Jake Stauch of Serval appears on theCUBE Research with theCUBE hosts to explain Serval’s approach to automating employee support that is native to artificial intelligence, AI and to discuss agent governance. Stauch outlines Serval’s two-agent model, code generation primitives, identity and access considerations and the operational harness required to deploy AI agents safely, and they situate these elements within enterprise deployment patterns.
The discussion addresses real-world IT service management, ITSM use cases and deployment patterns and explores how AI changes IT's role across organizations. It examines use cases across human resources, HR finance legal and operations and considers permissions memory and testing requirements for production systems.
Stauch emphasizes governance and a two-tier agent model as central to limiting risk. They argue that code generation provides extensibility but requires robust testing permissions and memory controls to be reliable in production. theCUBE analysts highlight how AI can convert IT from a backlog-driven support function into forward-deployed builders, unlocking productivity across human resources finance legal and operations.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: Mixture of Experts Series. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: Mixture of Experts Series.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Register For theCUBE + NYSE Wired: Mixture of Experts Series
Please fill out the information below. You will recieve an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for theCUBE + NYSE Wired: Mixture of Experts Series.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: Mixture of Experts Series. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: Mixture of Experts Series.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Sign in to gain access to theCUBE + NYSE Wired: Mixture of Experts Series
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: Mixture of Experts Series. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Jake Stauch, Serval
Jake Stauch of Serval appears on theCUBE Research with theCUBE hosts to explain Serval’s approach to automating employee support that is native to artificial intelligence, AI and to discuss agent governance. Stauch outlines Serval’s two-agent model, code generation primitives, identity and access considerations and the operational harness required to deploy AI agents safely, and they situate these elements within enterprise deployment patterns.
The discussion addresses real-world IT service management, ITSM use cases and deployment patterns and explores how AI changes IT's role across organizations. It examines use cases across human resources, HR finance legal and operations and considers permissions memory and testing requirements for production systems.
Stauch emphasizes governance and a two-tier agent model as central to limiting risk. They argue that code generation provides extensibility but requires robust testing permissions and memory controls to be reliable in production. theCUBE analysts highlight how AI can convert IT from a backlog-driven support function into forward-deployed builders, unlocking productivity across human resources finance legal and operations.
>> Palo Alto studio connection, Silicon Valley and Wall Street. I'm John Furrier, co-host of theCUBE here and Dave Vellante, my co-host. Hello, I'm John Furrier, host of theCUBE out of theCUBE's NYSE studio. Of course, we have our Palo Alto studio connecting Silicon Valley to Wall Street. This is our mixture of experts here. We talk to leaders who are making it happen, who are experts who are building the technology and the innovation to bring in AI to everything that we do. Jake Stauch is here, co-founders here of Serval, company really in the middle of what I call the biggest hype cycle, but also the biggest value opportunity. So unlike other hype cycles, this one has a lot of meat on the bone. Jake, thanks for coming on theCUBE and our Mixture of Expert series, kind of a play on AI, but you're an expert. Thanks for coming on.
Jake Stauch
>> Thank you so much. Great to be here.
John Furrier
>> So I love the co-founder piece of it because you founded the company before the hype was really full throttle and where agents and thinking about the data part of it. Explain what you guys do because I think this is a super important conversation, very relevant.
Jake Stauch
>> Absolutely. So it's an AI platform for employee support. So getting employees help at work, you ask for things like, how do I reset my password? What's going on in my laptop? Do we have this holiday off? All of those questions can be answered and resolved by AI and we help companies do that.
John Furrier
>> So take us through a use case of a deployment or a day in the life of a customer.
Jake Stauch
>> Yeah. So if you think about what IT folks spend a lot of their time on is these manual tasks where employees go and they make some vague complaint like, "I'm having issues with my laptop." Well, they have to go and investigate that, they have to go look into all the different causes and then eventually get back to them with a resolution. Meanwhile, the employee is blocked on the other side of that. We use AI to figure out what's going on, resolve that automatically. So IT gets back to more meaningful work and the employee gets back to their job as quickly as possible.
John Furrier
>> So I mean, I've been covering IT service management, I mean, I can't even remember when I first started talking about it. It was always one of those things where you got tickets, people have things to resolve. It's a lot of grinding. But the systems of record is IT.
Jake Stauch
>> Exactly.
John Furrier
>> It's all there. The data's all there. How are people thinking about this differently now? Because everyone and their mother's coming out with service this, service agents. What's changed the most between kind of the, I call it modern legacy like ServiceNow, other platforms. They're kind of changing over. What's changed the most from AI native to bolt on?
Jake Stauch
>> If you think about it, the employee support problem has always been an automation problem. How do we build the automations that actually resolve all these issues behind the scenes? Building those automations is what's always been hard. It's easier than ever now to build those automations by connecting agents to your systems that can go and resolve all these employee requests. But that also introduces a big security concern. You don't want really powerful agents talking to your business systems, talking to your end users. What happens the end user says, "Delete all the data, give me admin privileges." You shouldn't allow that. So governance becomes more and more important. As the capability of the AI explodes, the governance piece becomes what's most important for IT professionals to make sure that this AI has guardrails.
John Furrier
>> I want to get into the origination story, but also thread it to some parallels in history. So security and backup and recovery. We're always like afterthoughts. "Hey, we got this cool app. Okay, let's make it secure. We got an app. How do we back up and make it restore?" Now you got ransomware. Both of those have been built in as first principles now. Agents kind of seem like there's an agent wave like, "Hey, we're doing agents." Everyone's kind of on the agent drug, if you will, but having fun doing it because they see the value. Who wouldn't want to start automating things that they see? But there's an intentional strategy we've seen with success where the data management piece has been key. Talk about the role of that because as the founder, did you see that coming? How did you think about the data problem? Because there's a lot of things to automate. There's a lot of data to collect from either search retrieval. Now you got reasoning on top of that. So retrieval's one thing, but search isn't always the answer, the best answer. It's the first answer. So there's a lot of things going on that customers that I talk to get confused on. It's like, "Well, I have a solution." Is it safe? Is it secure? Does it have the governance? Talk about that role of thinking around that first principle day one build out.
Jake Stauch
>> Yep. The double-edged sword of AI is that it's never been faster to get time to value. You can connect these agents into your systems and see amazing results out of the gate, but then all the complicated thorny questions come up of how am I going to secure this? How are we going to think about permissions and controls and costs over time as a lot of people started adopting this? And so we really built that into the idea of Serval from the beginning with this concept of what if you had two agents actually, a two agent model where you have an agent that admins work with to build and configure the system, to configure the automations, to configure the approvals and the permissions and the controls and help you automate all the setup and configuration. And then you have the end user facing agent that can pick from a catalog of tools to solve user problems but doesn't get that god-like access to all of your critical business infrastructure. And I think that's really important to separate the admin control agent from the end user facing agent. We think that's the right approach. And we see other companies starting to adopt this kind of two-tiered approach to get agents in front of their users that are very powerful but still controlled with an enterprise -
John Furrier
>> So you basically peak the agents on certain things and have them talk to each other to resolve it.
Jake Stauch
>> Yeah. You have one agent that builds tools for the other agent to resolve. That means that the end user is never talking to an agent that actually can delete the users, reset passwords, do things that the admins don't want it to be able to do.
John Furrier
>> Yeah. And under the covers there, just to kind of put the scope out there, there's a lot of complications. The complexity involves identity, access management. I mean, these are things that are normal IT things, but now they're out in the wild.
Jake Stauch
>> Exactly.
John Furrier
>> How are customers getting their arms around this? Take us through that.
Jake Stauch
>> Yeah, it's really challenging. As identity explodes as a problem, we are just starting to get our arms around the idea of identity and really tech forward companies were implementing tools for identity governance and who has access to what applications and roles. Now you introduce agents into the mix and agents are starting to outnumber employees at many of these organizations. And one, you have to pay attention to governance just as much, really more so because an agent can do a lot more damage in a short period of time than any human's able to do. And so we see a lot of the principles from identity governance going beyond just human identities to agent identities and securing those and protecting access for those. But I think a lot of the same technologies work just as well for agents. They just become more important than they work for.
John Furrier
>> Yeah. I mean, let's talk about the business. You guys had great success. You went from zero revenue to over $1 billion valuation in 18 months, great run, validation for sure. And this is the beginning of the journey, a lot more room to grow. How did that happen? What was the key to success for you guys?
Jake Stauch
>> I think there are a lot of keys, but I think the fundamental thing is we listen to customers. We sat down with the IT buyer, the leader, and we said, "What can we solve for you?" More importantly, we said, "If you were going to hire a bunch of people today, what would you ask them to do for you? Because we are going to deploy AI to do that for you instead." And so we spent all this time working with IT and building this understanding, this empathy for their problems and then we just built the technology to solve the problems. And then we built that in this iterative feedback loop so that during that process, it seems like you're not making a lot of progress, no revenue, no customers, but one day all of a sudden it starts to click and people start to buy. And that's how the business accelerated so quickly is because we invested all this time and effort understanding the customer so that when we really shipped a solution, it was exactly what they'd been looking for all along.
John Furrier
>> So you did the work.
Jake Stauch
>> We did the work.
John Furrier
>> You guys went down into the trenches.
Jake Stauch
>> That's right.
John Furrier
>> All right. Talk about the concept of forward deployed engineer in context to, I'm just going to say it, forward deployed IT person, because I love that question. If you had to hire people, everyone can rationalize that. I never have enough RECs funded in IT. So if you deploy IT with agents, in a sense, you're forward deploying the intellectual capital of IT because agents are essentially workers.
Jake Stauch
>> Yep.
John Furrier
>> What's the scene like in the IT community around this? Because it's an opportunity because now they can extend their superpower or their knowledge base to have more of those workers.
Jake Stauch
>> Yeah, absolutely. IT has always been kind of limited by these resources and there's always been a bigger backlog than they've been able to tackle. That's why so much of IT work ended up being IT support. Even though most folks didn't go into IT to do IT support because those problems are urgent, those become the bulk of what you spend your time on. But if you can clear those out with AI, all of a sudden IT gets to work on these other problems across the organization. So we're seeing customers actually deploy their IT teams into different parts of the organization to be these like forward deployed engineers within the organization. Let me figure out how to deploy AI to the legal team, to the HR team, to the finance team. And so IT gets to be builders and Serval is a technology that enables that behind the scenes.
John Furrier
>> Yeah. I mean, basically you're deploying IT to everybody. You're basically replicating and creating a digital twin of IT with intelligence. It's funny you mentioned the psychology. I would agree with you that a lot of IT folks that are friends of mine all kind of say that, "I hired, this new job. I'm going to be the lead IT engineer for the cloud native, blah, blah, blah, agile infrastructure, all this." Selling the dream, looks good on paper. Then they end up doing support work. They're grinding, doing all this heavy lifting, like drudgery support, which is critical because that's the productivity angle that they're supposed to be.
Jake Stauch
>> The entire organization depends on them. So it's really important work.
John Furrier
>> They're supposed to make everyone productive, that's the role of IT. How has the psychology changed? Give me some examples of customers that have literally flipped the script on that kind of capability.
Jake Stauch
>> Well, we transform IT from this support function into a builder function. So instead of providing services, support services for the last organization, they are building tools for the rest of the organization. Reusable tools, automations, onboarding flows, off-boarding flows, reporting flows so that the whole organization gets enabled. So IT is not a blocker. IT is not someone who you're waiting on to finish some task for you. IT is a resource IT is the person who's going to build that really cool automation for your department. They're going to give you access to AI and it's enabling them to be builders, which is fulfilling for IT, but then it's unblocking the whole org-
John Furrier
>> They're happier.
Jake Stauch
>> The whole company's happier.
John Furrier
>> They're happier. Okay. So this is the role of IT. There's an old book that was written, Does IT Matter? Dave and I always talk about this book all the time. It's kind of an old data center, old school IT book. And it was basically making the argument, do we need IT? And I think here with the productivity gains, it's key. If someone's watching this, what do they need to do or what signs in their organizations lend itself to being ripe for accelerating some of this innovation? What are some things that jump out? Backlog, give some examples.
Jake Stauch
>> Here's a quick test. Go and ask for help at work in your IT department, HR department, finance department, how long does it take you to get what you need? Because it should be automatic. It should be instantaneous. If it's not, then there's a problem. There's a problem that AI can probably solve.
John Furrier
>> All right. Talk about the founding story. Were you sitting around having a couple beers saying, Hey, I'm going to go do this. Did you come in from a specific angle? What was the motivation? What was the origination?
Jake Stauch
>> Yeah, I was working at a company called Verkada. I was working in product and I was spending a lot of time with IT leaders as part of that business asking them, "What else could we build for you? What kind of products we could build for you?" And completely unrelated to the company I was at, I heard these complaints about, "Get me out of the help desk, help me with onboarding, help me with off-boarding, help me build all these automations I have in my head." And we thought there's got to be an alternative to what they're doing today because they were stuck in these drag and drop workflows and these really complicated build patterns. And realized we could actually build the technology that enables IT to build out all these automations, deploy AI to the organization. And so it just made sense.
John Furrier
>> So what happened next? You say, "Okay, I got to start a company." Get some funding.
Jake Stauch
>> I did. I was a founder before that. I knew I was going to be a founder. And so we took this kind of half baked idea to investors and said, "We think we've got something here." And it turns out ServiceNow has proved this is a pretty massive market. And so we think the time is now, the timing could not be better. AI is enabling this disruption and we're the team to do it.
John Furrier
>> Who were some of the funders? Can you share the names?
Jake Stauch
>> Yeah. First Round Capital and General Catalyst, they were the amazing early believers, alongside Box Group and some other fantastic angels that were the first ones to no revenue, no idea, no team, just me and my co-founders saying, "I think we can beat one of the biggest software companies of all time." And they said, "I believe you can too." And they were kind enough to believe in us.
John Furrier
>> Yeah. So what happens next? Because one of the things that I've observed in the service side is that once you nail that, it's pretty sticky. You get loyalty, you have access to the people who are now building. What's next for you? Because when you get builders building and IT builds-
Jake Stauch
>> IT builds....
John Furrier
>> that's a good trend. What are they doing? What's next for you? Is there a sequence to a broader market opportunity? What do you see? What's your vision?
Jake Stauch
>> Absolutely. Because it spreads across the organization, IT is an entry point because they hold the keys to a lot of these business systems. But once you work with IT, you start working automatically with HR, with finance, with legal, with operations teams, with sales and marketing. And the average customer of Serval is deploying Serval in 13 different departments. So it goes across the organization. It becomes an enterprise automation platform for supporting all employees all the time. And so the end result here is people are automating boring, tedious work and people are getting back to the work they enjoy. So it unlocks a much broader opportunity outside of ITSM. Yes, we do all the ITSM software, but we extend it across enterprise and then we do the underlying labor to get people back to the work that actually is meaningful.
John Furrier
>> What's funny, we've been doing theCUBE, it's our 17th year interviewing people. Dave Vellante and I were talking on a CUBE pod like a couple months ago and I said, "Dave, we now have a new category called modern legacy." Because we always say legacy systems. And modern legacy by our definition is basically early cloud. So I mean, go back to 2015. So the question is that you're coming in fresh, you're tackling, got beachhead, got a great solution. What's the big change from the modern legacy view that customers are dealing with? Because enterprises have legacy. They have opportunities to abstract away complexities with what you're doing. What's the big change from a execution standpoint, technology, architecture? What's your observation perspective on that?
Jake Stauch
>> Absolutely. I think one of the biggest shifts is this idea of code generation, AI code generation as a fundamental primitive. I think the future of software is increasingly going to be building AI generated code in real time and using that to accomplish different functions. And so building an application around code gen is very different than the legacy way of doing things where maybe you have a proprietary programming language or a query language built in your product and that's how you customize things or you have some kind of complicated no code infrastructure to build and customize. That's how the legacy of early cloud adopters did things. But now you can build an entire software application that includes code generation as part of its fundamental primitive, which means that it's infinitely extensible. Which means that every idea that a customer has for an integration, for an automation, for an extension becomes possible because you can generate the underlying code to power those capabilities.
John Furrier
>> And you mean because they don't have to take it offline or have a meeting the code is generated. How does someone do that? That's a question I get a lot. How do I build an AI native platform or tool or solution with code generation in mind as a first principle?
Jake Stauch
>> Yeah. So code gen, what's great is the models are taking care of a lot of the difficult parts of the actual generating the code. Now the harness becomes much more complicated of how do you make that reliable? How do you make sure that's durable, that it's tested, that it's safe? That ends up being a lot of what we do to make sure that companies can trust the code that's being generated can build in the approvals, the permission checks, the governance. All of that is actually the hard part. The generation of the code is often handled by the underlying models, but a lot of the long tail stuff is all the things that make that complex.
John Furrier
>> The code gen is easy once it's baked and verified through either governance, hygiene or policies.
Jake Stauch
>> Yeah. I won't say it's easy, but it's straightforward.
John Furrier
>> It's a lot of work, but you get a payoff.
Jake Stauch
>> Simple, not easy. Yeah.
John Furrier
>> It's a payoff.
Jake Stauch
>> It's a payoff.
John Furrier
>> If you do the work, then you have confidence and reliability.
Jake Stauch
>> Exactly. And then you've got a platform where people can build all the things they've always dreamed of building. That onboarding workflow that's 120 steps, you can just describe that natural language and the AI can build that full system for you, and that's just magical.
John Furrier
>> What's the hard thing about the harnesses right now? That comes up a lot. When you said harness, that's the tricky part, what do you mean by that?
Jake Stauch
>> Yeah. So it's all the infrastructure around the underlying models to get these to work in production applications is how you can think about this. And I think there's a lot of hard problems in the harness. I think one of the increasingly relevant things that people are thinking about right now is the concept of memory. We want the AI agents in some cases to know a little bit about the previous things you've asked, context about your organization and then in other cases you want it to kind of start from a clean slate. Don't use all that information. That's not relevant anymore. And so the boundary of what we keep as memory or keep as context in the system is a very challenging problem to solve. And because it's not clear cut when you should retain information, what is relevant from previous sessions, what needs to be retained, but you also don't want to start from a clean slate every time.
John Furrier
>> That's where reinforced learning could come in. You know the use cases that might need memory, some that don't. How do you handle those use cases?
Jake Stauch
>> Yeah, right now there's a human in the loop that often says, "Hey, we have this suggestion that we want to make to store this information, this skill that we think is relevant for future work in the platform." Now we'd like that to be more automatic, but it turns out that people have different preferences on what they want the AI to retain and what they want the AI to kind of forget about and not reference in the future.
John Furrier
>> Well, Jake, great to have you on. I love this AI native code gen in mind because I think that's a first principle. And again, like I said, we've seen other ways where things get bolted on old, not because they weren't thinking about it, just because, "Oh, I've got to add that in." You're bringing it in from day one. What's next for you guys? What are you optimizing for? What's your focus? Next round of funding on the horizon. How many rounds have you done?
Jake Stauch
>> We've done a seed, a Series A and Series B, so three rounds of funding. We love our investor partners. I'm sure there will be future rounds of funding. We're growing at an incredible clip. So we're super excited to expand the team. Growth and hiring is our big focus. We've got a massive market to go after and we need a lot of talented people to make it happen.
John Furrier
>> And you got some great backers, First Round Capital. Josh's team there for 20 years doing great job. General Catalyst, world-class.
Jake Stauch
>> Sequoia at our most recent round.
John Furrier
>> Sequoia came in. Yeah, you got pedigree.
Jake Stauch
>> A great point as well.
John Furrier
>> Got some nice pedigree.
Jake Stauch
>> That's a great pedigree. It's great people to have in your table-
John Furrier
>> A lot of pressure. A lot of pressure....
Jake Stauch
>> in your corner. I don't think so. I think these are all supporters. They're all rooting for us and helping us behind the scenes.
John Furrier
>> Yeah. And they're smart. What are you focused on for the rest of the year, team-wise?
Jake Stauch
>> Building the team. Yeah, we're going to probably three exit team from where it is today. We've already four exit from the beginning of the year. So incredible hiring growth while also keeping the talent bar very high. It's very challenging, but it is what I have to spend most of my time on.
John Furrier
>> And key areas, Bay Area? All over the world?
Jake Stauch
>> We are hiring in the Bay Area in San Francisco, as well as in New York. We have an office here. It's growing very quickly. Probably open our first international office towards the end of the year. Yeah, hiring across all roles, sales-
John Furrier
>> All roles....
Jake Stauch
>> engineering, deployment, you name it.
John Furrier
>> All right, cool. Congratulations. Great story. Thanks for coming on theCUBE.
Jake Stauch
>> Thank you so much.
John Furrier
>> All right. I'm John Furrier. Host of theCUBE. Mixture of Expert series here, the NYSE Wired programming community. Thanks for watching.
>> Palo Alto studio connection, Silicon Valley and Wall Street. I'm John Furrier, co-host of theCUBE here and Dave Vellante, my co-host. Hello, I'm John Furrier, host of theCUBE out of theCUBE's NYSE studio. Of course, we have our Palo Alto studio connecting Silicon Valley to Wall Street. This is our mixture of experts here. We talk to leaders who are making it happen, who are experts who are building the technology and the innovation to bring in AI to everything that we do. Jake Stauch is here, co-founders here of Serval, company really in the middle of what I call the biggest hype cycle, but also the biggest value opportunity. So unlike other hype cycles, this one has a lot of meat on the bone. Jake, thanks for coming on theCUBE and our Mixture of Expert series, kind of a play on AI, but you're an expert. Thanks for coming on.
Jake Stauch
>> Thank you so much. Great to be here.
John Furrier
>> So I love the co-founder piece of it because you founded the company before the hype was really full throttle and where agents and thinking about the data part of it. Explain what you guys do because I think this is a super important conversation, very relevant.
Jake Stauch
>> Absolutely. So it's an AI platform for employee support. So getting employees help at work, you ask for things like, how do I reset my password? What's going on in my laptop? Do we have this holiday off? All of those questions can be answered and resolved by AI and we help companies do that.
John Furrier
>> So take us through a use case of a deployment or a day in the life of a customer.
Jake Stauch
>> Yeah. So if you think about what IT folks spend a lot of their time on is these manual tasks where employees go and they make some vague complaint like, "I'm having issues with my laptop." Well, they have to go and investigate that, they have to go look into all the different causes and then eventually get back to them with a resolution. Meanwhile, the employee is blocked on the other side of that. We use AI to figure out what's going on, resolve that automatically. So IT gets back to more meaningful work and the employee gets back to their job as quickly as possible.
John Furrier
>> So I mean, I've been covering IT service management, I mean, I can't even remember when I first started talking about it. It was always one of those things where you got tickets, people have things to resolve. It's a lot of grinding. But the systems of record is IT.
Jake Stauch
>> Exactly.
John Furrier
>> It's all there. The data's all there. How are people thinking about this differently now? Because everyone and their mother's coming out with service this, service agents. What's changed the most between kind of the, I call it modern legacy like ServiceNow, other platforms. They're kind of changing over. What's changed the most from AI native to bolt on?
Jake Stauch
>> If you think about it, the employee support problem has always been an automation problem. How do we build the automations that actually resolve all these issues behind the scenes? Building those automations is what's always been hard. It's easier than ever now to build those automations by connecting agents to your systems that can go and resolve all these employee requests. But that also introduces a big security concern. You don't want really powerful agents talking to your business systems, talking to your end users. What happens the end user says, "Delete all the data, give me admin privileges." You shouldn't allow that. So governance becomes more and more important. As the capability of the AI explodes, the governance piece becomes what's most important for IT professionals to make sure that this AI has guardrails.
John Furrier
>> I want to get into the origination story, but also thread it to some parallels in history. So security and backup and recovery. We're always like afterthoughts. "Hey, we got this cool app. Okay, let's make it secure. We got an app. How do we back up and make it restore?" Now you got ransomware. Both of those have been built in as first principles now. Agents kind of seem like there's an agent wave like, "Hey, we're doing agents." Everyone's kind of on the agent drug, if you will, but having fun doing it because they see the value. Who wouldn't want to start automating things that they see? But there's an intentional strategy we've seen with success where the data management piece has been key. Talk about the role of that because as the founder, did you see that coming? How did you think about the data problem? Because there's a lot of things to automate. There's a lot of data to collect from either search retrieval. Now you got reasoning on top of that. So retrieval's one thing, but search isn't always the answer, the best answer. It's the first answer. So there's a lot of things going on that customers that I talk to get confused on. It's like, "Well, I have a solution." Is it safe? Is it secure? Does it have the governance? Talk about that role of thinking around that first principle day one build out.
Jake Stauch
>> Yep. The double-edged sword of AI is that it's never been faster to get time to value. You can connect these agents into your systems and see amazing results out of the gate, but then all the complicated thorny questions come up of how am I going to secure this? How are we going to think about permissions and controls and costs over time as a lot of people started adopting this? And so we really built that into the idea of Serval from the beginning with this concept of what if you had two agents actually, a two agent model where you have an agent that admins work with to build and configure the system, to configure the automations, to configure the approvals and the permissions and the controls and help you automate all the setup and configuration. And then you have the end user facing agent that can pick from a catalog of tools to solve user problems but doesn't get that god-like access to all of your critical business infrastructure. And I think that's really important to separate the admin control agent from the end user facing agent. We think that's the right approach. And we see other companies starting to adopt this kind of two-tiered approach to get agents in front of their users that are very powerful but still controlled with an enterprise -
John Furrier
>> So you basically peak the agents on certain things and have them talk to each other to resolve it.
Jake Stauch
>> Yeah. You have one agent that builds tools for the other agent to resolve. That means that the end user is never talking to an agent that actually can delete the users, reset passwords, do things that the admins don't want it to be able to do.
John Furrier
>> Yeah. And under the covers there, just to kind of put the scope out there, there's a lot of complications. The complexity involves identity, access management. I mean, these are things that are normal IT things, but now they're out in the wild.
Jake Stauch
>> Exactly.
John Furrier
>> How are customers getting their arms around this? Take us through that.
Jake Stauch
>> Yeah, it's really challenging. As identity explodes as a problem, we are just starting to get our arms around the idea of identity and really tech forward companies were implementing tools for identity governance and who has access to what applications and roles. Now you introduce agents into the mix and agents are starting to outnumber employees at many of these organizations. And one, you have to pay attention to governance just as much, really more so because an agent can do a lot more damage in a short period of time than any human's able to do. And so we see a lot of the principles from identity governance going beyond just human identities to agent identities and securing those and protecting access for those. But I think a lot of the same technologies work just as well for agents. They just become more important than they work for.
John Furrier
>> Yeah. I mean, let's talk about the business. You guys had great success. You went from zero revenue to over $1 billion valuation in 18 months, great run, validation for sure. And this is the beginning of the journey, a lot more room to grow. How did that happen? What was the key to success for you guys?
Jake Stauch
>> I think there are a lot of keys, but I think the fundamental thing is we listen to customers. We sat down with the IT buyer, the leader, and we said, "What can we solve for you?" More importantly, we said, "If you were going to hire a bunch of people today, what would you ask them to do for you? Because we are going to deploy AI to do that for you instead." And so we spent all this time working with IT and building this understanding, this empathy for their problems and then we just built the technology to solve the problems. And then we built that in this iterative feedback loop so that during that process, it seems like you're not making a lot of progress, no revenue, no customers, but one day all of a sudden it starts to click and people start to buy. And that's how the business accelerated so quickly is because we invested all this time and effort understanding the customer so that when we really shipped a solution, it was exactly what they'd been looking for all along.
John Furrier
>> So you did the work.
Jake Stauch
>> We did the work.
John Furrier
>> You guys went down into the trenches.
Jake Stauch
>> That's right.
John Furrier
>> All right. Talk about the concept of forward deployed engineer in context to, I'm just going to say it, forward deployed IT person, because I love that question. If you had to hire people, everyone can rationalize that. I never have enough RECs funded in IT. So if you deploy IT with agents, in a sense, you're forward deploying the intellectual capital of IT because agents are essentially workers.
Jake Stauch
>> Yep.
John Furrier
>> What's the scene like in the IT community around this? Because it's an opportunity because now they can extend their superpower or their knowledge base to have more of those workers.
Jake Stauch
>> Yeah, absolutely. IT has always been kind of limited by these resources and there's always been a bigger backlog than they've been able to tackle. That's why so much of IT work ended up being IT support. Even though most folks didn't go into IT to do IT support because those problems are urgent, those become the bulk of what you spend your time on. But if you can clear those out with AI, all of a sudden IT gets to work on these other problems across the organization. So we're seeing customers actually deploy their IT teams into different parts of the organization to be these like forward deployed engineers within the organization. Let me figure out how to deploy AI to the legal team, to the HR team, to the finance team. And so IT gets to be builders and Serval is a technology that enables that behind the scenes.
John Furrier
>> Yeah. I mean, basically you're deploying IT to everybody. You're basically replicating and creating a digital twin of IT with intelligence. It's funny you mentioned the psychology. I would agree with you that a lot of IT folks that are friends of mine all kind of say that, "I hired, this new job. I'm going to be the lead IT engineer for the cloud native, blah, blah, blah, agile infrastructure, all this." Selling the dream, looks good on paper. Then they end up doing support work. They're grinding, doing all this heavy lifting, like drudgery support, which is critical because that's the productivity angle that they're supposed to be.
Jake Stauch
>> The entire organization depends on them. So it's really important work.
John Furrier
>> They're supposed to make everyone productive, that's the role of IT. How has the psychology changed? Give me some examples of customers that have literally flipped the script on that kind of capability.
Jake Stauch
>> Well, we transform IT from this support function into a builder function. So instead of providing services, support services for the last organization, they are building tools for the rest of the organization. Reusable tools, automations, onboarding flows, off-boarding flows, reporting flows so that the whole organization gets enabled. So IT is not a blocker. IT is not someone who you're waiting on to finish some task for you. IT is a resource IT is the person who's going to build that really cool automation for your department. They're going to give you access to AI and it's enabling them to be builders, which is fulfilling for IT, but then it's unblocking the whole org-
John Furrier
>> They're happier.
Jake Stauch
>> The whole company's happier.
John Furrier
>> They're happier. Okay. So this is the role of IT. There's an old book that was written, Does IT Matter? Dave and I always talk about this book all the time. It's kind of an old data center, old school IT book. And it was basically making the argument, do we need IT? And I think here with the productivity gains, it's key. If someone's watching this, what do they need to do or what signs in their organizations lend itself to being ripe for accelerating some of this innovation? What are some things that jump out? Backlog, give some examples.
Jake Stauch
>> Here's a quick test. Go and ask for help at work in your IT department, HR department, finance department, how long does it take you to get what you need? Because it should be automatic. It should be instantaneous. If it's not, then there's a problem. There's a problem that AI can probably solve.
John Furrier
>> All right. Talk about the founding story. Were you sitting around having a couple beers saying, Hey, I'm going to go do this. Did you come in from a specific angle? What was the motivation? What was the origination?
Jake Stauch
>> Yeah, I was working at a company called Verkada. I was working in product and I was spending a lot of time with IT leaders as part of that business asking them, "What else could we build for you? What kind of products we could build for you?" And completely unrelated to the company I was at, I heard these complaints about, "Get me out of the help desk, help me with onboarding, help me with off-boarding, help me build all these automations I have in my head." And we thought there's got to be an alternative to what they're doing today because they were stuck in these drag and drop workflows and these really complicated build patterns. And realized we could actually build the technology that enables IT to build out all these automations, deploy AI to the organization. And so it just made sense.
John Furrier
>> So what happened next? You say, "Okay, I got to start a company." Get some funding.
Jake Stauch
>> I did. I was a founder before that. I knew I was going to be a founder. And so we took this kind of half baked idea to investors and said, "We think we've got something here." And it turns out ServiceNow has proved this is a pretty massive market. And so we think the time is now, the timing could not be better. AI is enabling this disruption and we're the team to do it.
John Furrier
>> Who were some of the funders? Can you share the names?
Jake Stauch
>> Yeah. First Round Capital and General Catalyst, they were the amazing early believers, alongside Box Group and some other fantastic angels that were the first ones to no revenue, no idea, no team, just me and my co-founders saying, "I think we can beat one of the biggest software companies of all time." And they said, "I believe you can too." And they were kind enough to believe in us.
John Furrier
>> Yeah. So what happens next? Because one of the things that I've observed in the service side is that once you nail that, it's pretty sticky. You get loyalty, you have access to the people who are now building. What's next for you? Because when you get builders building and IT builds-
Jake Stauch
>> IT builds....
John Furrier
>> that's a good trend. What are they doing? What's next for you? Is there a sequence to a broader market opportunity? What do you see? What's your vision?
Jake Stauch
>> Absolutely. Because it spreads across the organization, IT is an entry point because they hold the keys to a lot of these business systems. But once you work with IT, you start working automatically with HR, with finance, with legal, with operations teams, with sales and marketing. And the average customer of Serval is deploying Serval in 13 different departments. So it goes across the organization. It becomes an enterprise automation platform for supporting all employees all the time. And so the end result here is people are automating boring, tedious work and people are getting back to the work they enjoy. So it unlocks a much broader opportunity outside of ITSM. Yes, we do all the ITSM software, but we extend it across enterprise and then we do the underlying labor to get people back to the work that actually is meaningful.
John Furrier
>> What's funny, we've been doing theCUBE, it's our 17th year interviewing people. Dave Vellante and I were talking on a CUBE pod like a couple months ago and I said, "Dave, we now have a new category called modern legacy." Because we always say legacy systems. And modern legacy by our definition is basically early cloud. So I mean, go back to 2015. So the question is that you're coming in fresh, you're tackling, got beachhead, got a great solution. What's the big change from the modern legacy view that customers are dealing with? Because enterprises have legacy. They have opportunities to abstract away complexities with what you're doing. What's the big change from a execution standpoint, technology, architecture? What's your observation perspective on that?
Jake Stauch
>> Absolutely. I think one of the biggest shifts is this idea of code generation, AI code generation as a fundamental primitive. I think the future of software is increasingly going to be building AI generated code in real time and using that to accomplish different functions. And so building an application around code gen is very different than the legacy way of doing things where maybe you have a proprietary programming language or a query language built in your product and that's how you customize things or you have some kind of complicated no code infrastructure to build and customize. That's how the legacy of early cloud adopters did things. But now you can build an entire software application that includes code generation as part of its fundamental primitive, which means that it's infinitely extensible. Which means that every idea that a customer has for an integration, for an automation, for an extension becomes possible because you can generate the underlying code to power those capabilities.
John Furrier
>> And you mean because they don't have to take it offline or have a meeting the code is generated. How does someone do that? That's a question I get a lot. How do I build an AI native platform or tool or solution with code generation in mind as a first principle?
Jake Stauch
>> Yeah. So code gen, what's great is the models are taking care of a lot of the difficult parts of the actual generating the code. Now the harness becomes much more complicated of how do you make that reliable? How do you make sure that's durable, that it's tested, that it's safe? That ends up being a lot of what we do to make sure that companies can trust the code that's being generated can build in the approvals, the permission checks, the governance. All of that is actually the hard part. The generation of the code is often handled by the underlying models, but a lot of the long tail stuff is all the things that make that complex.
John Furrier
>> The code gen is easy once it's baked and verified through either governance, hygiene or policies.
Jake Stauch
>> Yeah. I won't say it's easy, but it's straightforward.
John Furrier
>> It's a lot of work, but you get a payoff.
Jake Stauch
>> Simple, not easy. Yeah.
John Furrier
>> It's a payoff.
Jake Stauch
>> It's a payoff.
John Furrier
>> If you do the work, then you have confidence and reliability.
Jake Stauch
>> Exactly. And then you've got a platform where people can build all the things they've always dreamed of building. That onboarding workflow that's 120 steps, you can just describe that natural language and the AI can build that full system for you, and that's just magical.
John Furrier
>> What's the hard thing about the harnesses right now? That comes up a lot. When you said harness, that's the tricky part, what do you mean by that?
Jake Stauch
>> Yeah. So it's all the infrastructure around the underlying models to get these to work in production applications is how you can think about this. And I think there's a lot of hard problems in the harness. I think one of the increasingly relevant things that people are thinking about right now is the concept of memory. We want the AI agents in some cases to know a little bit about the previous things you've asked, context about your organization and then in other cases you want it to kind of start from a clean slate. Don't use all that information. That's not relevant anymore. And so the boundary of what we keep as memory or keep as context in the system is a very challenging problem to solve. And because it's not clear cut when you should retain information, what is relevant from previous sessions, what needs to be retained, but you also don't want to start from a clean slate every time.
John Furrier
>> That's where reinforced learning could come in. You know the use cases that might need memory, some that don't. How do you handle those use cases?
Jake Stauch
>> Yeah, right now there's a human in the loop that often says, "Hey, we have this suggestion that we want to make to store this information, this skill that we think is relevant for future work in the platform." Now we'd like that to be more automatic, but it turns out that people have different preferences on what they want the AI to retain and what they want the AI to kind of forget about and not reference in the future.
John Furrier
>> Well, Jake, great to have you on. I love this AI native code gen in mind because I think that's a first principle. And again, like I said, we've seen other ways where things get bolted on old, not because they weren't thinking about it, just because, "Oh, I've got to add that in." You're bringing it in from day one. What's next for you guys? What are you optimizing for? What's your focus? Next round of funding on the horizon. How many rounds have you done?
Jake Stauch
>> We've done a seed, a Series A and Series B, so three rounds of funding. We love our investor partners. I'm sure there will be future rounds of funding. We're growing at an incredible clip. So we're super excited to expand the team. Growth and hiring is our big focus. We've got a massive market to go after and we need a lot of talented people to make it happen.
John Furrier
>> And you got some great backers, First Round Capital. Josh's team there for 20 years doing great job. General Catalyst, world-class.
Jake Stauch
>> Sequoia at our most recent round.
John Furrier
>> Sequoia came in. Yeah, you got pedigree.
Jake Stauch
>> A great point as well.
John Furrier
>> Got some nice pedigree.
Jake Stauch
>> That's a great pedigree. It's great people to have in your table-
John Furrier
>> A lot of pressure. A lot of pressure....
Jake Stauch
>> in your corner. I don't think so. I think these are all supporters. They're all rooting for us and helping us behind the scenes.
John Furrier
>> Yeah. And they're smart. What are you focused on for the rest of the year, team-wise?
Jake Stauch
>> Building the team. Yeah, we're going to probably three exit team from where it is today. We've already four exit from the beginning of the year. So incredible hiring growth while also keeping the talent bar very high. It's very challenging, but it is what I have to spend most of my time on.
John Furrier
>> And key areas, Bay Area? All over the world?
Jake Stauch
>> We are hiring in the Bay Area in San Francisco, as well as in New York. We have an office here. It's growing very quickly. Probably open our first international office towards the end of the year. Yeah, hiring across all roles, sales-
John Furrier
>> All roles....
Jake Stauch
>> engineering, deployment, you name it.
John Furrier
>> All right, cool. Congratulations. Great story. Thanks for coming on theCUBE.
Jake Stauch
>> Thank you so much.
John Furrier
>> All right. I'm John Furrier. Host of theCUBE. Mixture of Expert series here, the NYSE Wired programming community. Thanks for watching.
John Furrier