334: From Special Ops to Cyber AI Warfare: Lessons in Grit, Leadership, and AI Cybersecurity from Snehal Antani, Horizon3 CEO
43:44 Share

If you don't care deeply about what you do, if you don't love what you do, if it feels like a job, you're doing the wrong thing. And so this idea of being passionate and obsessed at what I do and mastering my craft because I love it, and the grit to overcome the problems and obstacles that come along with any hard problem, that really is my origin story and that really underpins my background and every decision I've taken.

Good morning, good afternoon, or good evening, depending on where you're listening. Welcome to AI and the Future of Work. I'm your host, Dan Turchin, CEO of PeopleRain, the AI platform for IT and HR employee service. Our community is growing thanks to you, our loyal listeners.

If you like what we do, go ahead and click subscribe on the newsletter button in the show notes for today's episode. If you do, you'll get some additional fun facts and tips and tricks that don't always make it into the main episode. Also, if you like what we do, please tell a friend and of course, give us a like and a rating on

Apple Podcasts, Spotify, or wherever you listen. If you leave a comment, as you know, I just may share it in an upcoming episode like this one from Bruce in Seattle, Washington, who's an HR director at Amazon and listens while commuting. Bruce's favorite episode is the one with Dr. John Boudreau, a real future of work pioneer. That's a great one about the new definition of work and the future of organizations.

Of course, we learned from AI thought leaders weekly on this show and the added bonus, you get one AI fun fact. Today's fun fact, Lewis Columbus writes in Venture Beat that enterprises run the risk of losing the AI arms race to adversaries who weaponize large language models and create fraudulent bots to automate attacks.

Adversaries are using generative AI to create malware that doesn't create a unique signature, but instead relies on fileless execution, making the attacks often undetectable. GenAI is extensively being used to create large-scale automated phishing campaigns and automate social engineering.

with attackers looking to exploit human vulnerabilities at scale. The real challenge is that AI-powered attacks aren't a single event, they're a continuous process of reconnaissance, evasion and adaptation,

My commentary, unfortunately, bad actors will always exist within and beyond the cyber realm. And social engineering is the biggest threat vector. Within the cyber realm, they'll always exploit both human and technology vulnerabilities. Security training is more critical than ever.

And so is making it harder for criminals to use ransom tactics and shadow currencies to benefit from successful exploits. Sadly, this is a problem that goes far beyond cybersecurity. Thankfully, the brightest minds in tech and policy as well are working hard to stay a step ahead. This is very irrelevant to today's conversation.

And of course, we'll link to that full article in the show notes. Now shifting to today's conversation with the great Snehal Antani, who's an entrepreneur, technologist, and investor. He's the CEO and co-founder of Horizon 3, a cybersecurity company that uses AI to deliver red teaming and pen testing as a service. Snehal also serves as a highly qualified expert for the U.S. Department of Defense, driving digital transformation and data initiatives in support of special operations.

Prior to his current roles, he was CTO and SVP at Splunk. He had multiple CIO roles at GE Capital, and he started his career as a software engineer at IBM. Stan Hau has a master's in CS from Rensselaer Polytechnic, co-engineers.

and a BS in computer science from Purdue, go Boilermakers. And he also holds 16 patents. Full disclosure, I'm an investor in Horizon 3. PeopleRain uses Horizon 3 for pen testing. And perhaps most important, this doesn't require full disclosure, but I consider Snehal both a friend and a mentor.

And I'm so excited that you all get to meet Snehal Antani today. Without further ado, Snehal, welcome to the podcast. Let's get started by having you share a bit more about that illustrious background and how you got into the space. Yeah, I know. Dan, thanks so much for everything. And it's funny, I learn from you in every interaction, especially your deep care for people.

And that is before I truly understood what PeopleRain did. And that just was incredibly infectious and inspiring and crazy.

crucial at the time because we started Horizon 3 right before COVID. And then COVID hit. And then you flip from being a face-to-face company to suddenly having to scale a remote company. And not knowing how to do that, how to preserve culture. And so I think that a lot of the care you had for people that I was exposed to prior to COVID is

created lessons and experiences I could draw from during the tumultuous time of scaling people in a remote environment with all sorts of challenges there. So thank you for that inspiration as well. But yeah, so my background, software engineer by education and trade. My freshman year at Purdue, I saw Drew Brees play Tom Brady, which is pretty awesome. And then I started my career at IBM as a software engineer, eventually became a CIO at GE Capital, and then moved out west to be the CTO at Splunk.

Left industry to serve within special operations for about four years and then finished that role in 2021. And then shifted my focus to building and scaling Horizon 3. And so we've been at it now for about five years. And it's been an incredible journey of scaling from my co-founder and I, Tony, to the early engineering team or founding engineers of about eight of us.

to over 200 employees today, growing revenue 100% year over year, driving efficient sales growth and all the things one hopes to pull off in the era of cybersecurity and hyper growth companies. So before we started recording, you said, what do you want the audience to get out of this? And I said, I want them to meet you, just the authentic Snehal that I know and the one who has a really unique leadership style. How has your

your roles at JSOC, Splunk, etc. How has that influenced the culture and your leadership at Horizon 3? It actually goes way before that. So we all have people we look up to or looked up to in our lives. And for me, it was my dad. And what my dad instilled in me at an early age was,

was this idea of grit and passion. So grit, typical, he immigrated from India in the 70s, typical immigrant story. You come with nothing, you're self-made, you go to school, you work part-time to cover tuition and everything you've got to do in that era. And grit's about

overcoming whatever obstacles that throw in your way to achieve the goals you set out to achieve, right? In many ways. And that's what my dad did. And in every story that he would tell me was something about how he would go to his master's classes during the day and he had no money. So he had to work. And so he would get out of class, work the second shift grinding plastics, uh,

at a local manufacturing facility. And then would work the third shift cleaning out the garbage trucks at the end of the day, just to get a little bit more money. So he'd be in a full rubber suit and power washer spraying the inside of garbage trucks. And he'd come back and then he'd do his master's graduate school work. He'd do the same thing the next day and the next day. And that's how he put his way through grad school. I bring that up because a key part of what drives me is

It's justifying the sacrifice that he went through, my parents went through as immigrants. And hey, it was worth it. Look at what you were able to create as opportunity for my brother and I. And so grit in this chip on my shoulder to achieve goals and justify what he went through. And they never asked me to do this. They never said anything. My parents never did of expectations. I didn't have any pressure to pursue a profession. It was...

a quiet inspiration of just me observing how they operate and how my dad operated in particular. And he recently passed, and so it's very close to me. And the other part is passion. Man, no matter what, my dad never, he never went to a day of work.

meaning it was never a burden. He loved what he did in every way. When he would come home from being an electrical engineer, he would bring broken toys, give me a multimeter, a soldering iron, and a screwdriver, and then teach me how to fix it. And that's how I learned systems thinking, troubleshooting, design. And what I learned from my dad there was,

If you don't care deeply about what you do, if you don't love what you do, if it feels like a job, you're doing the wrong thing. And so this idea of being passionate and obsessed at what I do and mastering my craft because I love it, and the grit to overcome the problems and obstacles that come along with any hard problem, that really is my origin story. And that really underpins my background and every decision I've taken, whether it's career or academic or anything else.

That's a great background. I've never heard you share that story. So roll back the clock many years, mutual friend Hunter Moeller introduced us. And it was Horizon 3 was you and Tony and a spreadsheet with some names of customers that aspirationally you wanted to sell to. One of the things that always impressed me in the early days is that the growth was always incremental. You always had a very

humble outlook. There was always a clear next milestone and you were always open about where you were, where you weren't, where you want to be. You always ask questions. You know, it was very receptive to coaching. And so,

Sanja, who was a CTO, etc., but now a first-time CEO, I saw you in real time learn to be a world-class CEO. Talk us through your perspective on every day, running into a wall, something that you had never experienced before, but continuously just picking yourself back up and growing a team in that image.

Yeah, and this is all pre-ChatGPT, where you had to learn subjects the hard way, which is like Googling it and picking up books and all that stuff. So break kind of my career or skill sets into a set of Lego blocks. And this is something I also learned from my dad, which was everyone wants to, especially millennials, want to flip the button and become CEOs straight out of college.

But the reality is there are a whole bunch of Lego blocks or skills that you really want to go learn and experience and make mistakes in before you're put in the CEO seat. And that's just something that I had applied in my approach to being a founder. And so while I was the youngest in a whole bunch of career accomplishments, I was considered an old founder at 39 when I started Horizon 3. But

The first Lego block I focused on early in my career was building technical expertise and credibility at IBM. In middleware, working on a web sphere, working in research and development, it was all about mastering a particular subject and being viewed as an expert in a bunch of areas, in this case around distributed computing and distributed systems.

And then I took on lateral jobs. So I actually took jobs at a pay cut or flat pay scale or lateral move from a career standpoint. Much to the chagrin of my other colleagues and friends at IBM, I left an R&D job and I took a product management role. And

The question was why? Well, I wanted to learn how to launch products. And I was an awful product manager for the first six months or so. I didn't know what I was doing. I just had to learn and figure it out. But I had to take that risk because there was a set of skills I wanted to learn. Every job I took, I took with the intent of learning some critical Lego. I felt I needed to be a good or great CEO.

After that, when I joined GE Capital, I'd never been a manager before. And GE Capital made the bet on me where I became a CIO and they said, look, we will teach you how to manage people. And they were world class at that.

At the same time, I came from a very deep technology background and at GE Capital is about helping them drive a digital transformation. And so I wanted to learn how to lead organizations at scale and leverage technology to create business value. And I was awful the first three, four, five months, but I figured it out. You keep your mouth shut, you listen, you learn and earn the right at the B on the team every single day. And I'm going to come to that story in a moment.

I moved out to Splunk, amazing opportunity. It's a company I deeply admired with the CEO, Godfrey, that I deeply admired as well. And the key lesson I wanted to learn was how to scale go-to-market. So I focused on being an outbound CTO versus a down-and-in R&D CTO.

And I wanted to be an outbound CTO because I wanted to be on the road working alongside and learning from the chief revenue officer, the CFO, the CEO, and how to scale a company. And you're not going to learn how to do that unless you carry a bag and you're under the gun. And then I did something totally unconventional. I took what we all, I think, hope to do is a 99% pay cut.

Left industry, couldn't talk about what I did, couldn't update LinkedIn or nothing, and left to serve within the special operations community, supporting the counterterrorism mission amongst other things like that.

And I did so taking this pay cut because one, I wanted to work on problems that truly mattered, that made a difference in people's lives. And going after terrorist organizations and keeping people safe was a problem that mattered. I was 20 years old when 9/11 hit. From that day, like most Americans at that age, we wanted to find a way to do something about it. And I found a path to having an impact.

But the other thing is, I basically viewed my time at JSOC as a PhD in leadership. I got to sit next to the greatest leaders the world will never know about, never hear about because of the organization they're part of.

Listen, learn and become better on a daily basis by working alongside these folks. And I feel those were the key Legos that allowed me to earn the right to start a company and be responsible for the livelihoods of 205 people today. Because when you're a founder, you know this, when you're a CEO, there are families depending on your ability to lead the company through good times and bad times.

There are families that are looking at that vacation house or that new car or whatever, imagining the payouts they're going to get because of how stressful and difficult startups can be. But it's high risk, high reward. And so I feel if you're going to lead and have that kind of influence on families, you better put in the homework and earn the right to do so. And that's what I focus a part of my career on.

So the culmination of, I didn't realize this, like literally decades of being in the batter's box, so to speak, getting warmed up for a seat in the CEO chair led to Horizon 3. Why did you pick that particular problem to solve? Red teaming as a service is certainly innovative, but the surface area of problems you could solve coming from that background is really, really wide. Why Horizon 3? No, it's a great question. So there were...

There were three ideas I was thinking through as opportunities I thought would be interesting. So the first idea was around the orchestration of things. And we're actually starting to see Anderall do that today.

And that problem was more around, you're going to have different kinds of drones and autonomous systems that are specialized at solving different kinds of problems. You'll have a fixed-wing drone good at doing long-range surveillance of something. You'll have quadcopters that can do precision maneuver. You could have ground vehicles that can do things like till the soil.

And if you're a Vintner in Napa, you can swipe a credit card and run like Amazon drone services, if you will, that says increase the yield of grapes in this field. And that should trigger a drone to do surveillance, find dry patches, understand water dynamics and so on, get a micro drone to drop sensors all over the place and use that to take some action. So I thought this was going to be an incredibly interesting problem to solve. I still think there's a massive market there.

But in 2018, 2017, being early is the same as being wrong. And so I said, I think that this can be a really valuable problem, but not for another 10 years.

And I think that's where it's gonna turn out to be. There just wasn't enough diverse things. I had another idea around edge cloud analytics, and then this problem around cyber. But what I realized early was the idea itself doesn't matter. What matters in the early stage of a startup is the team you can assemble.

And so I was ready to start a company and I bounced around trying to find the right early team and I just couldn't find them. Now, I kind of did this speed dating thing out here in Silicon Valley with people in my network and people of people in my network. And I just didn't find that click with a co-founder and early team that got me excited.

So my time at JSOC, my deputy CTO, Tony, he and I clicked really well. We worked on some really hard, really stressful problems. This idea of pen testing and hacking was a problem dear to my heart. I'd been a CIO. I was tired of having no idea if I was secure until the bad guys showed up. I wanted to verify my security posture for every patch Tuesday. I wanted a pen test Wednesday, and there was no way to do it.

And my bet was could I build an autonomous system or can I use AI to hack an organization? And that was the gamble. And Tony and I started brainstorming and the world we came from, special operations and intelligence community, we had access to offensive cyber experts. And we found that early team of like-minded individuals that love to work together, that were the right culture fit, that worked well under pressure. And once again,

The team matters the most and then the problem. And so for me, it was finding the right team and aligning around a problem we all cared about and then moving out from there. So you mentioned you launched the company essentially pre-LLM. Gosh, since then...

The threat vectors have changed, the frequency, the patterns, the whole approach. And this is not unique to Horizon 3, but just everything about the business that you launched into changed. What was the process for recognizing that and innovating around these rapidly changing cyber patterns? Yeah, I mean, back to you want to get exposure to trends and new capabilities, right?

in a way that allows you to form a good hypothesis for where you think the market's going to go. So at DoD, I was lucky to be working closely with Project Maven, which is the AI task force at the Department of Defense. And they were in the news for a bunch of stuff back in 2016, 2017. And my exposure to Project Maven and Drew Kukor was running that project. He's a brilliant marine. He's now, I think, at JP Morgan. But

He and I went back and forth and spent a lot of time talking about AI and the value chain and what matters in that problem set. He said this was 2017, 2018-ish, 2019, around then. Pre-LLM, pre-ChatGPT, pre-OpenAI.

And it was clear to us that the weights and models in an AI system don't matter. What matters is the training data. That's the crown jewels. And so if you're going to build an AI company, you better have access to proprietary training data. That was the X factor. And a great example of that in industry is Tesla. There is no public training data set for full self-driving.

And so what did Tesla do? They installed cameras on every car and they started collecting full motion video off of every vehicle on the road, whether they paid for FSD or not, didn't matter. And they built this massive corpus of training data.

Knowing that compute would and the cost to train was going to decrease over time. The weights and models were going to be throw away and you're going to refresh them every year, maybe every six months, maybe every three months as the rate of innovation increases. But the training data was going to be durable and persistent. So I viewed Horizon 3 not as a pen testing company. In many ways, we're a data company. Pen testing is just the sensor because

Because pen testing gives you ground truth of your security posture. So we run more pen tests a day than big four consulting firms run in a year.

We ran more pen tests last year than the entire history of computing. And every pen test is collecting telemetry and training data that only we have access to. And so the way that we built the system was around this data first platform, pen testing as a center. And now we're able to move into adjacent opportunities because if you're running a pen test and you want to deploy honey tokens or deception, well, that's just

different questions of the same data pen testing results are a map and compass that allow you to drive defensive actions and that had been the the vision from the jump in 2019 and uh so that hasn't really changed the team has scaled up the culture has scaled up and we haven't had to do any pivots or any major changes or rewrites because we're a data company pen testing is a sensor so uh

You mentioned a joint special operations man, JSOC. You wish that there was such thing as a pen test Wednesday to follow up a patch Tuesday. I like the way you said that. For those who are outside of the cyber realm, just maybe talk about the role of red teaming and then the innovation behind red teaming as a service that supports the volume of pen tests that you're talking about. Yeah, so the whole problem in cybersecurity is that

And actually the term red teaming is a military term from way back in the day. So if you have battle plans as a military unit, well, all those battle plans look great in PowerPoint. They look great on paper, but rarely do they survive first contact. Everyone's got a plan until they get punched in the face. Well, you can either wait for the enemy to punch you in the face or

Or you can hire a friendly team and have them shadow box you or punch you in the face on your own terms first to make sure that the plans are actually going to be robust and work out well. And so the whole idea of red teaming and military planning operations was to look at your plans through the eyes of the adversary and understand your weaknesses and understand how to make your plan that much better. Well, the same thing is true in cybersecurity. Look at your network through the eyes of the attacker and

the attackers would go after to steal your data or destroy your systems. And that allows you to then harden your defenses. Now, the next thing you can do is, if you're paying somebody to break into your house to figure out where to install locks and other security systems, well, if they find a broken door or broken lock, they can fix the lock along the way. They can install ring cameras along the way. They can do a whole bunch of things where while breaking into your house, they're improving your defenses.

And that's where I think the real value is, is don't just tell me I've got a broken window. Fix the window for me while you're already in there. And so that's really what we ended up doing as a company, was finding, fixing, and verifying the remediation of security problems that we found in your environment. I never thought until you said it as a Verizon 3S data company, but collecting all that telemetry,

gives you an amazing advantage when it comes to the remediation brilliant brilliant so i talked in the opener about increasing use of ai by the bad actors to exploit new vulnerabilities identify exploit new vulnerabilities but obviously the good actors are learning as quickly as the bad actors

Talk to us about that kind of cat and mouse game, but then specifically on the defensive side, how Horizon 3 is using AI to combat this increasing velocity of threats. Yeah, let's split that up into a couple of different sections. So the first section is around this idea I talk about, cyber-enabled economic warfare. And once again, I have this national security experience now, even though I was a tourist, and I will always view myself as a tourist experience.

in the special operations community. Because my military experience prior to serving at JSOC was watching Jack Ryan in Tropic Thunder. I didn't come from a military background. And I showed up on day one after being assessed and selected to serve in that unit as its first CTO and had to earn the right to be on that team every single day, which was amazing. But when you think about cyber-enabled economic warfare, there's a story from 2022 where the Japanese had provided support to the Ukrainians.

And allegedly, the Russians got angry and ransomware at a small company in Tokyo. And that company in Tokyo provided cup holders to Toyota.

Due to just in time logistics and lean manufacturing, Toyota had to shut down 28 production lines, causing nearly $400 million in economic harm and damage over cup holders. Who would have thought? Now, the flex by the Russians was not that they could ransomware a small company. It's that they knew where to apply the least amount of effort to cause the maximum amount of pain all below the threshold for war.

Because the Japanese or the Americans are not gonna spin up the 82nd Airborne and go to battle over cup holders, right? It is not an act that is egregious enough to justify military action that the population would approve. So if you take this at scale, it's not just criminal organizations trying to extort you through ransomware.

Now you're starting to see cyber attacks in support of national objectives by highly resourced countries, the Russians, the Chinese, the North Koreans, the Iranians, so on and so forth, and then Western countries as well. And so we're in this completely new era. The next part of that era is the role of AI in speeding up attack. Now,

The primary use of AI today is almost like Google search on steroids, where a threat actor with not a lot of effort can learn a lot about your organization. They could learn a lot about the supply chain for building Ford F-150s. And with very little effort, you can quickly figure out that there is a small company that supplies all of the rear view dimming mirrors for Ford. And that company,

If you attack them and cause disruption of their ability to produce rearview mirrors, we'll shut down the F-150 production line.

It took me about 10 minutes of searching in ChatGPT to figure this out. So you can use AI to accelerate reconnaissance of a target. You can use AI to personalize phishing emails and social engineering to increase the chances of getting through the filter, the defensive filter like, hey, I don't think this is a real person filter that the Nigerian princes struggled to overcome back in the day.

But the real value in AI for cyber is using AI to accelerate decision making. So when we pioneered AI for hacking, it took us 7 minutes and 19 seconds for our AI engine to hack a bank. No humans involved, no knowledge of the network, point, click, shoot, and took over the entire bank and all their Gucci security tools didn't work and stop us in 7 minutes and 19 seconds.

That was in 2022, about a year later, it took four minutes and 12 seconds. Point, click, shoot. And we actually had a nine-year-old kid as a proof of concept, click on the button to prove that you needed no senior skills to pull this off. The barriers for entry in cyber attacks had been completely eliminated. And so now it's sub 60 seconds. So in 60 seconds or less, can a defender at a bank or a healthcare company or anywhere else

detect the attack, stop the attack before the attacker succeeds? And the answer is probably not. So the future of cyber warfare is algorithms fighting algorithms with humans by exception.

And so this speed of attack combined with cyber enabled economic warfare, I think puts us in completely new territory that is gonna make the problem quite worse. Not to be the bearer of pessimistic news, just I'm a realist in this regard. But it creates a ton of opportunity to rethink how we do defense.

i like the way you put it the the first of all i hope the future of warfare is bots fighting bots because um i'm in the yan lakum camp that says if we're smart enough to create this technology we're smart enough to turn it off and uh i'm not exactly henry kissinger but it seems like you know the solution might be you know if you can find the uh the plug to pull from the bad actors that are powering you know the uh

the bots, the bad bots, then maybe we wouldn't need so many good bots as well. I realize that's pathetically naive, but what is the long-term fix? I mean, this just proliferation of bot arms, the bot arms race, it can't keep escalating. If you don't have hope, then why are we here, right? So there's always hope. I think what's interesting is

In 2022, there was an updated national defense strategy around this concept of integrated deterrence. And if you look at cyber threat actors today, you're either a criminal organization trying to make money, or you are a criminal organization acting on behalf of a state actor, right? So if you take Salt Typhoon, there is no direct connection between Salt Typhoon and the Chinese government.

And allegedly, if you have access to classified information, they've got those clear connections. But to the public, it just looks like a criminal organization. There are Russian proxies like Cozy Bear and others. They look like criminal organizations. There is no concrete way to prove that they are in fact arms of a nation state.

But there's a lot of, you can't see a black hole, but you can measure the effect that black hole has on the stars around it. And so there's plenty of ways to indirectly observe that a lot of cyber activities are on behalf of nation states. And so if you want to dramatically change cybersecurity and cyber attack risk, you've got to kind of bring those big players together and create enough deterrence that the activities just stop. And there's really,

countries, five countries that are behind the bulk of the cyber attacks. Whether it's the Chinese for espionage and IP theft or whatever else that we see there. Or North Koreans for ransomware in order to generate funds and value. Or the Russians in order to retaliate or pester particular countries that are not aligned with their point of view of the world or whatever else there might be. And so integrated deterrence in part is around synchronizing

activities across the State Department, the Treasury Department.

covert action and cyber initiative action in order to say, hey, look, we could do this too if we wanted to. Why don't you cut back, we'll cut back and we'll create this off ramp. That's like the optimistic view of the world, that there's enough deterrence to create off ramps for these things. But the bulk of offense or bad actor cyber activity can still be traced to mostly nation states.

using criminal organizations as a proxy to achieve their objective. And how you had me at hope.

I agree. And I do believe that there's a diplomatic and certainly a peaceful way to de-escalate the arms race. So I'm going to just here on the show, I'm going to declare Snehal Antani king of the world or something like some title like that and give you a crown, a scepter. What do you do? What are the first few things that you do to

start us down the path of a diplomatic solution. One of the big challenges, I mean, I had a front row seat for some of this on the government side for a bit. It is very difficult to truly synchronize activities across each of the agencies I described, right? We're seeing, forget about political affiliation for a moment or not, right? Just look at the disconnect between State Department, the Department of Defense, Treasury, and USAID.

as an example, where those four organizations and the trillion plus dollars they were spending each year aligned towards a single policy objective or a set of policy objectives at the executive branch level? The answer is no. So the hardest part of this problem is driving alignment across these agencies to act as one cohesive entity.

Now, we'll see a very different effect in other countries, right? So for example, between the Chinese Belt and Road Initiative and other initiatives that the PRC or the Chinese government goes off and runs. For the most part, there's a significant amount of alignment across all of the agencies and entities and initiatives that

on the Chinese side of the house. And so I think what you're seeing is a much better one plus one is three outcome for them than for us. I think what we see in the United States is one plus one equals 0.25 because of that lack of alignment. And so I think the really big thing is how do we drive better alignment of executive branch activities to start and then actually get meaningful collaboration between the executive and legislative branches.

A lot more hope to hope for there.

Hey, when you're the king of the world, it'll be your decision to make. Good answer. I know it's tricky. That's a good starting point. Hey, unbelievably, we're almost out of time here. I know we're just getting started and I feel bad that I sent you some discussion topics because I don't think we got through any of them. But you're not getting off the hot seat without answering one last important question to me that's probably in the minds of a lot of our listeners. So

We constantly are talking about the rise of agentic AI, actually giving some agency to the bots to go and make some of their own decisions. And many people kind of fast forward and say, you know, it's demise of a lot of jobs. It's, you know, we're making them do more and more human like tasks. When Snehal and Dan are back, you're having a version of this conversation and it's 2035, not 2025.

What does the cyber war look like? What's left for the humans and where are agents doing the work and where are humans doing the work? It's a great question. I would actually generalize it to let's just talk about tech jobs or professional jobs in general. I think we're going to see a pretty important bifurcation of careers.

So on the one hand, you've got people with specialized expertise and experiences. And take whether it's you or take some of the work that I do, or even right now as CEO of the company. For me, ChatGPT gives me unlimited interns. It has dramatically increased my capacity to get my job done. It has become an absolute force multiplier for me. That's because I spend 20, 25 years now in industry,

trying to accrue highly specialized skills and experiences. So if you're a senior engineer or a business leader, and you utilize AI and agentic workflows and so on as having unlimited interns at your fingertips, you're going to increase your capacity to have a bigger effect on your business, your company, your job, and

you're going to have more time to go home early. You're going to have more time to solve problems that are really difficult that only humans can go off and solve.

But to do that requires two things. One is you have to be a learn-it-all that is willing to figure out how to utilize AI as unlimited interns and be that force multiplier for you. And some people will naturally do it, but a whole bunch of people don't. So last week or two weeks ago, we had our company kickoff in Austin at Horizon 3. And it was four days or two days of company kickoff and then two days of sales kickoff. So

So the two days of company kickoff, the first day was just us talking through what we're going to go do.

The second day, I blocked off three hours. So think about it. The entire company for three hours completely blocked off. And we did an AI hackathon. And it wasn't about developers. This wasn't coders. We had every single person in the company. I said, you are all struggling. You all have problems to go solve. And we gave examples of how to use ChatGPT and Gemini and Grok and so on to help a sales rep write a better call script.

help an email marketing person write a better email or a software engineer write a better test case. And we said is this technology is going to enable you to completely change how you do your work. And we forced people to come up with ideas and actually got to present and propose and record and talk about all the things they went off and did.

We then had everyone write, here's the problem statement. Here's how they think they can fix it. And here's an example or how they did it if they did it.

And they all dropped in the Slack channel. We then copied and pasted all of those hackathon submissions, dropped it into ChatGPT and said, and summarize all of these into key themes. And we got eight themes out of them with the ball. And then what theme had the most submissions? And it was all around company onboarding or customer onboarding. So what we did was twofold. We crowdsourced an interesting problem to go solve.

And two, we forced people to figure out a way to utilize AI to make their jobs better. So that's kind of one area. The second area, though, is the folks that don't yet have that specialized expertise. The first group to get hit hard, I think, were boot camp web developers.

So these are people that tended to become web developers as a second career. So they were doing something else first. They went to an eight to 12 week boot camp, and they're now entry level front end developers. Unfortunately, that entire job category I think has been destroyed.

Because between Cursor and Replit and other tools, I'm going to get better front-end code for a fraction of the cost right in Go. So that's an example of AI causing job destruction. So now these poor folks just jumped from a first career to a short-lived second career. They're probably in debt from paying for those boot camp classes. Now they've got to go figure out what else to go do.

Some of them have it in them to learn how to utilize coding agents to make their jobs better. A bunch of them are going to struggle with that. And so I think we saw that in front-end web development first. We're starting to see that in law. Junior legal analysts are going to have a hard time moving up the value chain to building that specialized expertise unless they are naturally...

ambitious and focused and demand of themselves to become better at their craft. And there's always a percentage of people that are like that, but a whole bunch that don't. They are not going to push themselves to become experts in a particular area. And I think that's where we're going to see a bifurcation in the workforce. You're going to see a bunch of people that

accrue specialized expertise and become a force multiplier. You're going to see a whole bunch that are going to quickly jump through their second, third, fourth careers and

until they find a job that truly requires interpersonal communication. You can't replace a nurse with an AI entity. You can't replace a plumber or an electrician or a mechanic. So there are all of these jobs, and I think the role of trade schools is going to become incredibly important in the age of AI. I think you're basically going to have two groups of people, those that go to trade school and those capable of mastering and expertise,

utilizing AI as a force multiplier. It's a perfect way to come full circle because I'm thinking back to what you said at the beginning about your dad never really worked a day in his life because he loved what he did. And there's always going to be a need for people who are passionate about doing a thing, solving a problem, whatever it is, and that humanness will never leave.

no matter how much we automate, the junior tasks, the dot, dot, dot, that essence of humanity is waking up in the morning, being excited about solving a problem and doing it with other similarly committed people. So this has been great, Snehal. Man, thanks for hanging out. I just wish we had twice or three times the amount of time, but this is great. Amazing, Dan. Thank you for the time. And I think to summarize what you just said, and I say this internally at Horizon 3,

If you don't wake up every day passionate about becoming an expert at what you do because you love what you do, then let's have a conversation because no one should suffer. You should be working on things you truly love and care deeply about.

Yep, go out there, be a learn-it-all, not a know-it-all, right? As you said. Exactly right, exactly right. Good stuff. Where can the listeners learn more about you and the great work that your team's doing? So one, find me on LinkedIn. I post often about leadership and culture and team and tech and so on and startups and experiences. A lot of my posts are lessons learned because there's a lot of scar tissue from making mistakes. And I very openly talk about those lessons learned. I think it's important to build in the open. My company, horizon3.ai, come check us out.

But LinkedIn is a great way to engage. You send me a message, I'm most likely to reply unless it gets buried between the span that comes in there nowadays. But yeah, LinkedIn is probably the best way to find me.

Good. Well, there you have it. The great Snehal Antani from Horizon 3. And gosh, that is more than all the time we have for this week on AI and the Future of Work. Of course, we're back next week with another fascinating guest. Amazing. Thank you.