Oct 23, 2025 · 1h 23m · in-depth
Go hard early: How lessons from Verkada shaped this AI-native IT platform | Jake Stauch (Serval)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Jake Stauch, co-founder and CEO of Serval, details how lessons from scaling Verkada shaped his strategy for building an AI-native enterprise IT platform. He explains how natural language code generation, forward-deployed engineering, and uncompromising hiring standards allow startups to disrupt legacy enterprise software giants.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Brett holds 16.6% of the talking time here. How this is scored →
Brett as informed peer 4.6 Guest teaching 6.0 Guest disagreement 1.1 Brett pushing back 2.1
speaking balance: gold is Brett, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 1:00:24 Rejecting conventional product prioritization orthodoxy
Jake rejects standard product management prioritization practices, asserting that excessive prioritization blinds teams to neglected P2 issues that ruin user experience.
Hardest push from Brett ▶ 55:08 Pushing back on chaotic shipping in forward deployed teamsBrett directly challenges Jake on how embedding engineers with customers without product managers avoids devolving into an uncoordinated mess of feature slop.
Biggest teaching moment ▶ 16:10 The psychology of IT problem solvers during discoveryJake educates Brett on why traditional discovery questions fail with IT professionals whose pride makes them describe broken manual processes as solved problems.
Brett holds their own ▶ 15:14 Calling out founder bias in customer interviewsBrett jumps in to articulate the exact trap of early customer development where founders unconsciously steer interviews to manage toward their desired outcome.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Brett as informed peer | Guest teaching | Guest disagreement | Brett pushing back | Why |
|---|---|---|---|---|---|---|
| Targeting Existing Spend with a 10x Better Product | 4 | 5 | 1 | 2 | Brett asks Jake to reflect on core lessons from his time at Verkada. Jake explains the power of targeting existing budgets rather than trying to invent new categories of spend. | |
| The Verkada Demo and Consumer Enterprise Contrasts | 4 | 5 | 1 | 2 | Jake shares the Verkada demo moment of texting a camera feed to an enterprise buyer, explaining how consumer technology expectations bleed into enterprise software. | |
| Going Hard Early and Architecting for Enterprise Compliance | 5 | 6 | 1 | 2 | Jake outlines the philosophy of tackling hard infrastructure problems early, explaining why Serval built enterprise compliance and permissions from day one unlike Verkada. | |
| The Genesis of Serval: From Device Reboots to IT Helpdesk | 3 | 6 | 1 | 1 | Jake details the initial failed concept of rebooting offline cameras as a service and how customer feedback redirected them to helpdesk ticketing. | |
| Reframing Customer Discovery Questions for IT Professionals | 5 | 7 | 1 | 2 | Jake explains why asking IT professionals about their pain points fails due to pride, and shares how reframing the question around who they would hire unlocked real insights. | |
| The Automation Paradox and Eliminating Workflow Friction | 5 | 6 | 1 | 2 | Jake discusses the automation paradox in IT where building workflows manually takes longer than doing tasks once, creating the threshold Serval needed to beat. | |
| Building a Code-Generation Workflow Engine First | 4 | 6 | 1 | 2 | Jake explains why Serval tackled the hardest technical risk first by building a code-generation workflow engine rather than a superficial UI ticketing layer. | |
| Maintaining Conviction and the Workflow Builder Demo Moment | 4 | 5 | 1 | 1 | Jake describes holding conviction during slow early traction and reveals Serval's demo turning point when live natural language generates deployable code. | |
| Disrupting Enterprise Incumbents Through Programmable Configurability | 5 | 6 | 1 | 2 | Jake contrasts startups that fail by building overly simplistic tools against enterprise giants with Serval's approach of offering greater configurability via raw code. | |
| Architectural Scalability Across Mid-Market and Enterprise | 5 | 6 | 1 | 3 | Brett probes how Serval manages both mid-market and enterprise without splitting the product. Jake explains their API-agnostic code architecture handles both effortlessly. | |
| Engineering Velocity, Consistent Roadmaps, and Eliminating Thrash | 4 | 5 | 1 | 2 | Jake attributes their high shipping velocity to maintaining an unyielding roadmap without thrash, combining product intuition with embedded engineering. | |
| Enterprise Sales Dynamics: Top-Down Mandates and IT Builders | 4 | 6 | 1 | 2 | Jake explains enterprise go-to-market dynamics requiring top-down executive mandates alongside empowering bottom-up IT professionals as creative makers. | |
| The Indispensable Human Role in Enterprise AI Workflows | 5 | 6 | 1 | 2 | Jake explains why humans remain essential in translating nuanced business logic and varied company policies into AI workflow guardrails. | |
| Rapid Improvements in Enabling LLM and Code Generation Tech | 4 | 6 | 1 | 2 | Jake breaks down the massive technical strides in foundation model instruction following and code synthesis over the past 18 months. | |
| Building Moats Against Future AI Advances and the IT Vacuum | 5 | 6 | 1 | 3 | Brett raises whether foundational models could become too capable and erode software moats. Jake outlines attacking hard UI and click-ops automation. | |
| Reinventing the Role of Forward Deployed Engineers | 6 | 7 | 1 | 3 | Brett asks why forward deployed engineering is surging now. Jake explains how modern AI configurability replaces traditional implementation consultants. | |
| Caliber Management and Real-Time Customer Delight | 5 | 6 | 2 | 4 | Brett challenges Jake on how a decentralized engineering team avoids shipping chaotic product slop. Jake argues top-tier engineering talent maintains product integrity. | |
| Delaying Product Management and Rescuing P2 Features | 5 | 7 | 3 | 2 | Jake critiques traditional product management for neglecting P2 features and explains why Serval avoids traditional PMs to keep developers shipping direct fixes. | |
| Cultivating Customer Empathy: The Verkada Panic Button Story | 4 | 7 | 1 | 1 | Jake illustrates extreme customer empathy through his Verkada experience visiting payday loan cashiers, which made mobile panic buttons an obvious requirement. | |
| Hiring for Agency, General Intelligence, and Technical Literacy | 4 | 5 | 1 | 2 | Jake discusses prioritizing raw intelligence, high agency, and quantitative fluency over traditional domain experience in hiring. | |
| High-Bar Interviewing and the 'Better Than' Standard | 5 | 6 | 1 | 2 | Jake reveals their internal debrief rule requiring interviewers to name who at the company a candidate is better than to avoid settling for average hires. | |
| The Candidate Recruiting Multiplier and Talent Flywheels | 5 | 5 | 1 | 2 | Jake explains evaluating candidates on their likelihood to attract future talent, creating a flywheel of high-caliber referrals. | |
| Aggressive Go-To-Market Scaling in the AI Land Grab | 5 | 6 | 1 | 2 | Jake describes their decision to build an aggressive go-to-market engine early to capture the limited window of enterprise AI platform selection. | |
| How AI Unlocks Startup Simplicity and Enterprise Customization | 5 | 7 | 1 | 2 | Jake synthesizes lessons from Verkada leaders Philip and Hans, explaining how combining established TAM awareness with first-principles AI unlocks new market leaders. |
Statements from this episode (44)
Stauch: Verkada sold AI software via physical camera budgets
“And Verkata owned the entire platform, and so it was really cool to be able to sell a better camera to people that were going to buy cameras. And then you could also, by nature of selling cameras, also sell them these really cool AI capabilities, this really c…”
Stauch: Displacing enterprise software requires a 10x experience
“In some cases, It's not gonna be enough to just build a nicer, cleaner UI and a faster experience. You have to fundamentally look at what's been missing. And so in my opinion, the experience just has to be 10 times better. It has to be undeniable. It can't jus…”
Stauch: Tackling hard operational challenges early builds defensible platform moats
“Seek out the hard things that are harder for other people to do that unlock a lot of customer value. And then once you've built them, it's so much harder for somebody to come in after you and build those same things because you decided to like go hard early.”
Stauch: Serval prioritized enterprise compliance on day one, unlike Verkada
“Verkata waited several years to support things like SCIM and SSO and SOC II and various compliance frameworks. We did all of those things from day one because we knew that that was going to be a blocker to sell into the large accounts that unlocked the bigger …”
Stauch: Automated device troubleshooting is not a sustained enterprise IT problem
“We took that to a lot of customers. I did customer discovery and nobody cared. It turns out that was a very Verkata customer focused problem. And that was very top of mind when you're talking My customer is doing a big device deployment, but after those device…”
Stauch: Ask prospects what they would delegate to an assistant
“One of the questions I started asking that I thought was really interesting is if you could hire somebody today to just sit next to you and do work for you, what would you have them do? And it's another way of getting at what are you spending time on that you …”
Stauch: IT buyers underreport pain points due to pride in workarounds
“The IT buyer in particular really prides themselves on figuring things out and being a problem solver. And so they don't think about a lot of this process as being problematic because they figured it out. It works and it's quite effective.”
Stauch: IT automation fails unless setup is faster than manual execution
“Automation doesn't actually work for this IT use case unless it's faster to automate something forever than to do it manually once.”
Stauch: Better UX than Jira or ServiceNow is not a differentiator
“It would have been very easy to start with an IT ticketing system and you could make it much better UX than Jira and ServiceNow. And you could try to sell that. We didn't think that that was really going to be the difference maker. We said, if we can build an …”
Stauch: Enterprise buyers reject stitched-together or B-plus IT tools
“No one wants a mediocre ticketing system and a mediocre workflow builder and a mediocre access management system all kind of stitched together. And they don't even want a B plus version of any of those by themselves.”
Stauch: Serval's demo only clicks when showing the workflow builder
“The demo can be kind of broken up into everything that comes before showing the workflow builder and then everything that comes after, because before you show the workflow builder, people are bored and everything you show them, they think they've seen before. …”
Stauch: Disrupting legacy enterprise software requires attacking its configurability head-on
“I think so much of this legacy enterprise software, it's powerful and it's sticky because of its configurability. And if you're going to disrupt them, you have to disrupt them in a way that actually attacks that configurability head on.”
Stauch: Large enterprises demand software that adapts to existing internal business processes
“And I think what large enterprises want to do is not adapt their process to your software, but adapt your software to their business process. And these large enterprise software platforms like Workday and Salesforce and ServiceNow, they're very good at that. T…”
Stauch: AI boardroom pressure opens large enterprises to early-stage startups
“These enterprises, which historically are very late to the party, they're feeling a little bit more momentum around inter internally and pressure from the board and the C-suite to start implementing or at least exploring AI solutions that those kind of two fac…”
Stauch: Serval supports the entire Microsoft stack via API context
“For us, it's as simple as giving our system context on the Microsoft API, and now we support everything Microsoft. And that architecture decision allows us to just, you know, serve those customers instantly and with very few Hours of development”
Stauch: Serval pairs every customer with a core engineer
“Every customer, in addition to having a forward deployed owner, also has a member of the regular engineering team as an assigned person. So every engineer on the team has a deep appreciation, understanding of what we're building, who it's for, why we're buildi…”
Stauch: Serval's product roadmap has remained unchanged since seed pitch
“Our roadmap hasn't really changed in a year and a half. When we first pitched first round, those slides are basically our roadmap today. You know, priorities shifted, but The same core things that we're trying to build are true then as they were true now.”
Stauch: Software delays stem primarily from roadmap thrash, not engineering speed
“So I think a lot of delays on software teams is because of thrash, because you change direction and you try different things and you pivot around. I think we had a lot of conviction on what we were trying to build here. And we've been able to just like hammer …”
Stauch: Enterprise IT sales require top-down executive mandates over bottoms-up adoption
“There has to be top-down alignment. That's what we found, is that there has to be alignment from senior leaders that this has to be done. Because as you get into larger organizations, the actual folks doing the implementation, it can be a very large team. And …”
Stauch: AI capability is not the bottleneck; translating business rules is
“I think the gap here that humans are solving Is translating the business rules and requirements into the underlying product. What's really interesting about all of these AI products today is they can basically do anything. You're not really limited on the capa…”
Stauch: IT workflows vary too widely across companies for fully autonomous AI
“And there's shocking diversity from company to company, not, not even mid-market enterprise, but even within mid-market companies within enterprises on how they want password resets to work on how they want people to get access to applications. All of that end…”
Stauch: Prompt-to-code generation has advanced to where it is hard to break
“Code generation works super well now. So when we take these prompts to code, there used to be so many gotchas where the code wouldn't actually work, whether the API is not well supported or it makes a couple mistakes here and there. And now It's hard to break …”
Stauch: LLMs do not need to improve further to build a massive business
“I'm really happy with where the models are at today because they don't have to get any better for us to build a massive business. How they work today is actually good enough for our use case.”
Stauch: Future AI could allow startups to rapidly copy software
“The competitor I'm most worried about is the one that doesn't exist yet. The one that comes along a couple of years from now. And maybe these tools have unlocked so many crazy capabilities that they're able to move very quickly and build so much of what we bui…”
Stauch: Most enterprise processes requiring automation lack clean, structured APIs
“Obviously the universe of things that you might want to automate is much bigger than the things that have very neat, clean APIs.”
Stauch: IT teams deploy AI for other departments while lacking their own
“IT is witnessing sales have amazing AI tools and design having amazing AI tools and engineering, of course, having amazing AI tools. And they're just sitting watching, often implementing those tools for the org and paying for them, but not having their own too…”
Stauch: Forward-deployed engineers replace legacy software implementation consultants
“I think in many ways, what the forward deployed engineer role has actually replaced are the implementation consultants for large organizations, large companies like Workday and ServiceNow and Salesforce that are customizing that software to meet these organiza…”
Stauch: Traditional software customer feedback loops take months to reach engineering
“Traditionally you've got solutions engineers, you've got customer success staff, sales, AE, they talk to customers. Maybe the SE here is a good idea. They communicate that to a product manager. Product manager talks to the engineering manager about that. Maybe…”
Serval expects forward-deployed engineers to spend 20%+ time talking to customers
“There is very little difference other than we expect the forward deployed engineer to be spending 20% or more of their time talking to customers.”
Stauch: Product-oriented forward-deployed engineers will reduce PM and SE hiring
“I think the need for a lot of product managers is probably going to go down. If you hire really product oriented for deployed software engineers, then they're going to be increasingly making the decisions on what makes sense to build for the max Some number of…”
Stauch: Verkada scaled to 1,000 employees with only two product managers
“I mean, Verkata was a thousand people and had two PMs.”
Stauch: Ignoring P2 features erodes product experience despite good prioritization
“And it's also a mistake that actually the really good product leaders do is because they're so focused on prioritization, they never touch the P twos. P two stack up and you can get a lot of them that just completely erode the product experience. And if you're…”
Stauch: Product teams must embed directly with customers, not rely on 30-minute interviews
“You have to be embedded with a customer and it's not, oh, I'm going to schedule a customer interview and I'm going to have an engineer talk to a customer for 30 minutes because we're doing a customer interview and they've got a list of questions they're going …”
Stauch: Serval biases hiring toward technical generalists over domain experience
“No, I think we've definitely decided to bias more on really smart generalists that hopefully are technical or quantitative in some way, because I think all these technologies are going to change very quickly. The way of doing things is going to change very qui…”
Stauch: AI code generation makes coding literacy more valuable, not obsolete
“Shockingly, I think there's higher order returns to being technical and being able to code than there were. I think there was this moment where maybe we thought everything would be no code. And I think this era of these models being very good at code generatio…”
Stauch: IT staff with basic coding skills are now exponentially more valuable
“Now an IT professional, for example, with some degree of programming experience, Is incredibly valuable to the organization because of all the things that they can build for the company. Whereas five years ago, they were maybe marginally more valuable than the…”
Stauch: Interrogating candidate life decisions reveals personal agency vs drifting
“I love a storytelling interview where I ask people to tell me their story, starting with where they went to school, why, what they did there, internships, why they went and took their first job, their next job, and then I just double click into things that I f…”
Stauch: Hiring merely 'good enough' candidates rapidly degrades company talent
“When we're looking at candidates that we're bringing in, I like to make people tell me on the interview panel, who are they better than at the company? Because I think what often happens at your company. What often happens is coming to scale is you bring in pe…”
Stauch: Serval evaluates every job candidate on their ability to recruit
“We evaluate every candidate on their ability to recruit the next candidate, and we consider that very highly. And that could be them actually being somebody that goes out and makes phone calls and recruits, or also just being somebody who's so good that we kno…”
Stauch: Some Serval Employees Make More from Referrals Than Salary
“There's some folks on the team that have done quite well financially that are probably making more money from referrals than they are from their salary.”
Stauch: AI Software Categories Will Be Dominated by Single Winners
“And I think the same thing is going to happen in AI. And in this case, a lot of times it might be the incumbent that still wins, but there are going to be AI native competitors creating a massive, massive value in these categories. And I don't know That it's g…”
Stauch: AI lets startups combine product simplicity with enterprise configurability
“AI gives them a chance to compete both on simplicity and user experience and also configurability because you're using AI to set these configurations or maybe achieve value propositions that these other platforms just can't touch.”
Stauch: Legacy enterprise software giants will naturally lag in adopting AI
“They have this classic innovators dilemma that it makes it hard to steer the big ship to add these capabilities while still protecting their existing product and revenue not making drastic changes that are not backwards compatible. And so they're just naturall…”
Stauch: Startups should synthesize category history with first-principles design
“Weighing those two is, is more valuable than adopting either one wholesale. I think a lot of companies struggle to create new categories from scratch and they kind of ignore the fact that we should maybe look at why there's a category called ITSM and maybe loo…”