study with Carnegie Mellon in Columbia, so it's like a very neutral benchmark. 70% of the lines of code being introduced with AI agents are introducing security issues. So on one end, we're already reeling with this security debt for the last twenty years of software, but we would have hoped that we wouldn't have this problem moving forward and AI is so powerful, it would just do things more securely, but that's not...
We're not even talking about that. We're all talking about what it's finding and how much it can attack systems. The defensive side of, AI is is it actually doing things better for us is not the case today. I would argue...
I've been talking to a lot of customers. Most customers, bug bounty programs, are paying two to four times more today than they were six months ago.
And so, you know, we have this existential problem. We're writing code faster than ever before. We're finding vulnerabilities faster than ever before, but we haven't found the secret sauce of how are we just gonna develop things correctly from the from the onset so we don't accumulate this debt in the past. That is a problem I'm obsessed with.
That is a problem that close to 200 people at Endor Labs are personally obsessed with and is our mission is how do we help companies develop fast and develop securely? Because historically, it's been a trade off. I can either move fast, and it's a series of compromises on security, or I'm secure and really slow and frustrating to engineering teams. We've gotta be able to, and the technology today allows us to do both.
We just need to kind of insert the right capabilities in the right workflows and design them for an agentic first development future, and that's what we're doing.
Fantastic. And and, you know, I spend a ton of time trying to dig into all of this, the whole conundrum of speed to market and features and convenience and driving down cost prioritized over security. And, you know, as you said, that's why at the end of the day, it's more of a software quality issue. So I wanna hear more about Endor Labs, and I wanna ask you what you're most excited about. But why don't you give us the whole Endor Labs? What does that mean? What was the... What's behind the name?
Give us that that story.
Yeah. Certainly. Look. Maybe just even pre Endor Labs. I've been building cyber companies for the last sixteen years.
Endor is my third. I was very happily employed after Palo Alto Networks bought my last company called RedLock, which was basically the creator and defining moment of cloud security posture management. Sold that to Palo Alto Networks in 2018. I was running a fastest growing business at Palo Alto Networks called Prisma Cloud.
Life was great. Right? COVID happened. I had twins in 2020.
Everything was busy and hands were full, but I... I've just always somehow run towards problems, not away from problems. And so 2010, my first company was the advent of enterprise SaaS how would customers control their data when they're using enterprise SaaS. 2015 was obviously moved from data center to cloud. In the middle of twenty twenty one, you might recall, the SolarWinds incident happened.
And, obviously, it was the first time we as an industry really started speaking about at the highest level software supply chains. And what does that mean?
And the little known fact is my team, within a thirty day period, there was SolarWinds, but also I had a 450 engine... Person engineering team was hit by another smaller supply chain incident with a company called CodeCov that was also a public incident that happened.
And I was like, what is going on? What is all of the supply chain stuff? Like, we have all of these controls at an enterprise level, but you're telling me our developers can go pick anything off the Internet and start using it, and we have no idea about it?
And, okay, then how are we analyzing and evaluating in the, you know, the application security testing tools we were using at the time were completely, broken. Like, they were compliance checkboxes. We ran some scans. They generated hundreds of thousands of findings a quarter. Nobody wanted to look at them because it was a waste of time.
So by August, September, I'd kind of formed in my head. I'd let Nikesh know, hey. Like, my entrepreneurial bug is back. I'm gonna go back to building.
Give him a few months of that. And then came the time where, you know, the VCs were ready with term sheets. I was blessed. I had a good track record.
It didn't matter what I was gonna build. The term sheets were there. Needed a company registered over the weekend so I could have them fill in the name of the term sheet. Well, as it turns out, pre ChatGPT days, it was hard to find a company name and a domain name that you could actually afford and want to use.
And so we went through this for several hours, and I was just, like, annoyed at this point. Needed something done. We were watching Star Wars on TV. My brother-in-law is like, well, let's just, for kicks, go start messing around with Star Wars related things.
And, yeah, we ended up finding Endor, and and I certainly wanted it to be kind of more of a labs kinda concept, like it was technical, the domain was complicated. We were gonna solve technical problems that had never been solved before, and so we kinda want...
Thought it was appropriate to have labs. So that's kinda how the name started. I did not expect five years later to still have the name, but I will tell you, Jen, it...
For people that love Star Wars and a lot of techies do, it works really well in our favor. And we do lightsaber giveaways and lightsaber training with Hollywood stuntmen and stuff, so it's great.
I have to tell you, I'm a Star Wars fan. And that was another reason I was excited to have this convo. Now, I'm a... I don't... Do you watch Andor?
I I haven't watched Andor and I'm embarrassed to say that I need to.
So this is my gift to you. You must watch Andor. I think it's like the best series. It's sort of the pre, you know, what actually started the original Star Wars. So I wanna pause on something and go back to something you said that I I wanna make sure nobody misses because I'm not sure I have this right, but it sounded like in 2021, after you had just had two twins, you decided to start a new company.
Yeah. So I had twins in May 2020. They were on their basically first birthday or the day after their first birthday.
05/29/2021 is when I I think told my wife and said, hey, honey. You're not gonna like this, but I think I'm ready for my third baby.
And it was not an easy conversation.
Sometimes when she's angry, even now, it's brought up.
But, look, that... That's just me. I I think she knew who she married, and I do the best I can to kind of balance being a good father, good husband, and a great founder and entrepreneur. It's hard, but what it's kind of taught me is just manic prioritization. I don't do things for the sake of doing things. It's gotta be impactful.
You know, every trip I take away has to count on why I'm gonna spend, you know, days on an airplane. It has to be impactful. It has to be high value.
But, yeah, like, I did it. I was crazy enough, and my wife was amazing enough and is amazing enough to continue to support that.
Yeah. I mean, that's... That is somewhat crazy. One of the decisions I like to ask founders is what decision looked crazy at the time but turned out to be right? So I think you've just answered that one, but it sounds like this is almost hardwired into your DNA to be able to take that risk and to go after something. So that's fantastic.
I guess in terms of what you're working on right now, what is the most exciting part of it?
You know, as I kind of said earlier, I like to run towards problems, not away from problems. And I think, Jen, the biggest problem we have right now is also our biggest opportunity, which is AI is here. It's creating havoc. You know, we're gonna have a very painful next few years of dealing with the legacy of twenty, thirty years of software that is riddled with vulnerabilities.
I think the problem everybody has is the influx. You know, we've been talking about noise in security for a long time. But in the last six months, I mean, that nozzle has turned to max, and everybody's getting hit harder than they ever have on all kinds of bug...
Bugs in their software being reported, new software being created at the same time that they have no controls over an understanding of. And I think fundamentally, there's two main problems. How do I know within minutes if there's a problem that I need to lose sleep over and drop everything else and address? And how am I gonna go turn this equation of discovery outpacing fixing by 16 x or the fact that our mean time to remediate software vulnerabilities is closer to six months when the mean time to exploit is down to ten hours is how are we gonna create the systems and processes and the technology thereby to support accelerated remediation of the last twenty years of accumulated security debt?
But more importantly, I always tell people, while you're fighting that fight, you need to start today on making sure all of this newly created software with AI, with humans, obviously, mostly AI, is secure by default. And I kind of always tell people the best way to address remediation is vulnerability eradication upfront. Like, don't allow classes of vulnerabilities and bugs to even enter your system. And I believe now the technology and workflows allow us to do that.
Now is it the same frontier model that writes your code that should also review your code? Certainly, there's...
You know, we could debate that. I I don't think that's a good idea just like with cloud. And cloud security, you needed kind of an independent fabric for cloud security because you would bring in different clouds, different services.
You need a consistent independent verification layer. You need the same for software development. Right? A year and a half ago, Copilot was the hottest thing.
A year ago, Cursor. Six months ago, Claude. Today, it's Devon. Like, these models and these harnesses for development will keep changing.
The security controls need to kinda be fundamentally plugged in to be able to support all of those platforms and changes. And maybe...
I I don't know if you'll agree or not on this one, but I fundamentally believe cybersecurity is a deterministic problem to solve. And I feel like having a 100% probabilistic systems to address that just doesn't fly with how regulation is, how auditability, verifiability, repeatability needs to work. For example, if I run a security scan on my code base on a Monday with whatever model I pick, and then I run the security scan on the same mod...
Code base on a Tuesday or Wednesday, and I get different results, which I will, that's just how probabilistic systems work, how do I explain that to my auditors? What changed between these things? And so I think those are the kinds of problems I am very focused on is, one, businesses have to do what they need to do, which is innovate. We can't completely stop innovation because we have this onslaught of security issues for the last six months.
So manic prioritization, we need to remediate as machine scale the things that matter.
And moreover, we need to start eradicating classes of vulnerabilities. So a year or two years from now, we are kind of past this problem, not dealing with a even higher magnitude problem in terms of scale and volume.
I wanna pick up on two things you said. I wanna go back to sort of the classes of vulnerabilities because one of the things I've been a little obsessed with over the past couple of years are sort of memory safety violations and coding weaknesses that are, like, I think 60 to 70% of most software vulnerabilities are things like SQL injection and cross site scripting and all of that. So I'd love to get your thoughts on how you can actually go from memory unsafe to memory safe. But I wanna go back to something that, know, all of the security folks that I talk to, the biggest problem right now is just this radical compression between the exposure of vulnerabilities and the weaponization of vulnerabilities.
And you talked about accelerated remediation, which is a nontrivial problem, right? Because the ability to be able to identify, confirm, validate, remediate, or put compensating controls around a significant vulnerability that you may be exposed to, again, is is a complex endeavor. So talk to me about the solution for accelerated remediation because that's increasingly becoming the holy grail for security leaders.
Totally, Jen. So I think it starts with first and foremost really understanding what you need to fix. Right? Historically, the way people have done it is they draw a line at criticals and highs that you'd you'd never have time to go below that line. And then you're kind of spending hours and hours of security time, then development time to even figure out, am I affected?
And we've got to get to that answer, like, instantly. Like, at the moment that I know there is something, I need to know is it reachable and exploitable in the context of my application just because a certain version of a certain very popular open source library, as an example, has a vulnerability doesn't mean you're using it in your application in a manner that will be vulnerable or could be exploited. So this concept of reachability analysis, it turns out, which is don't just tell me if I'm affected because I have an imported library in my application, but tell me if in the context of what lines of code in that application, in what circumstance is used in an application, will this be exploitable?
Doing that analysis allows you to weed off about 92% of your vulnerabilities that are currently getting reported to you through your historic mechanisms of how you do code testing, code analysis. So big bang for your buck is understand upfront what affects you. So once I know that, the second thing is if I know exactly how you're using this library in the context of your application, which we had to solve for to deliver the reachability and noise reduction, then I also know exactly what parameters are you calling in which library and how is it used. And I'm also tracking 1,600,000,000,000 lines of code on open source every single day.
Like at Endor, we evaluate 75,000 new package versions in open source every single day within ten minutes or under of them being launched. So now we know all of the code changes that are coming in these new versions and we can run these compatibility assessments to say, if you are affected and you need to fix it, is there a possible path for you to upgrade to a version that would not break your application otherwise? It wouldn't introduce API changes. It wouldn't introduce version conflicts.
So we deterministically run that analysis for our customers and say, great. Good news. I got a path for you. Go to...
Go from version two one six to version two seven nine because that path will mitigate the vulnerabilities and not have breaking changes. Don't try to go to latest. It's got too many changes. It will not work.
So we now give them a deterministic workflow, which now their agents can just go rapidly execute with that intelligence. Now not everything is easy. If it's a spring framework, right, it's something really complicated down to a foundation, you can't upgrade out of it overnight. It might take you eighteen months.
So the next thing we're looking at is because we have this very deep intelligence of your code graph, we can say, is there something I can do in the first party code that can mitigate any exploit being triggered for that vulnerability? And it turns out 80% of vulnerabilities in Java, for example, in open source in Java can be mitigated by some change in that path up in the application you control, the first party code. And median lines of code is under 10. So in less than 10 lines of code change, you can entirely mitigate a class of vulnerability from being exploited in a very complicated library that you've been stuck with for ten years and you can't get out of.
So great, that's intelligence. Knowing that I'm having that choice gives me some flexible options to work with. Now if I can't do that either, then kind of we move to last resort. And the last resort is getting backboards of fixes in, like, new versions of libraries to the twenty year old library you're stuck on and you can't upgrade out of for another six, nine, twelve months, we will take that fix the maintainer's already created, backboard it to the old version, verify it, and give you now what is the minimum viable batch version of the library you're on.
So that's kind of flexibility. Right? This whole idea of, look, if you have a problem, you just gotta upgrade the latest is you can't. It's an engineering problem.
You can't do that in ten hours, And so you need flexible options, and you need the intelligence to figure that out.
Super helpful. Do are you doing anything with Accretis as part of the... I've heard of you guys on that with them?
Yeah. So we were the founding members of Accretis. We're on the governing board. I sit on the governing board of Accretis.
We started that with amazing comp...
Organizations, like, obviously, Pat Opad from JPMorgan and Anthrop, Big OpenAI, Microsoft. And so, yes, I think we're we're glad to see that come together. I think, obviously, right now, there's a lot of angst about this zero day problem, which Akriti's is very focused on. And I think that will be a problem, but I think this is also gonna be a very high upfront problem.
I think the steady state of this problem will not be as high because at some point, we're gonna find all the vulnerabilities we possibly can. The long tail will be how do you get maintainers to fix it, and that's why all these flexible options I discussed become important. But I'm really glad to see this organization come together because this idea that every company that finds bugs just throws them towards maintainers who owe us absolutely nothing to go deliver them on a certain timeline or cadence and having this intermediary organization that can build trust with those maintainers, help them with that work, and also do a lot of the triage and deduplication will go a long, long way.
Yeah. Totally agree. I think, you know, when Mythos was announced earlier this year, my biggest concern was on the open source side, having seen this when I was at CISA. You mentioned SolarWinds.
You know, I parachuted into that job in the middle of the sort of aftermath of SolarWinds, and then we were dealing with Afnium, JBS Foods, Kaseya. So it was just one series of supply chain intrusions after another. And then, of course, we had Log4J in December 2021. So this became a very real thing.
Do you all deal at all with, like, how to do...
Go from memory safety violations to memory safe code like Rust or other languages?
Look, we don't as a base... Like, that's not core to our business. Obviously, lot of the companies that are focused on agentic coding, you know, I know Anthropic has an offering of, like, you know, application migration. I know people are using Devon from Cognition heavily. I think what a lot of the AI coding assistants are helping with or long lived agents is really kinda entire rewrites of tech stacks.
So I think that's where the acceleration's coming from. But, like, our goal in sitting in all of that is working with our customers to create the safety harnesses because, again, all the new code being written by agents isn't magically secure by default.
In fact, the opposite at the moment. So we are creating all the safety guardrails to say, okay, let's make sure we don't entirely introduce new classes of problems when we're writing in Rust. It may not not be the memory problems, it might be something else, It might be logic problems, Authentautz issues, iDOR, all that. So we wanna create all the safety harnesses because one of the biggest struggles I'm seeing, Jen, and, like, you know this, hundreds of companies got access to Glasswing.
Lots of them had free credits to run these because they're mission critical. But what everybody's realizing is these tools are great at adversarial testing. They can be used in in lieu or for conjunction with pen testing. Like, you can run these massive scans, they cost you millions to tens of millions of dollars, you kind of run them once in a while when there's some very material changes, but you can't run them for speed, cost, and efficacy reasons on every line of code being produced.
Like, it it is not a proactive, like, scan every line of code as it's being billed, every pull request with those models. So I think what everybody's left within the industry right now is great. We got a taste of what amazing things these models can find, but I'm not gonna spend $50,000,000 a year running these on every pull request, and by the way, it might hold up a pull request for three hours to twenty hours because of the time it takes. So what is the answer to the defensive, like, when I have 7,000 pull requests a day in my company, what am I scanning those with?
And obviously, historically, we run these, you know, static code analysis, SAST like tools.
They have been interesting is what I would... Be the word I would use, like the check boxes, but they... The best ones fully tuned generated 50, 60% false positives because they didn't have the context of your entire environment and application.
But more interestingly, they didn't catch interesting bugs. Like, they caught basics, OWASP top 10 injection, you know, across the scripting, but not the things that put companies in the news, like completely missing authentication authorization or, like, logic issues were never caught. Right? The logic issues were either caught in human reviews, but most importantly caught in bug bounty programs or pentas.
The reality is we are seeing today, and we're doing it for our customers, that you can find those business logic flaws with the right LLM powered technologies, but it's not just pick up a model off a shelf and run it because you're gonna blow it through the bank. So our customers are using Endor Labs and our assistants to create these defensive harnesses to scan every pull request, every line of code defensively. So when those rewrites are happening from memory unsafe to memory safe languages or wide coded apps are being created by marketing and finance, we're scanning all of that code and we're deflecting all of those security issues and kind of that's where I was going with eradicating these vulnerability classes because the best vulnerability is one security never has to track on their backlog.
100% on that one. Let me go back to just the whole founder leadership piece of this and ask you, what's the hardest thing you've had to change your mind about since you started on this journey?
Look. All three journeys have been different. I think the first two were category creation. Like, when I started, there was never...
SaaS security didn't exist. Cypher Cloud was the defining company that started two years before Skyhigh and Netskope to even create this cloud access security broker category. And in 2015 when we started RedLock, nobody had budgets for cloud security. Like, they still didn't understand whose problem is it.
Why can't Amazon just solve this thing? Why do we need a product to detect when s three buckets turn loose and are publicly exposed? Right? So I think those two experiences were much more around education.
Like, I needed...
It took us years to explain to companies in the industry that were adopting these technologies really, really quickly what were going to be the problems? Right? You're telling people about problems they haven't faced before, and that's kind of complicated because it's like, okay, why should I believe you? And, you know, over time, what I found in those cases is leading with a lot of security research and insights of, like, data.
You know, for example, at RedLock, we were the first ones that reported that Tesla's cloud environments were being used for crypto mining.
Right? And so, like, being able to take those scales or massive Fortune five hundreds data were leaking on open s three buckets. So a lot of it was education, education, education. But there, when you're creating a new category, you can start with a feature or a set of features that are really powerful and then build your way over multiple years to a platform. In contrast, at Endor Labs when we were starting, there was a well defined application security testing market. It had been around for ten to twenty years.
And so when you're doing kind of a displacement as a as a starting point, you've gotta build everything the legacy players have and then double double on that because people aren't gonna go change technology for 10% better or 20% better. And so the capital you need, the patience you need to build something much larger, deeper, wider is different when you're kind of in a mature category but trying to displace it entirely.
I would say across all of them, though, the the two things that have been kind of very consistent to me or or three things maybe. One is people matter more than anybody else. As I say, when you start a company, you have no product, you have no customers. The only thing you're betting on is to hire the right people that will help you build the best platform.
And so I always say don't skim on the quality of talent, what you need to pay them, what you need to give them in equity because a 100% of zero is zero. A lot of people are like, well, you know, I'm gonna get a few, interns and get started. It's like, okay. If you have no other choice, I get it.
But if you can, you've gotta figure out how to get the best people in because also those best five, 10 people will go get you the next 20, next 30. The other thing was the diversity of talent. We were solving, as we've been discussing here, a hard engineering problem. So what I wanted to do is make sure...
The easy answer for me would have been go get the first 10 engineers from RedLock and Palo Alto Networks who've been working with me. But instead, we went down the the high road, if you will, which is I kinda made a commitment but to myself, which is no more than two engineers in the first 15 would be from the same company. So we kinda got two from Uber, two from GitHub, two from Meta, two from Cisco, two from Palo Alto Networks, and that's kinda how we built this team because we want the diversity of that engineering excellence and experience to build the company on.
The the second thing is obviously being customer obsessed. I I I have seen this mistake be made especially in cyber a lot. Right? Really smart people, really sharp brains wanna solve a problem for tomorrow but don't have a way to connect it to the customer's problem of today.
And I say if you don't build a bridge, if you're too far out there, you might have the best technology solving the most advanced threat detection management problem, but connect it to the the here and now. Otherwise, you're too far ahead. Everybody in security, the buyers, you know them better than I do, but, like, they're always, like, fighting the fires of today, trying to make sure the decisions they make will hold them together around the corner...
The next corner, but you're not gonna go buy the thing for the next corner if that's not a problem for today. And so I think that's a thing people get wrong a lot, and and that's kinda where I have a fundamental difference of, like, a lot of people are like, well, I'm in stealth. I can't tell you what I'm I'm doing. Like, okay, I get it.
I respect that. My perspective is build in the open because if somebody can just take what you tell them in fifteen, twenty minutes and copy you, then shame on you and and good for them. What you're trying to do is validate that idea as much as possible and hopefully you have the better execution and the best intellectual horsepower to solve that problem that sharing the idea or the concept in fifteen minutes doesn't give your whole company away.
Yeah. I love that. I love that. It's very different than most of what's out there, frankly. I really love the transparency and what you said about if somebody can copy it, you ain't adding any value. So what what founder trait do you think is wildly overrated?
Oh, hot seat.
My vent...
My VC network is very strong. I've kinda heard a lot of people come in and say, look. I can start a company in cyber. It's easy.
You know? Cyber's hot. I'm gonna come start a company in cyber. I've got a strong network.
I'm gonna get funded. I I think that's obviously great. I think cyber as an industry, though, I always tell people is a lot...
Is built on a lot of trust and reputation. And the the better advantage for you would be you have done things over the years that people will respect you for in cyber, not, hey. I can raise a bunch of money so thereby I can build a successful cyber company.
I think that's probably the biggest fallacy. The other side is also true, which is I have seen a lot of former CSOs, leaders start companies, and their biggest realization, though, on the flip side is how hard building a company is versus building a technology. And we can get into that. That's not the question you asked me.
But, like, on the overrated side, hey. I have a great investor network. I can raise money. It's, like, great.
You'll get your first round. I don't think that's necessarily a claim to a successful company.
Yeah. It sounds like, you know, it's been quite a journey since you started your first company, and I'm sure it's been a learning journey.
In your mind, is there a mistake that you made that you're really glad you made it?
I think it goes back to the the build in the open. I think in the...
In, like, my first run, it was a lot of hush, hush. Everybody's gotta sign an NDA if I'm gonna tell you more. I I think it was just friction that was unnecessary in the process, delayed the conversations I wanted to have. At Endor Labs, I've...
I never cared about getting an NDA. If you as a customer want an NDA to talk to me, we'll do it. So I would say that's probably the biggest mistake in the past. I think the other thing is something that I learned very quickly, which is around sales, is the hardest thing, Jen, is a lot of us builders can suss out very quickly good engineers from bad engineers, good product people from bad product people.
Sales is hard because sales requires intellect. It requires athleticism or or just competitiveness.
It requires grit. It requires somebody to be able to grind. They have to be in the right place in their life to be able to grind. Most founders, it's not natural to us to figure out the right from the mediocre, from, you know, maybe a red flag situation.
And so I have always found, like, if you hired a single salesperson in a company, you're starting out, you wanna start sales, you get the first hire, you are banking on that person to effectively tell you if your product strategy and your product market fit is there or not. And if in the next six months, it's not working, what you don't know is if they are not working for you or truly the product is not working. And so I always now give this guidance to people, which is hire in pairs when you're hiring sales, hire in a couple of different territories like different markets. Right?
So maybe a West and an East, hire people with slightly different, think an AB test to your sellers.
Because if both sellers with slightly different backgrounds, maybe one's been at a start up, maybe one's just got an incredible track record of some larger companies and has a strong network, maybe one is earlier in their career, maybe one has that established channel partner, another ecosystem, and none of them can crack the code, then the problem probably is the product of the market. But if one of them is crushing it and the other has nothing going for themselves, then you know where the problem is.
Totally. That makes great sense. I'm sure you do a lot of mentoring. What's the one piece of advice that you give to aspiring founders?
First and foremost, try to scare them away because I think a lot of them underestimate, especially in cybersecurity, that having a great idea doesn't necessarily mean you're gonna have a great company. It's a crowded space. It's an unsaturated space. It's overfunded.
Building the...
Even if you have a great product, getting it to market is the harder part of the equation. And so I kinda try to leave them with that upfront that don't underestimate the second part because I just see a lot of people tap out in a year because they built a great product and they're like, damn. Like, I didn't realize this from the other side how hard it is. And I'm like, well, karma, when you were the buyer and, you know, the vendors used to reach out to you, you're ignoring them.
Now you have a different title. People don't wanna talk to you. You're on the other side. So I think that's certainly one.
But in all seriousness, I think it goes back to the the bridge building. Like, don't work on an idea so far out that people, customers, practitioners can't connect to that problem based on what their roadmap is right now and through the next twelve months. So kind of...
I say, look, if you wanna have a really, really big idea, like, that's your that's your ultimate landing zone, you gotta build a bridge to get there. Like, what is the problem that is here now that you're gonna connect to? Just way too many times people are always, like, in the I'm on the other planet. I'm like, dude, like, people don't know how to get there.
They they can't resonate with this problem. And I think that that happens all too frequently because we all wanna change the world, and we wanna change it a little too fast. And that's not how, you know, the cybersecurity industry works.
So coming down to the wire here, and you just teed up my next question, which is if your company succeeds beyond your wildest dreams, what changes in the world?
I think it kinda goes back to where we started, Jen, the conversation, which is security. It's always been this adversarial tension...
Adversarial such contentious relationship between engineering and security. It's always a or, not an and. And, ultimately, for me, I think we will...
If we're successful, the world will just be creating secure software by default without us moving through unnatural hoops in lots and lots of tickets and large massive workforces of people that are doing stuff that is heroics at the end and putting Band Aid together on a release candidate. It would be just vulnerabilities are eradicated, software is securely created by default. Doesn't mean we're never gonna find problems after the fact, you know this better than I do. We will always find problems, but that'll be the 10% of the time being spent.
90% of the time is all on secured defaults, which is I don't even think about as a human squiggly lines of ID and annoyances of tickets. Stuff is happening. We have self healing software that is just magically secure, and I don't think about security any differently as...
Than a quality problem. And I think what that will also require...
I I I fundamentally believe security of software needs to be a CTO's problem, not just a CSO's problem. And so I think a lot of organizational shifts need to happen as well.
Yeah. Amen, brother. I actually think it needs to be a CEO problem as well, to be honest.
Alright. Last question for you. You've been on our stage at the Innovation Sandbox three times, and it's a pretty stressful experience getting up there.
So we're about to open our window for new companies to apply, which I'm super pumped about. It's going to be the best ever. What's your best advice to somebody who is both submitting, but also, like, getting on that stage? Pretty intimidating. Like, what what would you advise them to get that pitch right?
Look. I think the authentic, all of us as founders have this passion in us. That is why we started this company. I think what happens in the process of preparing for RSAC and innovation sandbox is we're, like, trying to memorize every word to get it down to the t, and what happens is people get nervous because they missed one word, and they can't remember the next thing is, remember, we're gonna find that inner passion.
You started this company because you truly believe in a large scale problem and you wanna be the change agents to fix that world. Bring that passion out. And, obviously, you gotta practice, you know, the the time management is the biggest thing, but, you know, trying to rehearse and remember every single word is challenging, and you will get nervous up there. So bring your bring your inner fire, bring your passion, you know, remember why you started this company, and let everybody resonate with that problem.
That's awesome. Well, I hope I will see you at Founder Circle coming up in the funk... In coming six months or so.
And, thank you very much for this awesome conversation.
Is there anything that I didn't ask you that, I should have?
No. I think this has been an incredible conversation. Thanks for having me. I think we're at an incredible transition point of, like, this whole...
Is cyber needed? Is it not needed? You know, obviously, the last few months of, public markets will tell you that I think the believers will believe that cybersecurity will have way more tailwinds now, with AI. Cybersecurity is not going away.
I think fundamentally, like you have with everything else over the last twenty years of technology transformations is you have the makers, you have the checkers. You've gotta have security as your independent layer in everything you're doing with AI.
You know, you can't have the models that are doing the work also be verifying their work. I think it's somewhat of a dangerous precedence as we have historically seen over the years. And I think just keep your heads down, stay focused, and, you know, I think we will all come out of this feeling like we have a better security maturity than we do today, but the next year is gonna be painful. And so hang in there, and we're all hanging there with you trying to solve these problems.
Fantastic. Well, thanks so much, Varun. It's great to chat. I really appreciate your time.
Yeah, likewise. Thank you.