← The Peel with Turner Novak
The Peel with Turner Novak
Inside Serval: Building the System of Intelligence for IT | Jake Stauch
01:25:20
Loading summary
Transcript277 lines
- [00:03]
A
Jake, welcome to the show.
- [00:04]
B
Thank you so much. Glad to be here.
- [00:06]
A
Yeah, I'm excited. I know we're going to talk a lot of fun stuff. Can you really quick, for people who don't know Serval, really quick, just give us an explanation of what it is?
- [00:14]
B
Yeah. Basically it's an AI platform for employee support. So I'm sure folks are familiar with all these cool AI tools for customer support. We support employees internally. So when you go in, you ask questions internally. You need help maybe resetting a password, getting access to an application. A lot of times that's it related. We often start with IT teams but more broadly we can support internal questions around, hey, I need to make this update to my benefits or I need an employment verification letter or I need this expense approved or send an NDA to a customer. So all the kind of internal questions. We have an AI platform to automate those resolutions and we make it very easy for the admins on the other side of that to build those automations.
- [00:51]
A
When you say you make it easy, like what is it like today? Like if I'm not using Servol, if I'm using a market tool that's out there, it's not easy.
- [00:59]
B
Yeah. The traditional way of internal support, what's called in the category like enterprise service management or IT service management, is you make your request, you often file a ticket or you make a request in a chat platform and then it creates a ticket for you. That ticket sits in somebody's queue, eventually it gets assigned to a human, that human might reassign it to somebody else, tag somebody, and it just becomes this, this ticket that moves through the system and eventually hopefully you get a resolution. But meanwhile you're kind of just stuck around waiting. Our ideal experience is you ask for help, you get help immediately. All that's automated.
- [01:33]
A
And why, why like, wasn't that a thing? Because that seems pretty straightforward. Like, I don't know, like if you
- [01:39]
B
want to make employees support, great, you have to automate the requests. Yeah, the challenge is building those automations. So there's all these great automation tools that are out there. Obviously there's, there's these cool drag and drop workflow builders, no end to those kinds of tools. They're everywhere and they are very powerful. But you think about the problem here is you have to go in, you have to say, okay, what triggers this event? Okay, what the rule is, if this, then that drag this, drop that maybe add some custom scripting and so you can build cool stuff, but there's so much friction in building the automation, there's so much friction in maintaining that because things change over time, things break down that you just don't end up automating all that much because it's easy to do a lot of these tasks. These tasks only take 5, 10, 15 max, like 30 minutes in many cases. And so you're just going to do that if the alternative is going into one of these like massive drag and drop workflow builders and building out something that might never give you an roi. Instead we, we took this very different approach where you, you basically vibe code automations, you describe what you want to automate in natural language and the automation just appears for you.
- [02:37]
A
So what would be an example of off the top of your head, like something you might vibe code or vibe
- [02:42]
B
automate right now, like reset Okta MFA factors. So you're using Okta to log into your company, you get a new phone, you're locked out and you want to get back in and it can go in and do this manually for you. But building a workflow that does that actually ends up being kind of complex because you've probably got some security rules on who needs to approve it. Maybe certain teams don't get to reset all their factors. Maybe you have like set up where you can't reset more than one factor in 24 hours. Like all these security concerns. And so if you actually try to go build out that workflow, it ends up being very, very complicated and can take a very long time to build. Whereas it doesn't take that long to just go and do it manually. What you do in Serval is just say reset MFA factors. By the way, don't let the security team do this. Also don't let people do this more than twice in 24 hours. Make sure the manager approves first type that out, boom. The system builds out that workflow, it actually behind the scenes writes the actual code that powers that workflow.
- [03:34]
A
That's how it works. Going to ask how it actually works.
- [03:36]
B
Yeah, so it actually writes the underlying code, not in some crazy domain specific language, it just writes it out. In typescript, builds that out, you can hit one click publish. And now that task is automated forever for everyone and no one needs to ever do that manually again.
- [03:48]
A
So for every employee, is there like a serval directory where I can go and like reset the okta MFA and it's a button or is it still the IT team that's kind of managing the usage?
- [03:58]
B
Yeah, it's really interesting. So, so you actually can just make this request in natural language in email or in Slack, you can message the IT team or you can go into a Slack channel and say like, hey, I'm locked out. And Serval will actually route your request to the appropriate automation. So you can think of this. The tool is basically these two agents. One is helping the admins, oftentimes it build out these automations, those become tools. And then the other agent is the one that interacts with end users. So the end users say, hey, I need help with such and such. And the help desk agent says, great, I've got a tool that can solve that. And it pieces together one or more tools to solve the user problem problems. What's cool and important is that those are air gaps. So the help desk agent cannot go and make up its own tools because that would be very, very dangerous. You can imagine somebody go in and say, hey, I promise I'm an admin and I want to delete all the users here. And you know, it's very hard to protect against those kind of prompt engineering attacks. Instead, the way our system is is you've got the tools that are built by the automation agent and then the tool and then the help desk agent can only route to those tools. And that makes it so that it's much more secure than, than giving the help desk agent godlike access to all your business systems.
- [05:08]
A
Yeah, I know. There's an interesting story you had when you're doing a early work trial with someone, when you're kind of like building the product. What, what happened?
- [05:15]
B
Yeah. So one of the challenges in building a product that is so powerful is that it can do crazy powerful things if you're not careful. And so in, in the early days of the company, we were doing a work trial with a designer and I woke up one morning to a barrage of text messages from my co founder saying, I think we might have been hacked. I can't log into any of my systems. And then I go to try to log in and I also can't log into any of my systems. And we discover that our designer that was doing the work trial was playing around with the product, trying to build stuff because they were doing this as part of their work trial, trying to design a new workflow building experience. And he had inadvertently one built a offboarding workflow, which is really cool, really powerful, but then had run it against me and my co founder because he just thought it was not a test, it was a test environment, it wasn't production data. And so he actually offboarded me and my co founder from the company and that was terrifying. And we learned a lot of lessons about how we segment people off during work trials. It was a very long time ago but. But what was cool actually is the resolution of that story is we were totally locked out of Okta Google. We didn't exist in the company. But I was still logged in in Serval, so I could actually use Serval to bring me back. And that was the only way I was able to like undo all that damage was actually using our own product to get us back into these systems.
- [06:36]
A
Geez. But I guess that gave you like a slap in the face of like this is the worst case scenario you can do with this product.
- [06:42]
B
Yeah. And it makes you think a lot about hey, like let's make sure that one only admin should have this experience. And of course that's true in the product. This was a special case because we just brought someone in for work trial and they're playing around an environment that they obviously should not have been in the production environment internally. And that doesn't happen anymore. But also when you're building the product, you have to think about these product decisions so that the system is more secure by default where you can't just have one click be one click away from some disaster. And so we think about that a lot as we build product of hey, this is so powerful. Let's make sure that people can't make small mistakes that have a very big impact. And one example of that was we actually built a feature that we rolled back because we felt it was too powerful. And so that feature was the way our product works as I described is like you can vibe code, these automations describe what you want to automate and do it. But what happens if somebody comes in with a request and you don't have an automation for it today? It just escalates it as a manual ticket. So though someone comes in and they say, hey, I want to delete this user, offboard them from the system. And we don't have an automation for that necessarily pre built by the IT team. And so what ends up happening is that gets escalated and then somebody handles that manually. We thought, you know, it'd be really cool. Automatically build the automation for them and have it ready to go so that it just needs to say one click yes, run this automation.
- [08:04]
A
This is before anyone's requested. It's like pretty rebuilt.
- [08:07]
B
You're saying it's after the request comes in, but an automation has not been built for it yet.
- [08:11]
A
Oh, so it makes it and then it like approves Exactly.
- [08:14]
B
It approves it and it runs it automatically. So it says, hey, this user asked for this other person to be off boarded. Do you want to approve and run that automation? And you could one click and approve it. And it worked and it was amazing. But when you thought about it, man, we are one click away from somebody saying something like hey, delete our entire AWS account, delete the company, delete the company, like Opera and our system would go and generate that workflow and then you're like kind of one click away from disaster. Now in practice you would not expose the API scopes that make that possible. There's all these protections before that would happen. Yeah, but we were very uncomfortable the idea that like you're one click away. So we rolled that back and instead made it so that we would suggest new automations to you automatically. But then you'd review and approve those automations and then those would be used in future requests. So you're not going to just like automatically hit a button and then have something, you know, major happen.
- [09:06]
A
It's interesting that this hasn't really become a thing until now because when you think of an IT professional, it's like somebody who probably just like loves technology, loves just like the computer, loves like Internet and all that stuff. And there's not that many automation tools for them. Like it's just not something that anyone's built yet. Like do you know why we haven't built very good technology for IT professionals?
- [09:30]
B
There actually are a lot of automation tools, but they all look the same. They're all these drag and drop work workflow builders and every product has them and a lot of IT folks do use them, but they end up only using them for these very complicated long running processes because that's the only place they can justify the upfront investment like
- [09:48]
A
an overnight migrate, a server or like
- [09:52]
B
some part of the onboarding process or the operating process, things that take a very long time because then you could potentially realize the gains from that automation. What you don't see a lot of automation for is the smaller and frankly the more common kinds of tasks because there's this questionable payoff. If the task only takes you 10 minutes to complete and the automation takes you a couple weeks to build, you're just never going to be able to prioritize building that automation.
- [10:17]
A
Yeah. One thing I feel like you've mentioned before is that a lot of automation isn't actually automation. What is, what does that mean? Cause it's like an oxymoron.
- [10:27]
B
Yeah. So it's interesting when, when you think about how a lot of these ticketing systems think about automation, the automation will be, hey, we automatically generated a ticket and we automatically assigned it to somebody.
- [10:40]
A
And like that's automated.
- [10:42]
B
Yeah.
- [10:43]
A
And still a lot of work to be done.
- [10:44]
B
Yeah. And from a, from a very legacy perspective, true, those things used to take a lot of time. Someone would have to go and take the user request and turn that into a ticket and fill out all these forms and it is valuable to take a request and automatically generate the ticket. But that is not really the work to be done. And we really think about automation as no, the user's problem is solved, not We've taken some steps in the software platform to automate pieces of the software journey.
- [11:09]
C
This episode is brought to you by numerl. Numerl is the fastest, easiest way to
- [11:14]
A
stay compliant with US Sales tax and global vat. It's easy to set up and they automatically handle all registrations, ongoing filings, and their API provides sales tax rates wherever you need them with all the integrations you need. Numero sports over 2,000 customers in both the US and globally and they pride themselves on White Glove High touch customer service. Plus they guarantee their work and they'll cover the difference if they mess anything up. They're fresh off a Fundraise closing a $35 million Series B from Mayfield, which they're going to reinvest into building an even better product. If you want to put your sales tax on a AutoPilot, check out numerl
- [11:48]
C
at their new domain numeral.com that's n u m e r a l dot
- [11:54]
A
com for the end to end platform
- [11:56]
C
for sales tax and VAT compliance. This episode is brought to you by Flex. It's the AI native private bank for business owners. I use Flex personally and I love it because they use AI to underwrite the cash flow of your business, giving
- [12:11]
A
you a real credit line.
- [12:13]
C
The best part is 60 days afloat, double the industry standard. Flex has all the features you'd expect from a modern financial platform like unlimited cards, expense management, bill pay that syncs with your credit line and their new consumer card, flexelite Flex Elite is a brand new ramp like experience for your personal life. A credit card with points, premium perks, concierge services, personal banking, cars and expense management for your family net worth tracking across public and private assets and a whole lot more fully integrated with your business spend. One card for your businesses, one card for your personal life. One card for everything. To skip the waitlist, head to Flex 1 and use My Code Turner to get an additional 100,000 points worth $1,000 after spending your first $10,000 with Flex Leap, that's Flex 1 and Code Turner for $1,000 on your first $10,000 of spending. Thank you, Flex. And now let's jump in.
- [13:08]
A
So an interesting thing that you mentioned is like you, you're basically building the code. Like are you, did you have to go out and train your own models
- [13:18]
B
to do all this?
- [13:19]
A
Like, because when you look at these companies like Cursor, Kodak, et cetera, like pretty big undertakings, obviously they work pretty well. People use them. Are you, is there like APIs you can plug into that helps you do this? Did you have to build this all yourself? Like, how did that go?
- [13:34]
B
We, we ended up building the entire agentic framework all ourselves internally. Now the models, we use off the shelf models, a lot of, of infrastructure around that to make that really reliable. We found that the models, especially in the, the latest generation just perform so well that we haven't, we thought we'd need to, to train our own model or fine tune an existing model that ended up not being necessary. And we're, we're less and less confident that's going to be an approach that we take in the future. We'll see what happens. Hard part of that is because the problem set is quite constrained. If you think about what we're doing versus say a lovable or more generic Vibe coding platform, we are building IT or enterprise service management automations. Generally. These are procedural scripts that are going and hitting a set of API endpoints with nicely documented public APIs. And so yes, our system ingests all these public APIs. We have a good understanding of how these APIs work. We have examples of good automations that have been built. And the IT automations you want are generally do this, then do this, try this. And that kind of script or program is built very reliably by AI Today.
- [14:41]
A
It's all just rules based. Follow what's out there.
- [14:44]
B
It's, it's rules based. There's, there's in the training set right in the, in the corpora that, that these tools are trained on. There's just a lot of examples of this kind of code and there's a lot less ambiguity about how you'd build these kinds of scripts versus using a vibe coding platform to build a SaaS application, which is, which is challenging and has a lot of architecture considerations. Using a Vibe coding platform to take a set of actions across a few different APIs, that's very reliable. And everything we built to make that more and more reliable over time is really improve the product.
- [15:19]
A
Yeah, and it's, it's kind of an interesting market or space because you might not think as an outsider, you might be like, it software like that can't be that big of a market. But I think there's like some publicly traded companies in the space. They're like some of the largest software companies. Like, is it. How big is the market? Do you know?
- [15:38]
B
The market is massive. And I had the same kind of impression outside in before we got deep in the space of is there a big enough market for this? Because I knew from my time at Verkada how big of a problem automation and IT service management was for our customers. But I didn't really understand the breadth and depth of this market. Obviously ServiceNow, one of the largest SaaS companies of all time. And this is predominantly what they do. And, you know, their stock price has taken a hit, so this is out of date. But they were a $230 billion company last year and one of the biggest publicly traded SaaS companies in history. They've taken a bit of a dive this year. Maybe related to Turbo, maybe not. We'll see.
- [16:19]
A
But just a coincidence, A coincidence.
- [16:21]
B
You know, we announced our Series B and I think they lost 40 of their share price. But, you know, like there's, there's, you know, many things at play.
- [16:27]
A
Who knows, like Wall street bets, traders. These guys are going bankrupt.
- [16:31]
B
Exactly. They thought they saw our video and they just.
- [16:33]
A
The launch of hundreds of billions. So.
- [16:39]
B
But yeah, it's a massive space. And even smaller players in the space like JIRA Service Management have a great business and ITSM Fresh Service has a great business in itsm. And so the category is quite massive. I mean, depending on the estimates, 12, 15 billion or more, just in the IT service management software sector. I think what's interesting about Serval is we're not limited to the ITSM or Enterprise Risk Management space, the software space. We're actually going after the labor here because, yes, there are, you know, 10 billion plus tens of billions spent on the software, but there are hundreds of billions spent on IT support and internal employee support. And that's really the space that we see us making a dent in is actually automating the work, not just building a better itsm.
- [17:25]
A
So then I feel like some people take that question like, oh, you're like, replacing jobs. Like people are losing jobs because of Serval. Is that true necessarily?
- [17:33]
B
Or we're not seeing that we are replacing Work, but not jobs. And we are not seeing actual job reduction or job loss or rifts. What we're seeing is one, we're enabling it to be this kind of internal automation center for the organization. We're empowering really talented, smart IT folks to go and build workflows and automations for the broader company. And that's so cool to see them actually have tools where they can show off their capabilities. If you think about like you work at IT support, how do you show that you're amazing and you're better than the person working next to you in IT support? If most of what you do are these manual kind of click ops type work operations.
- [18:13]
A
Yeah. But if your value is like I reset passwords, like that's kind of, it's
- [18:17]
B
hard to show off. It's hard to be better than your peer at resetting passwords. But if you have a tool like Serval, you become a builder. Like you're, you're actually able to build these really creative, complex workflows for onboarding, offboarding compliance reporting, help desk. And so you really are enabling folks to show their value. And yes, you are replacing and eliminating a lot of the drudgery of the work, but you're also enabling higher order kinds of work and making these people more impactful and more important and vital to the organization because they become the experts on these tools and they build automations for the broader organization. So we're really excited about that trend and it's been so cool to see our customers be elevated in their companies because they bring in a tool like Serval and they become like the hero because they fixed all these problems at the company and we're not bogged down by just a barrage of tickets.
- [19:05]
A
Yeah, it's interesting. I had, I think this is going to come up before I publish this episode with Gary Tan. Why? See what we talk about. So firewood used to be 25% of US GDP. Did you know this?
- [19:16]
B
No.
- [19:16]
A
Like that's like an insane stat.
- [19:18]
B
Yeah.
- [19:19]
A
And it was literally in like the 1700s and even in, I think in the 1830s it was 26.3, 26.3% of US GDP was the sale of firewood. So it's like today, I mean, it's like I don't even think it exists on like it's like 0.1% or whatever.
- [19:39]
B
Yeah.
- [19:39]
A
But it's just so fascinating how like the market does change. Like the economy does and will change. Yeah. And it's based on technology.
- [19:46]
B
Like you just add. I saw that was really Interesting is there are four times more people working in IT support than in software engineering. Really? And so when you think about the impact of these tools for software engineers, there's four times more people working in IT support. And that sounds crazy if you live in the kind of the tech world, but then you kind of think about, okay, think about retail and restaurants and manufacturing and healthcare. A whole lot of IT support folks running around, not a lot of software engineers. And so it makes sense that IT support is just a much, much bigger part of the economy. And it's been growing so quickly because everyone becomes an IT worker eventually. You know, if you think about, you know, in the past a lot of people did not interact with technology during the course of their day. Now there are very few jobs which don't interact with technology in some way, shape or form. And that's what's really created the boom in IT service management and internal IT supports.
- [20:40]
A
And it's true. It's like if you are operating in the economy and it's 2026 now, holy shit. If you're doing a job like participating in the economy right now, like you kind of have to use a computer
- [20:52]
B
no matter what the job is. You're using a computer, you're using a phone. At the very least you're interacting with technology in some way that's vital to your job.
- [20:58]
A
Yeah. Even even if you're like a truck driver. So like the truck is technology, but also. Yeah. You might be using a phone, GPS coordinates. Yeah, Sinking. Yeah, logging. Sinking without a drop off in the pickup. And then eventually I was talking to my wife about this last night we, she rode into Waymo for the first time and she kind of like the initial, her initial reaction was like, this is weird. Also kind of cool. But then also, oh man. Like what happens all people who are drivers. Well, you're like a truck like that, you still probably have a person in the truck, they just don't have to drive it.
- [21:30]
B
Yeah.
- [21:30]
A
Like a matt, like that's hard. 18 hour trip across the country driving a super heavy dangerous vehicle versus like the computer kind of does it for you. It makes your job a lot easier.
- [21:41]
B
Yeah. And there's so much work to do. I mean I think that that's what it comes down to is that there's a backlog of jobs to be done and work to do that we're just not doing because we're stuck with all this other work. And it's true in software engineering where we're saying like, yes, even when you bring in these AI tools. Turns out there's a huge backlog of demand for software engineering. Same is true in pretty much every category where there's so much more we would be doing if we had the resources. And so even though we're automating a ton of the work, there's still a lot behind that that needs to be done. And we're enabling and a lot of times that's more interesting work and that's what we're enabling people to focus on.
- [22:16]
A
Yeah, because it's like imagine if, you know, a decent chunk of your day is just, you know, sending people a link to, you know, reset their password or sending or, or spending a bunch of time helping people spin up their operating environment, like the new laptop, you know, just that might get kind of old. It might be fun when you first start, but if you're doing it for 20 years, might be kind of cool to not spend as much time on that.
- [22:40]
B
Smart people, I mean, as you said, these, these are people that have been interested in computers and technology a lot of times their whole lives. They got in this field because they find this stuff really fascinating. And then the actual job, the reality of the job does not require their level of intelligence and expertise. And we like to say, like, we unlock the most meaningful parts of the job, we unlock meaningful work instead of just focusing. And there's, there's always this gap Right. Between the idealized version of what a job is and what the job actually is. And we think a lot of the difference between those two worlds is the kind of work that AI and tools like Serval are automating away.
- [23:14]
A
Yeah. And that I think I want to talk about you, how you kind of came up with the idea you were working at a company called Verkada, but even before that you were working at or you had a startup you were in, I think. Did you go to Duke?
- [23:31]
B
Yeah.
- [23:31]
A
Yep. Okay, so just tell me the story about your very first startup.
- [23:34]
B
Yeah, I went to Duke. I worked in a neuroscience lab doing eeg. So reading brain waves non invasively with sensors, electrodes on the, on the surface of the scalp. Yep. And my first job in college, work study, job was, my job was to look at the screen while people were in the lab doing some kind of test on the computer and look at their brainwaves and see when they're about to fall asleep, because I could see that in their brain activity and then offer them a coffee or Coke because if they fell asleep, I ruined the experiment.
- [24:03]
A
That's actually pretty cool.
- [24:04]
B
So you can do that Study job as an 18 year old is like, cool, I'll watch the TV, I'll see like, oh, I see they're about to drift off. And I thought it was just so cool that I can look at the screen and I can tell whether or not this person is paying attention based on their brainwaves. And that became initially an idea that I had around. Oh, we could test advertising with this. Because at that time a lot of ad testing was not done digitally. It was done in these in person focus groups. So you're about to spend millions of dollars on an ad campaign distribution of that, and the way you test that is bringing a collection of 20, 30 people into a room and asking them what they thought about the ad.
- [24:40]
A
Yeah, crazy.
- [24:41]
B
That's what we did. Yeah, that was kind of state of the art at the time. And the state of the art was also like, you could get a dial that could, like you could while you're watching the ad, turn it up or down on whether or not you liked it or not. And so the initial idea was, hey, let's go and test advertising with brain scans and see when people are paying attention to the ad and what other kind of brain responses we can look at. And that might be a more interesting, more objective way of evaluating how these advertisements are doing. Started that business. Did, did quite well. Actually worked with a lot of amazing brands that were very excited about this. Very clear though, I could see the writing on the wall that this space was moving more and more towards digital ad testing where like, yes, we could bring these people together in a focus group and spend tens of thousands of dollars to test this single ad. Or we could just launch it in a digital format and get very quick feedback on what was working and what was not. And so that was becoming obvious. And I had this conversation with one of our customers that said, you know what I'd actually like to do? Actually like to take this home, put it on my kid who has adhd and see when they're paying attention to their homework and when their attention drops off. And I think that'd be really interesting to me. And we took that idea and we're like, I think this actually is really interesting. And what if we could give them kids feedback on when they're paying attention and when they're not, Maybe even give them a game to play or the more they focused, the more, you know, the faster their dragon flew or the further ahead in a tunnel they could see. And so you tied it to game mechanics. So we ended up pinning the company Building this company called Neuroplus that made brain controlled video games for kids with ADHD and other attention problems. So brain controlled video games to train attention skills. And it was really cool and it worked really well. We did clinical, randomized clinical trials found it was more effective than medications in helping kids improve attention skills.
- [26:18]
A
Why is that?
- [26:19]
B
Attention is a skill like you can practice paying attention, just like you can practice anything else. The problem with practicing that is you don't have feedback. You don't, you don't know when you're doing a good job. And so by giving you real time feedback on how well you're paying attention, you can, you can train yourself to get better and better. Now it's, it basically is diet and exercise for the brain. So yes, it is very effective, but it also requires a commitment and a lot of hard work and a lot of, a lot of effort. And it's very hard to compete with medication in the space. And we learned this the hard way. Seven years I spent building this business, you know, had great early traction. A lot of superfan customers, a lot of amazing families that we helped, but was never a rocket ship company and never had that kind of rocket ship trajectory. And just kind of slowly, slowly grew very, very slowly over time and always felt like we were one step away from really unlocking something. But it never really happened. I think fundamentally because the product market fit was not there. Our market was, you know, we thought our market was everyone who takes Adderall, everyone who has ADHD, which is, you know, increasingly 10 to 20% of the population depending on age group. And it turns out our market is actually the families that had their kids on these medications that had really severe negative reaction. These medications do have a long tail of side effects that include loss of appetite and trouble sleeping and in severe cases like seizures and other issues. And so those families loved our product, but that ended up being a very, very tiny percentage of the market. And most families, these drugs are safer and effective. Um, they're, they're often free or very inexpensive. They take zero time. And so it's hard to compete against that when, you know, what we were offering took a lot more time. Money, energy, effort.
- [28:00]
A
So was it like a cap that they wore?
- [28:03]
B
Yeah, we actually built the hardware. So we built a consumer EEG device that measured their brainwaves and then we also built the software, which is a set of video games they could play that responded their brain.
- [28:14]
A
Brain. They were at their brain or they have a controller.
- [28:16]
B
No, with their brain. So the more they focused, the More they were paying attention, the faster their dragon would fly, for example. Or we had one where like it would put wind in their sails as they sailed the pirate ship.
- [28:27]
A
So they just had to pay attention to the game and to like, yeah,
- [28:30]
B
wind blowing the sails. Basically they had to activate the parts of their brain that are associated with attention. So we're basically linking, there's, there's these correlations of, you know, when your brain is active in a certain way, the brain activity looks a little bit different than if it's, you know, if you're zoning out, if you're kind of like falling asleep. Those kinds of trainings. Interesting. And so we use that, but then we also use other mechanics. So we'd also look at your muscle tension to see if you're like tense. So we'd make sure that you were relaxing as well. And then also we tracked your body movement to make sure you're sitting still. So you basically had to relax, sit still, focus, and play this game. And I think the coolest part of the experience was I got to go to hundreds of families, homes and watch kids play this and help them get set up. And the feedback system and the game mechanics made it so that kids could be like, oh, wow, I can do this if I try. And a lot of the benefit, I think is just convincing people that they can pay attention, they can do these things if they put in the effort. And a lot of times they didn't think that they could.
- [29:30]
A
And so you mentioned you were doing this for seven years. You also told me earlier that there's no point in the company where you had more than three months of Runway.
- [29:38]
B
Yeah.
- [29:38]
A
And this is a hardware company. Like, I'm just.
- [29:41]
B
Hardware company. Yeah.
- [29:42]
A
Yeah. So how did that go? Just financing the business and like making it work for 10 years.
- [29:47]
B
I. It was a very hard company to fundraise for. I mean, one, I had no experience. I started in, I started in college. I dropped out of college to run the company. So, you know, I did not have a track record of success.
- [30:00]
A
And you were in North Carolina.
- [30:02]
B
I was in North Carolina. I was not in like a fundraising environment that was, that was really frothy or Right. You know, a great place to start a company for sure. I loved it there. I love Durham, love the triangle. But not the most lucrative funding raising environment compared to say, the Bay Area. I was also in consumer, I was also in hardware, I was also in healthcare. And we were not doing the FDA path. So it was kind of like this, no man's land of this is just not a category that anyone was really excited about. There's not really a fund that those thesis is like non FDA consumer healthcare products that require hardware and also game development. It was, it was like take all the things that scare off investors and like that was basically what we're doing. And then of course like we didn't have like crazy, you know, traction early on. It was like just absolute grind. And so investors, rightfully so, like were skeptical of, of the business and very hard to fundraise. I did put together several rounds of funding but the way those funding rounds often came together was they were tranche rounds where we, we'd raise like a little bit at a time and it'd be like unlocked with milestones or it was like kind of rolling closes where we couldn't put all the money together. So we'd raise 50k and then 100k and then 50k. And so what that meant in practice is for that seven year period, I really had never had more than three months of Runway in the bank to pay our employees. Which made it an incredibly stressful experience.
- [31:25]
A
Yeah.
- [31:26]
B
But also incredible learning experience. It's basically an inoculation against future stress and panic. When you just live in that world for so long, it definitely makes you more resilient to that kind of environment in the future. And luckily I've not had to have that same experience at Serval.
- [31:42]
A
Yeah, I mean it probably just like trains you like. What's the one Bane quote from Batman? Like I grew up in the pain basically like nothing, nothing phases. Yeah, exactly. Like I've seen way worse.
- [31:52]
B
I've seen way worse. And, and there's like nothing that can really faze me and no kind of like downturn that can, that can scare me because I've just seen, I've seen worse.
- [32:01]
A
And then you ended up getting a job at a company called Verkada.
- [32:05]
B
Yeah.
- [32:06]
A
Did you, did you like move to San Francisco and was like I gotta, I gotta figure out my life. Like it was the kind of like a relation type moment.
- [32:14]
B
I was trying to figure out my life. It was very clear that I was not on a rocket ship startup and I'd really come to terms that there was. We did not find product market fit and the path we were on was not getting us to probably market fit. And that was wrong about the market. And that took a lot longer than I'd like to admit to do, you realize? But eventually I did have that realization. And so then it's like, okay, what do I want to do next? I was very lucky to have a friend from college that had joined Verkada very early running marketing. And he actually initially called my wife to see if she would join Verkada,
- [32:47]
A
does she do more marketing type stuff.
- [32:48]
B
So she was a CEO of her own startup and, and had rolled off that startup that start, got acquired and then was at McKinsey doing some marketing stuff. And he said, I need you here at Verkada. Like it's crazy what's happening here, you've gotta come. And she said, no, but you should talk to my husband. And so, yeah, so she passed the phone over to me and I was like, sure, I'll, I'll chat with the team. I was not really interested in security cameras. You know, I, I had this perspective of I make brain controlled video games. Like I do the coolest stuff in the world.
- [33:17]
A
I'm above some bullshit.
- [33:19]
B
I'm not gonna build like, you know, commercial security security cameras. Like that is not interesting at all, but kind of like at the time it's silly, but I was like, as a favor to you, you know what, I'll meet the rest of the team, I'll like hear what they have to say.
- [33:31]
A
Yep.
- [33:31]
B
And then I end up chatting with the CEO Philip and just very impressed by Philip and the company they're building. I fly out to San Francisco to meet more of the team in this process of like the first call, There may be 15 employees. By the time I fly out to meet them, like 90 employees. And and so I, what I saw this, like saw this growth just in real time and I was like, wow, I've been, I've been grinding away for years and years and years. These people in my mind's like basically just started the company. I think it was like two years old or a year and a half at this point. But like in my, it's like, you know, they basically just started and they are just crushing. It's like I have to see what this looks like. I have not felt this kind of product market fit before. I, I got more excited about the category of, you know, cameras are actually pretty interesting, especially when you think about AI and computer vision. Think about like the impact you're having on safety and security technology was, was more interested than I thought. And, and I felt like I had a lot of value to add because I had been building in kind of the intersection of hardware and software for a long time. And so I, I came in as an early product leader initially on the camera side, but then eventually got to run a bunch of new Business units for Verkata and such an incredible learning experience.
- [34:41]
A
Yeah, it's a cool product. I had Philip on the show maybe like a year ago. I'll throw a link in the description so people can check it out. But he showed me a demo where basically you have this app and you can type in. If you're a security professional, you can type in red shirt. And just any footage of someone with a red shirt, it just pops it up. You can jump in, watch it, you can type Tesla, you can type whatever you want and it will just identify it in the footage. And there's a lot more stuff that obviously the Verkato product does. I think he has gotten into like the access systems on buildings, like not just the cameras, like all the different security tech and like building technology now. It's pretty interesting how it's evolved.
- [35:19]
B
Yeah. And becoming more and more proactive. So you can detect things that might be problematic or might be threats before they. They mature and evolve and, and prevent crime, prevent like people getting hurt. I think that's, that's really the promise here. And it's so exciting.
- [35:34]
A
And so what all did you learn at Perkata? I think you were there for like four or five years.
- [35:38]
B
Yeah, close to five years, man. It's like hard to imagine. One, I saw what product market fit looks like. I saw the impact you can have when you have a product that people are just buying and that there's actually that fit between what you're selling and what the market wants. I learned how to learn a lot about B2B and enterprise sales. And Verkata is a legendary sales team and got to be a part of a lot of those conversations and the growth of that company. I learned how to grow and build a team at a larger scale. We got to start a lot of kind of zero to one products at Verkada. And the way Verkata thinks about these things is they're very isolated products. We treat new products like new business units. And you're responsible for the revenue of that business unit and you're responsible for building out that. That team and product and what the vision looks like. And it was cool because I got paired with this director, I was director of product, I got paired with the director of engineering counterpart and we got to build these new businesses from zero to one within Verkada and one, built a great relationship. Two, just got to learn a lot about what was working and what wasn't. Launched a lot of things that worked incredibly well and sold incredibly well. Launched a lot of things that didn't and saw the difference between those approaches, those products. And you just got so many repetitions in. Right? You just got these reps in. And one, that relationship ended up like I ended up starting a company, starting serval with that director of engineering counterparts. So we got to basically do a five year trial run of what it was like to work together and kind of start something together. And two, I just got to see so much product that we built and shipped and saw what happened there. And then three, our customer was it this entire time. We are selling into IT teams.
- [37:18]
A
It's like security it, right?
- [37:20]
B
Or it's actually more often just generic it. So what happened in this space and what really helped Verkat as a tailwind is security cameras, access control systems, a lot of the building kind of security infrastructure. As these became network devices that people wanted to access remotely and connect in their network, the ownership of these purchases and usage migrated from more of a facilities team to more of the IT team. Now once you plug into the network, IT at least has a say and eventually becomes the owner.
- [37:51]
A
And so before like there would just be like a VCR, TV tape thing
- [37:55]
B
that a security camera, VCR initially and then it became an NVR, like a DVR, you know, kind of like a TiVo for security cameras that was on site, but it's still an on prem server that stores all the data, all the footage. It's all managed on this device. But you know, eventually you say like, oh, it'd be cool to watch this footage from my home and connect it to the Internet. And when facilities starts like plugging in devices into the network, into the switch, it says, wait a minute, like, let me get involved here. Yeah, and you know, oversimplification, but, but over time it takes on more and more of an ownership of this function. And so what it means for us is I was almost exclusively spending my time with it, basically doing five years of customer discovery with it, figuring out, what do we want to build for you next? What are your pain points? What are your problems? And of course I hear nonstop about the help desk and all the manual tasks that they're stuck with. Kind of like the worst parts of their job. And unfortunately at Verkata, my job is kind of like steer them away from that and talk about like, yeah, what about the things that are drilled into the wall? Like, talk to me about, you know, H Vac and speakers and like all these other potential product extensions that are more related to our overall product portfolio. But the lessons there really stuck with me of like I understand what their pain points are, I understand what their job looks like. And I think the biggest thing that stood out was the automation surface area was practically zero. They weren't automating anything. And that was such a disconnect for me because my perception being in Silicon Valley, seeing all these amazing tools that all these companies were building, it seemed like there's all these amazing IT enterprise automation tools and then that's juxtaposed with going on site with these customers and seeing it up, up close and personal and nothing's automated. Like they're not automating anything. And so what is the gap there? What's causing that disconnect that really kind of haunted me and my co founder and eventually became the inspiration for what became Serval is well, what happens if we solve that automation gap? What happens if we make it very, very easy to build the automation? We take the friction away. You'll actually end up with, with much bigger automation service area.
- [40:05]
A
And I know there's a question that you asked to get really close to that kind of like problem and answer. Do you want me to say the question or do you know what the question is?
- [40:12]
B
Yeah, I know what the question is. It was the most important customer discovery question that we had when we were exploring ideas was if you could hire someone to just sit next to you and do work all day long, what would you have them do? And when I started asking that question, it became very clear what people wanted, which is they want somebody to sit next to them. One build a bunch of automations, try out different automation tools like these tools. They're very excited about them, but they don't, you know, if you're handling tickets all day, when are you going to carve out time to build these things? And so they want somebody to go and build automations for them and then take on all of these kind of repetitive manual tasks, the help desk tasks, all like the kind of the low value, high effort tasks that it is often stuck with. And so it's like very interesting to hear those two things like I want somebody to go build automations for me, I want somebody to take over the help desk. And we believe that those two problems are actually linked, that if you actually had somebody building automations for you that were really good, you'd also make the help desk problem go away too. And so once we understood that this is fundamentally what people want, then it becomes a question like, well how do you deliver it? And that's what led to strval and
- [41:23]
A
the existing kind of Help desk platforms, like were they not picking up on this feedback or like was it like. Because it seems like they weren't solving for it necessarily.
- [41:34]
B
They weren't. I think in a pre AI world it was not an easy problem to solve at all and probably an impossible challenge to solve because they all offered workflow builders that could solve all of these problems. The challenge was not that the tools didn't exist, that you couldn't build workflows to automate all of these things. That has been around for a long time and I think if you're a product leader at these companies you might say well I don't know why they didn't build these automations. Like we gave them all the tools, they could have automated everything. Yeah, but that was the problem is that the tools outpaced the capability of the tools outpaced the time and energy that it had at their disposal to go and implement those tools. And so what you needed to solve was not yet another automation tool, yet another workflow builder. But how do we short circuit the process of building those workflows? How do we not give you automation, automate the automation so that that whole process of building out those workflows happens automatically.
- [42:27]
A
And this wouldn't have really been possible pre all the cogen tools.
- [42:31]
B
Exactly. Because our approach was hey, the only way to make this work and make this work reliably is to take your natural language description of what you want to automate and turn that into automation. We think the best way to do that is actually by writing the underlying code to power that automation in a world where that doesn't exist, very, very hard to make that work reliably.
- [42:49]
A
This is maybe interesting for people who are thinking about startup ideas, not serval related. Like where else do you think this could be applicable? Like creating products for customers by creating code like in a non even adjacent category. Have you if you thought about that much?
- [43:06]
B
I haven't actually. I've been so heads down in server. I've not thought about like the other things we could do. I mean I think obviously maybe like
- [43:11]
A
adjacent categories for serval then where.
- [43:13]
B
Yeah, so we think about a lot, you know we start with IT and security teams but then within a couple weeks of using our product usually they're branching this off into HR people, teams, finance teams, legal operations staff. Because the automations we're building are not limited it of course they actually expand beyond the company.
- [43:29]
A
Yeah, because it's like identity products, it's like security sort of in some cases I can definitely see like the adjacencies and like I don't know, hr customer facing, touching things.
- [43:43]
B
Yeah, exactly. I mean our mission here is build AI platform for employee support, get people back to meaningful work. And so that means yes, a lot of employee support. Is it related? Most of it. Is it related still. But people still have a lot of questions for the people team and a lot of questions for these other departments and I think that that part of the business is that part of the category is growing very quickly as you know these employee support requests continue to mount and people are always going to need support at work and you know we want to build a platform that makes it very easy to build the automations for those other teams, especially the non technical teams that don't have the resources to go and build these, these workflows themselves.
- [44:19]
A
So you had this insight from Verkada. You started building, it was just like off to the races. Everything clicked and worked. Customer signing up left and right. How did it go?
- [44:27]
B
Oh I, how I wish that was the case. So we started building and I think from the very early days there's a couple things we knew were true. One is that the technology was not there yet. So when we tried to build kind of a proof of concept of this kind of natural language to code workflow builder it fundamentally did not work.
- [44:46]
A
So what peer, like what time was this?
- [44:48]
B
This is April, May 2024. So like GPT4, you know going into GPT 4.5 I think is coming out, you know, maybe around that time or maybe a little bit later. But that's kind of like this era, it's pre GPT4.0 but this was like
- [45:06]
A
the summer of cursor right where like Codegen's like first like I guess working
- [45:11]
B
truly it was starting to, it was starting to a lot especially like the autocomplete and like I see what you're doing here and I can like get you the next step but and the, the true autonomous of like hey I want to go do this thing. The reliability of just generating that on the fly was just not quite there. And so if you were really on the rails and you said something simple like reset Google workspace, password and here's the Google Docs, it would do a pretty good job and that would work fairly reliably. So you know you had a set of cases where it would work reliably and that's what gave us confidence like okay, well it works for this set of cases where it doesn't work is as soon as you start to get creative and you kind of go off the rails. That's where it starts to break down. And so we were confident that we were just like one or two model iterations away from this, really starting to click. And so that gave us confidence. We felt like, hey, this works better, honestly than we thought it would. Not as good as it needs to in the future, but like we're not that far off. So we had confidence there. And especially once 4.0 came out, it felt like there were a lot of tools. Then obviously the cloud models started to be much better on the cogen side. But once that started to happen, we got a lot more confidence there. So one is, we knew that would be hard. We decided to do that problem first. You know, our vision from the beginning was no, we're not going to just build an AI layer, we're going to build a full ITSM platform, go after the legacy incumbents, build the full system of record. We felt that was very important from a product strategy, from a business strategy.
- [46:35]
A
Why did you feel that? Because some of the consensus wisdom is like build a small little wedge in like product that works, then you expand from there.
- [46:43]
B
You were like fundamental. I disagree with that. In this area, at least for our business and our customer. We just felt like a lot of the low hanging fruit problems are solved by a handful of point solutions. What you really need to do is build a better platform. And I think part of that just comes from looking at the market, looking at what people spend money on. People spend a lot of money on it. Service management and we want to go straight after those budgets. We don't want to convince somebody to create a new category, a new budget for us. We want to go after their existing spend directly. That was part one on the business strategy, on the product strategy. We just felt like you just cannot build the perfect product experience if you're tied to somebody else's platform. If step one to getting value out of Serval is setting up ServiceNow, setting up Jira, service management and the way that we need it set up, that's just not the ideal product experience. We want to be able to own that end to end experience from day one. And so we knew the product vision was very, very big, very expansive. We decided to start with the hardest problem. We knew we could build a great ticketing system, asset management, all these other things. But the hardest problem was going to be can you get this workflow builder, this AI code gen system to work reliably. And so we spent a lot of time on that. We convinced ourselves pretty early on that like hey, it's pretty close and it's going to be there and it's doing really cool things. And then we started building out the rest of the platform. But because we're going after this category of legacy incumbents, IT service management, there's a couple things that made it really hard in the early days. One is startups do not buy IT service management. Startups don't have IT teams.
- [48:15]
A
Yeah, it's like just the founders like oh you started like just here you go, here's your laptop.
- [48:20]
B
Let me know if you need to help provisioning stuff like go for it. And so you can't sell to a 10 person company or a 30 person company or really even like 100 person company.
- [48:29]
A
You need to sell to like corporations
- [48:31]
B
smallest company that would even have an IT function is going to be at least 100 employees. Probably like 2, 300 employees.
- [48:37]
A
What's the smallest customer you've got?
- [48:39]
B
Around 200 employees. Okay, that's, that's, that's really kind of like the early stage and usually only those customers when there's a high growth trajectory where they know they're about to get much bigger. And so what that means is that the customers you're talking to either don't have this problem because they're too small or are mature enough where they, they need the product to be really good.
- [48:58]
A
Yeah. And it really like test out because that's the thing like sometimes if you got other friend with a early stage company like let's try each other's products.
- [49:05]
B
Sure.
- [49:05]
A
Like give them feedback but you're probably none of that.
- [49:08]
B
Yeah, no IT or security leaders like oh yeah, I'll throw this in front of my users like what's the worst that could happen?
- [49:13]
A
Yeah, people just reset. They like lock this founders out of the company in order for it to
- [49:18]
B
have a lot of value.
- [49:19]
A
Yeah.
- [49:19]
B
It also needs a lot to have a lot of access and you know you need to connect it to like critical business systems. And so it was very challenging one just because the product service area just didn't do all the things. They're not going to replace their existing incumbent software with you when it's missing 80% of what the incumbent software does.
- [49:37]
A
So did you spend a long period of time just building all these different features that would like convert?
- [49:43]
B
Yeah, our vision was that this didn't really work until you actually had the workflow builder combined with the full ticketing platform combined with you know, access management, automating access requests which is a big part of the help desk and all these other capabilities and have like this full unified experience. And so from the beginning we had this vision of this has to be this big product or else there's just no reason for somebody to migrate off their existing tools. But that is really, really challenging because it takes a while to build that. Yeah. And for the first six months or so it's just me, my co founder, you know, he's spending, you know, all this time building. I'm, we're, we're talking to customers nonstop. But there's just no poll because what we're showing them doesn't come close to what it needs to do.
- [50:22]
A
So how long did it take to sign the first dollar of revenue?
- [50:25]
B
Over a year.
- [50:26]
A
Okay.
- [50:26]
B
Over a year of just building and building and building and not, you know, getting nice compliments of like this looks really cool, keep us posted. But nothing that feels like a buying signal or getting close to a purchase. You know, we had design partnerships, you know, building alongside customers, but even the design partnerships, it always felt like we were just so far away from this being an actual tool that they would purchase and replace their. Because again, it's always a rip and replace. There's not really a greenfield opportunity here. If you're a decent sized company, you have a solution in place and we're trying to displace it. So it was challenging.
- [51:02]
A
Were there any signals of like, hey, we're in month 11, no one's paid us for this yet. Like, but we think next month we might get it. Like how did you know that you could keep going?
- [51:12]
B
This is the hardest problem in startups, I feel like is knowing what is a persistence problems. Like you're actually just a lot of hard work and a couple months away from, from unlocking something and where you're, you're just diluting yourself. Which honestly my first company, there's a lot of like self delusion of like thinking that we were just one feature away, one product launch away and if we just built the next thing that we'd unlock some big part of the market. And I think, you know something I heard that I think is very true is, is, you know, good founders can often squeeze product, some degree of product market fit out of ideas that don't really have legs. And so you can very easily have this motivated reasoning that's like, oh no, this is going to work. We just need to keep building and we just need to keep doing stuff. And that could be completely wrong. In this case we were right that we just needed to keep building. But it was really hard to know that in the moment.
- [52:05]
A
So could you dissect the difference between your first and second company? Maybe. And maybe it's pretty obvious now in retrospect.
- [52:12]
B
Yeah, I think in the first company I was very optimistic and so every signal I got that was positive I think just like fueled me and I just kind of like was seeking the positive, just gravitating towards the positive feedback. Like wow, this customer is so excited. This mom who used it with a kid said it changed their life. We need to just keep going, keep going, keep going. And I think in the second company, very jaded, very skeptical. Like the positive feedback basically had no impact on me. I didn't feel it at all. I was like, whatever, I don't believe you. And then the negative feedback and like really gravitate towards that of like, okay, this person doesn't think that this is a good idea. Let's dissect that. Like understand that. And so just having an overall more skeptical view of the market and taking so much more positive feedback for me to feel like we're heading in the right direction.
- [52:59]
A
Yeah, but wouldn't that, you know, you're on month 10 of no customers that are actually like paying you and 10 months of just negative feedback. Wouldn't you say like, you know what, let's go to like this other thing. Let's go to a completely different area.
- [53:14]
B
Yeah. And I think it's like important that it's, it's not the, it's not negative feedback, it's just like no buying feedback. Right. It's like this is really cool guys. Keep me posted. This is awesome. I love it. You know, keep me up to date, send me, add me to your newsletter. You know, he just got these things of this is, this is very far away and it got, it got really tough because yeah, you just don't know if this is ever going to work out. And I remember going on a really long walk with my co founder of. Should we change course? Should we like maybe pivot to something where we could sell it faster? Like maybe it is just like this very thin AI layer that we sell to early stage companies that just like answers users questions in Slack and that's something that we could build and there would probably be like early stage companies that buy it. We thought about that. We're like, well there's not really a market for that. That's not an exciting space to go after. You can't name me a hundred billion dollar company that sells like a slack bot. And it feels like you get stuck in that category for a while, and then you'd be competing against a lot of other companies that we wouldn't really have a great advantage over. That didn't feel a good idea. Do you go like, large enterprise, you say, like, hey, I'm going to shrink the product service area, and maybe I'm going to find a narrow approach that goes for the large enterprise. I just felt like, well, then you're not really going after the real category again. You're kind of like taking a short cut. Just. It just felt like, hey, actually, I think we're right. I think you have to build the full product. You just have to keep building. And so we just kind of talked about it and we looked at all the different alternatives and tried to be really honest and real with ourselves around what the different options were, and then doubled down and said, no, we are on the right path. When we get all the stuff we want into the product, people will start buying it and we will have a business. And sure enough, a few weeks later, we got our first customer and then second customer and third customer. And then it just started to snowball and got to a really exciting place very quickly.
- [55:13]
A
And so the snowballing happened. At what time did it start to snowball? Cause I think you mentioned April of 25. You were still in this, like, April of 25.
- [55:21]
B
I did not know if it was going to work, and it very easily couldn't. And we're still considering, like, should we do something different? Should we, like, make some kind of radical change to the business?
- [55:29]
A
This is nine months ago.
- [55:31]
B
Nine months ago, May, we got our first customer. So that was a good signal, but it was very, very cheap. You know, crazy discounts, like, still not a great deal.
- [55:41]
A
What was the customer or the deal or the size like, you allowed to say?
- [55:45]
B
Yeah, it was actually perplexity.
- [55:47]
A
Oh, nice. Okay.
- [55:48]
B
And they were the first one that actually signed.
- [55:50]
A
And they're on that cusp of like, pretty large. Probably have. They're like a startup, but they have
- [55:56]
B
a lot of employees and growing, like at that point, hundreds of employees inflecting very, very quickly, about to triple. And so, you know, had a need for a product like ours, but, you know, they had an. They had an existing solution in place and we had to rip it out. And so that. That made it so that there's a lot the product had to do.
- [56:12]
A
How do you rip that out? Like, did you have to help them with kind of implementation and all that kind of stuff? Yeah.
- [56:18]
B
And yet, one, we went on site every single week to Meet with them
- [56:23]
A
in person, which sounds like a big deal, but actually some of my. The companies I've invested in that have had like early success with big, tough, kind of like lumpy, hard to build products like this. The going on site for implementation is like extremely important. It's kind of like show up. Yeah. The forward deployed engineer thing that's became come like a meme at this office.
- [56:45]
B
We literally sit around the IT leader's laptop and we just iterate and we get all the things that we're missing and we, we deliver that to the team and we just keep building and building and building and then eventually we get there and. And then it started. And then, you know, we got our second and third customer in June with
- [57:03]
A
these like, bigger contracts.
- [57:04]
B
Like, they started bigger. Like the next contract was like three times bigger than the last one and. And the next one was like twice as big as the one before that. So, you know.
- [57:11]
A
Yeah, it's like pretty quickly you're like, like, oh man. Like someone's finally buying. It's like, oh, wow, that one's way bigger. Holy shit.
- [57:16]
B
Like, yes. And then we had one that came through that didn't even pilot us, that actually bought us at list price without even a pilot. And that was, I think, one of the first silly signals of, oh, no, this is going to work. Yeah, this is going to work.
- [57:28]
A
That's awesome.
- [57:29]
B
And then we.
- [57:30]
A
How did that feel like, for somebody who's maybe in like the 11 months through this or the April of 2025, like internally, how does that feel as a founder?
- [57:37]
B
When you finally get that, it's this massive relief. But then it's also like immediately coupled with the anxiety of like, oh no, we have so much to do. It's like, they just bought this thing. You know, we have a lot to build to make sure they have a fantastic experience. Like, we are certainly not out of the woods just because we have three customers, you know, we, you cannot like, get too excited about that. We got a long, long way to go.
- [58:01]
A
But there's probably that like, extra motivation. Like, I can keep doing this now because it's finally, finally working.
- [58:06]
B
Yes. And then we ended up, you know, because we we got really great logos, we got, you know, five or six great customers and, and so many customers in pipeline at that point. Like we had five or six customers, 20 or so in pilot. And. And that's when we started getting preempted on our Series A because people were starting to take notice. A lot of enthusiasm around the category. I'm on the space.
- [58:29]
A
So do you know how that happened? Were your investors talking to, were downstream investors that were preempting you talking to your existing investors or was it like customers were telling the investors they were
- [58:41]
B
talking to customers so they were doing research on the space? I think this is not a clever idea to go after ServiceNow to go after this. Itsm right. Like you, a lot of people can come up with this idea of oh hey, there's this massive software company like maybe we can build a better version. And so I think this is part of a lot of investors thesis that there's going to be a better version of IT Service management, of enterprise service management built. And so there are always investors interested in what we are doing and they're kind of always on the background kind
- [59:08]
A
of like seeing how if it works maybe we'll talk.
- [59:12]
B
And once they started talking to our customers and hearing feedback, you know, some of which was like feedback they never heard before from customers. Just the level of enthusiasm for like an IT service management product was just off the charts.
- [59:25]
A
Yeah. And that's like the most, it's like a boring product you would think of like why are you excited about this?
- [59:29]
B
And these people are obsessed with it and said basically the worst thing that could happen in their life is this being taken away from them. And so with that enthusiasm, because the TAM is so large here, that was enough signal for a lot of investors because the TAM is insanely massive. You don't have to prove that there's a market here. You have to prove that you built something so special that you can actually make a dent in that market. And that's what we did. Ended up raising our Series A from the wonderful folks at Redpoint. They've spent a lot of time in the ITSM space. They got really excited about what we're doing and had a lot of options for a Series A but, but Redpoint was, was clearly the ones that had understood the space the best and felt like they had the most valued ad. And then it just felt like off to the races. Like at the, at the time we, we got our Series A term sheet, I think we're eight people as a company. And then very quickly started to grow, started to hire the customers, start to snowball. We hire our first sales rep and you know, I think if you zoom out obviously in the, the day to day everything feels chaotic and stressful and you lose deals, you win deals and it feels ups and downs. But then if you zoom out it looks like it was just like a crazy rocket ship Trajectory.
- [60:37]
A
Yeah, I'm sure somebody looks at sort of right now like, oh, it's like always been crushing it. Like I'm sure things are great, there's no problems. Like it's all easy. You just look at like the chart that you probably have. But as we said for the past hour wasn't the easiest journey. Yeah, for sure.
- [60:51]
B
It's like you zoom out, it looks like this crazy up into the right trajectory. You zoom in and you see like the split spikes and the ups and downs and all the things that feel like on a day to day basis that they're not going well. But yeah, if you zoom out, it's been just an incredible story.
- [61:04]
A
And then you said the day you announced a Series A, you got a term sheet for the Series B. Was that again just like people were like, we want to give you money.
- [61:10]
B
Like that was, that was, how did that go? So yeah, we, we delayed the announcement of series A a little bit so we ended up, you know, doing the Series A in, in August.
- [61:17]
A
But I feel like everyone kind of knew like all the investors you are for sure. Like everyone was like, we gotta, you can't keep secrets.
- [61:23]
B
I mean the, the, a lot of league, everyone knows everything it seems like. And so we raised a Series A closed out in August, announced in October. The day we announced the Series A in October though, I get a text from Sequoia saying hey, can you come to our office? I say, no, I'm in Orlando. I'm in an IT conference actually in Orlando. Then they text back and say, where are you staying? And ended up flying out to Orlando from San Francisco, one partner from San Francisco, one partner from London and meet me for dinner with the term sheet. And that's how the series B happened. And you know, the idea was they talked to all of our customers again at this point, many, many more customers than five. They talked to all of our customers. They talked to everyone that I've ever worked with, basically, you know, managers, peers, direct reports. They also talked to all of our competitors, customers and they had this full like picture of the market and they just, they just knew that like we're, we're going to win and so why wait on the sidelines, you know, for us to grow as a company, why not just come in as early as possible?
- [62:29]
A
And then did you, did you consider anyone else or you just like this is a really good offer.
- [62:33]
B
Like, like I initially said no at dinner mostly just because we just fundraised and it felt like, hey, we just raised $50 million Series A oh, my.
- [62:47]
A
Who's 50?
- [62:47]
B
You said it was $50 million Series A at the time. I think Maybe we're like 15 people as a company. Like, I don't have a way to deploy additional capital. It doesn't make sense for us to bring the Series B at this time.
- [62:58]
A
Yeah. So then what did they do to convince you?
- [63:01]
B
One is, I talked to a lot of references of people that have raised money from. From Sequoia. I thought about, you know, the impact it could have, not just the additional capital, but the support of a great new set of investors, great set of partners. And so, you know, as I considered, like, hey, what are the different options? What are the different things? Yes, we could postpone our Series B to, like, you know, where it makes more sense in six months, but there could be a lot of value in doing this now from a support perspective and from a signaling perspective and helping with hiring and customer acquisition and all these things. So I decided to that it made a lot of sense.
- [63:40]
A
And this is about three months later. Was it a good call? Do you feel like you've gotten what all they brought to the table?
- [63:46]
B
Absolutely. Absolutely. Great decision. I don't regret it at all. Support on the customer side, on the candidate side, recruiting side, strategy side, just across the board. And we felt that, honestly from all of our investors. I've been very nervous about taking every single check that we took, really, but you just don't need it. It's a big decision, you know, that these people are going to be on your cap table forever. These are people you're going to have to answer a phone call for the rest of your life, potentially. And so are these people that you really want to basically get into a marriage with. And it's stressful. It's stressful to make that kind of commitment. But every single investment we brought on board, I felt so lucky to have them involved. And we made the right decision every single time, I think.
- [64:35]
A
So when you think about good support from an investor, you know, helping with customer introductions or with hiring, what is the good relationship there look like? Like, are they, like, texting you a LinkedIn profile? Like, my cousin's looking for an internship. Like, can you interview him? Or, like, what is the. What does the good relationship look like for you?
- [64:55]
B
Yeah, I think it changes a little bit based on stage. First round was one of our leads in our seed round alongside General Catalyst. First round is incredible. At seed stage, I mean, there's just no one better. They. They really are so focused on that stage.
- [65:10]
A
They were probably really supportive through the journey of like we don't know if this is going to work exactly.
- [65:15]
B
And this is what they do is they spend time with companies before they become successful companies, before anyone knows what these companies are. And they've seen it all, and they've seen it all at that stage. And so it was so valuable to have first round Bill Trenchard over there as a partner. Constant meetings, constant one on ones. So much support, both like strategic and emotional support of like, hey, this is normal that this takes a while to figure out, but just actually very tactical. Like here's how founder led sales can be structured. You know, they actually hosted a retreat training founders on founder led sales and going through like an entire like four day intensive course.
- [65:51]
A
What was your biggest takeaway from that retreat?
- [65:54]
B
I learned a lot about running a successful pilot because we had tried to start these pilots that ended up devolving into kind of these endless kind of this limbo feature requests and it never kind of ends and it turns into kind of like a design partnership with no kind of line of sight to signing a deal.
- [66:12]
A
So is this just like, oh, we'll do another three months kind of thing? Like how do you fall into that trap?
- [66:18]
B
You meet with them and they're like, hey, what about this, what about that? And then you just kind of keep meeting with them. And it's hard to have a purchasing conversation, especially early on because the product legitimately is not mature enough. Right. And so everything feels kind of like a promise. And so it's hard, especially selling against a legacy incumbent where it's already entrenched. So you're saying, okay guys, well I'll do all these feature requests, but would love for you to rip out your existing system and then put this in. Even though it doesn't do all the things you want it to do, it's a hard conversation to have have.
- [66:49]
A
And so what did it change your approach?
- [66:51]
B
Oh yeah, basically setting up from the beginning. Here's the structure of the pilot, here's how we're going to evaluate. Here are the criteria for whether or not we're going to move forward, making that a structured conversation and then structuring the meetings of the pilot to make sure we're moving towards the evaluation and we're not just kind of like hanging out, meeting every week and hearing more about like what the product could do better. Okay, so that's a very tactical thing. First round made well over a hundred customer intros. So just like again, very tactically intro to the decision maker at ICP customers. Very valuable Recruiting. I mean, they put their resources behind us to help us recruit. They were able to source us several candidates.
- [67:30]
A
This was during that year of. Of no customers converting. And you were still. They're introducing you to people and everyone's bouncing off like, do. Were you able to like keep the relationships and then people like those people you met in that first year, they now have started closing over the past couple months.
- [67:46]
B
Oh yeah, for sure. I mean, it's funny, we. We just restarted a pilot with a customer that we got introduced to before we even incorporated Serval.
- [67:53]
A
Okay.
- [67:53]
B
And you know, just dragged on for a while, ended up like not going anywhere. And now they've come back and, and you know, just kicked off a pilot with us. So we're really excited about going back to that other customers. And what's interesting is the early feedback from these customers was not negative, not like, oh, this doesn't work. It's like, this is awesome. We love where you're going with it. Basically call me when you have a product because again, we're not this net new thing that can just sit on top of your stack and you can plug in and just get value out of. You have to pull out. Especially at this time. Now we've made this migration process much easier. You don't have to rip and replace your system of record, but at the time it was like you had to pull out your full system of record and you had to migrate over to Serval. And that's just a big. That's a big ask, especially where we were as a company.
- [68:39]
A
You changed how you did sort of the replacement, the rip and replace. You mostly didn't have to or they kind of automated. It was. So what exactly did you do?
- [68:48]
B
So one of the challenges in this, this category in the all AI companies face this is are you going to be the system of action layer on top of the system record or are you going to be the system of record? There's advantages to both approaches. We decided from early on we wanted to do both. We wanted to actually be the AI layer on top and the full system record. And in the early days, we kind of really pushed forward to like, hey, you should. You should implement us as a full system of record. Now what we found is, and we knew this would happen, you know, then you're sitting around kind of waiting for them to be ready to replace your system of record. So instead we said, you know what, let's do both really well and we can be the AI layer on top of your existing system of record. We can Sit on top of a ServiceNow or Jira service management. But the way we do that is very unique in that it's not just a simple layer, it actually becomes a mirror. So it's not a layer, but a mirror where we're mirroring the data in both systems so all the tickets that flow through actually show up in both platforms. Okay, why that ends up being important is the day, the moment you decide, hey, could we just use Serval? All the data is already there. There is no migration because you've already been floating everything through the system. And also incentivizes people to think about it in that way because they start working out a Serval and they notice, hey, Serval has all my data here, it has all my configurations. It is a replica of my system of record. Why again are we using these two systems? And so we stopped having the conversation with customers of like, hey, we're going to replace your system record. We come in and say, no, it's fine, we'll just be a layer on top. What we find is every single one of them, once they get implemented, start saying, oh, we should just use Serval for everything. And so that's been a really, really important. And we knew we were going to do that from early on. It's not like we had this complete realization. But once that was actually implemented in the product in a really seamless way, that really helped us to not have a tough conversation early on. But instead say, hey, you can implement this as your full system record, you can implement this as a layer. Doesn't really matter to us. We know where you're going to end up, which is you're going to use the full platform.
- [70:45]
A
So did you build agents that automate like the syncing of data?
- [70:49]
B
Yeah. So we have a really robust syncing system with health checks. Like all the data synced between two systems configurations on is it a one way sync, a two way sync, when does the sync happen? What fields are synced? How do we interpolate things where we don't have a data match? Like do we use AI to like match fields? Like really, really robust. We decided to invest and make that a core part of the platform.
- [71:10]
A
What is the, like the, the timeline? Like, is it like if I type something in Serval half a second later it shows up in that. Okay, so I'm curious, like what, what tool did you guys use to do that? It was like a certain software.
- [71:21]
B
We just built it internally.
- [71:22]
A
Oh, you built it internally?
- [71:23]
B
We built it all internally, yeah.
- [71:24]
A
Okay, interesting. Yeah, I feel like because it's definitely a question that comes up a lot. Is this like, build system of record layer on top?
- [71:33]
B
Yeah.
- [71:33]
A
How do you.
- [71:34]
B
We think the answer is both. And the way you do that is actually build it not as a layer, but as a mirror so that you can ease that transition. And a lot of this is just deciding to build really hard things. Because when you look at this problem, it's like, this is impossible. All these fields, like all the mappings, how do you solve this? This is like so messy. API rate limits, like, you run into so many challenges and a lot of this is just saying, you know what, like, it's really hard and that's why we should do it.
- [72:01]
A
And I know there's a. There's the stat that we both kind of like thought was interesting, that it's something like a Harvard study that said 90% of AI projects fail or something.
- [72:09]
B
I think, yeah, the MIT story, story
- [72:11]
A
there, there, like, do you know why? Why do you think that is? Because it's a massive number. You would think AI is a massive bubble and this whole thing should pop if no one's actually using it. Right?
- [72:19]
B
Yeah, I mean, I think, like, if you take that at face value and I think there are like some concerns on, on how to think about that, that research, but I think if you, if you like assume that that's true, what we find is the technology has advanced to the point where the problem is not the models are not good enough, the problem is not the software is not good enough. The problem is translating the capabilities of the AI of the software platform to match the business processes, or reverse, you could change the business processes to match the capabilities and the way the AI and the software is structured. And so that mismatch between how businesses are run and what the software can do, that's where you see things like for deployed engineering and professional services trying to bridge that gap. Today, that is the hardest problem to solve in these AI implementations today is the actual implementation of. I run onboarding in this way. I've got a 72 page PDF on how onboarding works across the company. Cerval has this amazing AI platform for automating workflows. How do we connect these? How do I get all the cool capabilities of the Servo platform and use that to automate all my onboarding? That's where a lot of the work happens and that's where a lot of this can break down. And so I think what's going to be interesting over the next couple years is you're going to see a Lot more focus, especially from companies like ours, on how do we actually bridge that gap. How do we make it easier to translate the business requirements to the product capabilities without it being this huge implementation project that has to be owned entirely by a forward deployed engineering staff?
- [73:45]
A
I think you might have said this, maybe somebody else said this to me before, but it was like you almost make the implementation part of the product is like, so what have you found that works the best? Are you still navigating it?
- [73:58]
B
We're building some really cool stuff that are basically going to allow you to just tell the system, here's what I want to do, here's how we have things set up. And it can both map to your business processes, but also suggest changes to your business processes that make the overall structure. Because a lot of this is meeting in the middle where you might have a certain flow that makes a lot of sense in a world before AI, but with AI tools you could actually simplify this flow and it'll work a lot better. But a lot of large enterprises also say, you know what, the business process is not going to change, we're going to do it this way. And I need your tool to adapt to that. And I think you have to have a, if you're selling in a large enterprise, you have to have a tool that can adapt to what the, where the customer is and not force them to change how they do things. Though a lot of times those changes can be very helpful in making the system overall more reliable.
- [74:46]
A
But it's kind of one of those like the bigger the ship is, the hardest to steer it. It's like if it's just like 50,000 person company, it's like you gotta, they're
- [74:55]
B
not gonna change how they do things because a vendor came in and said, hey, it would be better if you did it this way. No, you have to come in and say, hey, we can do it exactly the way you want. And then we can talk later about how you can make some improvements.
- [75:06]
A
Yeah. And then so you've raised, I forget actually like the dollar amount I probably should like rechecked this morning where you've raised like it sounds like over a hundred million dollars.
- [75:15]
B
Over 125 million.
- [75:16]
A
125 million. You probably have. The majority of you haven't spent yet, but you're going to hire more people. So how do you think about hiring right now with the phase that you're at?
- [75:27]
B
I think what's really important is obviously we have incredible hiring goals and we went from eight people in August to 40 today, and we've got crazy hiring goals for the rest of the year, and it's not going to slow down. But in the middle of all that, you have to keep the bar incredibly high. You actually have to increase talent density as you hire.
- [75:45]
A
That sounds hard. It's like, founders are probably amazing. How do you hire people better than you?
- [75:49]
B
It's really hard. One is you have to have a vision that's really exciting. I mean, that's why we're going after this massive category. That's why we're not just happy to be an AI layer, a startup that sells some kind of narrow niche, but we're actually going after one of the biggest software categories in the world. And we're going to be much bigger because we're going after the services layer. So it starts with ambition, starts with, you know, building a team that people want to join. It also starts with actually having the talent density to begin with. So the early hires end up being so important because then the next person you want to bring in that's really, really talented, they need to see your team and feel like, oh, yeah, this is. This is the group I want to be a part of. Great people want to work with people that are better than them. Great people don't want to show up and realize, oh, man, I'm going to be, like, maybe the best person here. And that's what we want to build this culture, though, is where every person that joins actually is raising the bar. And one of the ways we do that is we actually make sure when we're evaluating candidates that we think about it in that way of, is this person actually better than the median person at the company today? Whatever the function is, do they actually raise the bar in some meaningful way? Because what often happens in these companies as they scale is you start hiring people that, like, meet the bar, the people that are better maybe than 40% of the people you've hired in that role. And that makes sense in the moment because they hire, they do well in the interviews, and they actually do better or are better than 40%, almost half of the people in the entire company. And so how could you not hire this person? They're better than, like, the people doing the interview. But if you keep hiring at the 40th percentile over and over and over again, well, what happens downward. Yeah, it basically just, like, continues to collapse and the talent density collapses. And you see this happen over and over and over, especially as companies get very large, is that they cannot keep raising the bar. And so they hire people that are good enough, that are better than a lot of the people in place, again, probably better than some of the interviewers, and yet they're not actually better than the median or the average. And so one of the questions we ask as part of the debrief on a candidate is, who are they better than? Because if you can't tell me who they're actually going to be better than
- [77:55]
A
and, and, oh, this is an internal question. I look like you asked the candidate, who are you better than?
- [78:00]
B
Oh, yeah, that would be great, right? Like, make them challenge somebody for their job.
- [78:03]
A
You don't know.
- [78:04]
B
We don't do that. We, we just, like, we do it as a forcing function, obviously, like collapsing somebody onto, like a single dimension of like, good or bad. Like, that's, that's not, you know, it's, it's impossible to do. But I think it's important to think about it in this way of saying, hey, are they actually raising the bar in some way? And maybe it's because they're super high slope, so maybe they're less experienced. They're not going to come in and be as productive as some of the folks on a team. But we believe that within a couple years, they could be one of the best engineers here. Or they, they bring a new skill set that we just don't have on the team and they raise the bar in that way. So I think there's different ways to raise the bar, but I think they have to come in and make us better. And I think the other thing they have to do is, is, are we excited for their start date? This is the best gut check is if they sign today and they've got their start date two weeks, three weeks from now. Are we counting down the days until this person starts or is this just kind of. Yeah, it's great. They're like, start this day. All the best candidates. You are, you are so excited for the day they started because you're going to feel that impact. And so it's really important to raise the bar on these candidates. It's, it's kind of our, our foundational principle is that every candidate raises the bar.
- [79:05]
A
Yeah, that's like a fun, really easy to remember. Just like gut check on a kid. Like, are you excited for them to work for you? Like, I mean, so simple.
- [79:14]
B
Yep.
- [79:14]
A
But there's so many times you're like, man, not really.
- [79:17]
B
Yeah. A lot of times you'll go through the debrief and everyone will pass them. They'll say like, yep, 3. 3 out of 4. 3 out of 4. 3out of 4. We should hire them, but no one's excited. And then you say, you know, you say, are we excited for this person to join? Like, who are they better than at the company? And you'll find very quickly like, oh, actually this is not a bar raiser. This is somebody who's good. And hiring an elite team means saying no to a lot of great people. And that's the hardest part of the job, is that these people come in and they're actually great and there's nothing wrong with them and they did very well, but they don't get a spot.
- [79:50]
A
And one kind of interesting hire, you made those. Maybe at this point it's not that contrarian, but you hired somebody who's one of the earliest employees to work on video production.
- [79:59]
B
Yes.
- [79:59]
A
So talk me through that.
- [80:01]
B
Yeah, it was really interesting. So one is, we knew we were hiring this incredible growth team. I actually hired my coo, it's my wife, and I hired her and she was, she was running growth at Rippling. And we knew we were going to have this incredible growth team at the company. But the growth team is only as good as, like the content you can equip them with. Right. You know, you can't put a bunch of stuff in front of people and expect it to convert if you're not really investing in content and storytelling. We also knew that because we're going after these large enterprises, we really needed to punch above our weight from a, from a brand and content perspective.
- [80:35]
A
Because if you look like a startup,
- [80:36]
B
if we look like a startup, no one's going to buy us because people do not buy from startups in this category. The companies we're competing against are massive, massive companies. Like, we have to punch above our weight and we have to equip our growth team with great content. And so for us it's like, actually we want to bring this in house and we want to have this be a core competency. And so what's funny is I reached out to the best video person I ever worked with, absolute legend. And I just asked like, hey, do you have anyone in your network? Because you're the best I worked with, you have to know people that are like, looking to join an early stage startup. And he reached out and he said, you know what, I might know somebody. And even though I had no intention and no idea that we could actually get our top choice or the person that we're most excited about, I just wanted him to introduce me to somebody. And instead he came to the Office and was so excited about what we're doing and decided that to join us and one of the best hires we made.
- [81:31]
A
Wow, that's awesome. And one last thing you said you focus on a lot is the people that you hire, are they also able to make more hires for you, help you recruit?
- [81:39]
B
Exactly. Yeah. So. So raising the bar is one thing. But then we also think about this from obviously team is the only durable advantage in this era. Like any product advantage insights you have about the market, it's not going to last very long.
- [81:50]
A
Yeah, like I could vibe code a decent version of server.
- [81:53]
B
There's so many amazing smart entrepreneurs out there. There. The funding environment is so rich. Like funding is not a differentiator talent. You know, like talent is the only true long term differentiator because it compounds and the people that use these tools, they just compound and compound and compound. And so we knew that talent is going to be really important. And so when you bring people in, yes, they have to raise the bar from their pure themselves as, as a, as an IC or as a manager. But what we also think about is how did they impact future hiring. So is this person likely to make it easier or harder for us to recruit more people that raise the bar? And so they can do that a couple different ways. So some people are like tactically they come in and they act as recruiters. They come in and they have great networks and they bring people along and everyone wants to follow them. And those. That is a big part of what we think about when we bring people in is like are they somebody that brings people with them, brings high caliber people with them. So that's like at the, at the top level, someone who actually brings the people along. And then there's also like okay, but maybe they're not going to bring a bunch of folks along. Maybe they're more junior. But do they make it more likely that when we have somebody visit, when we have somebody do an on site, do we make it more, do they make it more likely that person joins? Whether because they bring a level of energy, enthusiasm, kindness, are they, are they just so incredibly talented that people want to work with them because of their caliber? So we, we think about people through that lens of do they actually help us recruit either actually by being a recruiter physically or just like because they're here, we're more likely to get the next person that we're really excited about.
- [83:26]
A
Interesting. Yeah, I feel like it's, it's kind of like one of those things. There's like different ways to Be like, maybe they're like a very high profile person where like everyone will see like the news article or the LinkedIn update and be curious. Or it's literally just like they could text five people that they worked with in the past and just bring them all over.
- [83:42]
B
Or they're just so much fun and they bring like this lively energy office. You're like, oh, I can't wait for this person to have lunch with my candidates. Because candidates are going to meet this person and they're gonna decide to join. And the flip side is true, somebody could be very, very good. But we feel like, you know what, like, I don't know if I want this person to have lunch with our candidates. And I don't know if after that lunch experience, the candidate's gonna be more likely to join the company. And so we, we really think about that, that lens when we evaluate people.
- [84:06]
A
Yeah, it's almost like you ever watched the Silicon Valley show. Yeah, no. Guilfoyle, like, he's like super talented here, but like, candidate might be. And he was like, don't work here. Like, this place, place is terrible or something. Like, you don't wanna scare people.
- [84:18]
B
Yeah, exactly.
- [84:19]
A
Well, this was, this was a lot of fun. Thanks for, thanks for doing it.
- [84:22]
B
Thank you so much. This was a blast and I hope
- [84:24]
A
you had a blast.
- [84:25]
C
A quick thanks again to Numero and Flex for supporting this episode. Put your sales tax on autopilot numero.com and upgrade to Flex Elite to get $1,000 on your first card using code turner with the waitlist link in the description. If you enjoy this conversation, please, like, comment, subscribe and share with a friend who should use Serval to automate their help desk. Make sure to check out the back catalog of over 100 episodes with the founders of companies like Robinhood, Mercury and Box. Tune in over the next few weeks for guests that include Chatham Buddha, Gunta at Benchmark Mike and Nikhil at Footwork and Scott Stevenson at Spellbook, the fastest
- [84:58]
A
growing AI company in Canada.
- [85:00]
C
If you don't want to miss any of these, subscribe to my newsletter. The split linked in the description to get each episode plus a transcript emailed directly to your inbox every week. Thanks again for listening. See you next time.