The strange thing about enterprise automation is how much manual work it can take to automate something simple. Serval CEO and cofounder Jake Stauch remembers a CFO giving an IT leader a straightforward rule for approving expenses. Turning that rule into a working workflow took two months and hundreds of steps in Okta Workflows.

Serval was founded to attack that gap. Stauch and fellow former Verkada employee Alex McLeod launched the company in April 2024 with the idea that companies should not need a technical project every time they want to eliminate a repetitive task. That idea now sits at the center of Catalyst, Serval's new AI agent, which searches ticket histories for recurring problems, examines the systems a company has connected and determines whether those problems can be automated. It then generates the workflows, skills and access configurations needed to make the automation work.

"Catalyst focuses on the problem, not the solution. It analyzes the problems users are raising and asks whether they can be solved with automation. It's not analyzing how an IT team resolved a ticket and simply reproducing those steps. Rather, it examines the tickets themselves, what's possible through the available APIs and the capabilities of the Serval platform, and then builds the automation needed," Stauch told me in an exclusive interview.

He explained that Catalyst doesn't automatically publish what it builds. "They can examine each step, add permission checks so it runs only for designated users and require approval from specific individuals, groups or workflows before it proceeds. The administrator decides what ultimately gets deployed," Stauch says.

Serval raised a $47 million Series A in October 2025, followed by a $75 million Series B led by Sequoia. The two rounds brought the enterprise AI startup's total funding to $127 million and pushed its valuation to $1 billion.

Stauch claims the platform is an "AI-native, better alternative" to ServiceNow and says it has already started taking customers away from the dominant enterprise service-management platform. He argues that a platform built around modern code-generation agents can move faster than one that has spent decades adapting an older architecture.

"We've built our platform around code-gen agents as a fundamental primitive. It's a fundamentally different architecture from ServiceNow, one that wasn't possible if you started a company before 2024," Stauch says.

Mike Leone, vice-president and principal analyst at Moor Insights & Strategy, says Serval's bet on TypeScript is meaningful. "TypeScript gives you a new kind of automation that you can read, version and hand to an auditor. It can be built in hours rather than through a traditional services engagement," he says. "That's a real advantage, and ServiceNow does move more slowly in that area."

ServiceNow's latest results , however, tell a different story. In its second-quarter 2026 results, subscription revenue reached $3.88 billion, up 24.5% year over year, while its AI business crossed $1 billion in annual contract value. The number of customers with agentic AI in production increased ninefold in nine months. But Stauch claims Serval is hearing a different story from customers that use ServiceNow, saying less than 10% of the ServiceNow AI products they have purchased have been deployed.

"We hear from our customers that they've bought a bunch of ServiceNow SKUs and have never been able to get them implemented. ServiceNow is pushing them to reprioritize the budget so that it can show AI revenue." ServiceNow did not respond to a request for comment on the claim.

"Nearly every enterprise vendor sold AI into 2025 budgets faster than customers could staff the rollouts. The criticism is fair, and I hear versions of it constantly," Leone says. Two decades of accumulated customization give ServiceNow a significant advantage, Leone notes, but he cautions against treating Serval's early lead in code generation as a durable moat. Code generation is becoming a standard way to build enterprise tooling, and ServiceNow already has a code agent of its own.

"An agent acting on your behalf has to understand what your processes are, who approves what and which systems can break when it touches them. ServiceNow has spent 20 years collecting that context, while Serval is earning it one customer at a time."

Turning AI Code Generation Into Enterprise Automation

Serval's existing platform handles employee support across IT and other business functions, with requests coming through Slack, Microsoft Teams, email, web portals and other channels. Catalyst works one step earlier. An administrator can describe what they want to automate in natural language, and the agent generates the code and configuration needed to build it. The architecture builds on an idea Serval has used from the start. One agent creates the tools and automations, while another uses those tools to resolve employee requests. That separation gives administrators a clearer boundary between what AI can build and what can actually run in production.

The platform primarily uses models from OpenAI and Anthropic through zero-retention endpoints. Stauch compares Catalyst with Cursor, Claude Code and Cognition, which help software engineers write application code. "We are code gen for everything that's not software engineering," he says.

But model developers such as Anthropic increasingly have the technical ingredients to build software that can interact with enterprise APIs, reason over business context and execute multistep tasks. A sufficiently capable model could create a password-reset script in minutes.

Stauch does not dispute that OpenAI or Anthropic could generate the code needed to automate a task, but argues that generating the code is only one part of building an enterprise automation product. A production enterprise workflow needs security controls, approval routing and permissions. He claims that foundation-model companies are "unlikely to spend the enormous amount of resources required to build that layer because their business incentives point toward products where the model itself supplies most of the value."

If models become increasingly commoditized and more value shifts into the application layer, the incentives for OpenAI and Anthropic could change. Serval is betting that by the time they do, it will be "too deeply embedded in enterprise infrastructure and customer workflows for them to easily catch up."

Serval Wants to Replace ServiceNow, Not Sell Alongside It

Stauch's decision to target ServiceNow stems from the tech giant's roughly 90% penetration across the Fortune 500 and its pace toward about $15 billion in annual recurring revenue. Moreover, he believes the large-enterprise service-management market has effectively become a one-player market, leaving customers with few meaningful alternatives.

“Other companies have taken the approach of moving downmarket and focusing on smaller businesses, but we’re going direct, intending to fully replace ServiceNow,” he says.

Serval wants customers to consider the budget they already spend on service management rather than find new money for another AI product. Stauch compares that approach with selling a CRM against Salesforce, an HRIS against Workday or an ERP against SAP , where the existing platform sets the budget a challenger is trying to capture.

That positioning has already created some friction. Stauch says his company planned to sponsor ServiceNow's Knowledge 2026 conference but was told it could not participate.

ServiceNow is also investing heavily in the architectural gap Serval says gives it an advantage. The company paid more than $2.85 billion for Moveworks. Moreover, the company claims its Level 1 Service Desk AI Specialist is live with more than 40 customers and handles 80% to 85% of service requests without human interaction, while its AI-native products carry a 20% to 30% pricing premium.

Serval has previously reported an 80% help desk-automation rate.

ServiceNow collapsed five packaging tiers into three and bundled Now Assist, Moveworks, the Data Fabric and AI Control Tower into all of them. "Everyone with a contract bought AI whether they went looking for it or not, so a deployment rate measured against everything sold doesn't tell you very much," Leone says. "ServiceNow does disclose deployment numbers, but those are self-reported, so discount them accordingly."

Investors Are Betting On A New Era Of Enterprise Software

Statistics shared by Serval claim that its revenue increased roughly 500% between its Series A and Series B periods. Serval also says customers automate more than 50% of tickets, although the company has not publicly disclosed its exact ARR, profitability or detailed customer count.

The startup’s $1 billion valuation reflects what investors believe the company could become, not evidence that it has reached the scale of the enterprise software giant it hopes to replace. Stauch describes the investor thesis: "talent, market and timing." He compares the shift with the transition to cloud computing, when a generation of cloud software companies reshaped markets once dominated by older vendors. In service management, he points to the transition from BMC Helix to ServiceNow as an example of how one generation of enterprise software can give way to another.

"We don't know if that's going to happen yet, but that is the bet a lot of our investors are making. They believe this is another massive tech shift that's going to create a new generation of incumbents," Stauch says.

Leone notes that ServiceNow's IT service desk is sold by the seat, which creates a problem when AI can resolve tickets without a person involved. "Half of ServiceNow's new business is now priced on something other than a seat, and repricing at that scale is what a company does under pressure. Serval didn't have to displace anyone to change how the incumbent charges."

The startup is entering a market with a much larger cast of competitors than its ServiceNow-focused pitch suggests. Atomicwork is building an AI-native ITSM platform around its Atom assistant and has moved into agentic identity governance. Espressive and Rezolve AI remain independent, while Console and Ravenna are also appearing in head-to-head evaluations.

Stauch says most of those startups do not appear in Serval's enterprise deals. "We hear about them from the press and VCs, but they don't really show up in our deals. We see ServiceNow, to a lesser extent, then Jira Service Management, Freshservice and just a couple of others," he says. "Maybe Ravenna is the exception, since they're only a month or two behind us."

The Bigger Bet Is Enterprise AI Control

Serval's long-term vision extends beyond tickets. "We want to take everything we're building for these IT admins, HR admins, finance admins—these automations they're building—and allow employees to automate more and more of their work with their own personal agents that they can use at work," Stauch says.

That could move Serval closer to an enterprise control layer than a conventional ITSM platform. The company already sits around access requests, tickets and privilege escalation, giving it a potential role as employees ask personal AI agents to perform work beyond their normal permissions.

The models powering enterprise agents may become interchangeable, while the permissions, approval rules and workflows governing what those agents can actually do could prove harder to replace. Question is whether enough CIOs will trust a young AI startup to sit between their agents and enterprise systems.

"For a startup to genuinely displace ServiceNow, it needs a named enterprise reference outside the cloud-native cohort. It needs a real answer for the configuration database and change management. And somebody has to own the generated code once there are thousands of workflows and a regulator asking who signed off," Leone says. “Serval does deserve credit on access management, as they built it into the same product as ticketing, and ServiceNow had to buy a company to get there.”