Serval's Catalyst mines the ticket queue for automation work, not the project backlog

Serval, founded in April 2024 by former Verkada employees Jake Stauch and Alex McLeod, has released Catalyst, an AI agent that searches ticket histories for recurring problems, checks which connected systems could resolve them, and then generates the workflows, skills and access configurations to do it ^[2][3]. The notable move is not code generation but the sequencing: discovery shifts from a human backlog exercise to a search over tickets, which weakens the assumption that every automated workflow needs its own funded technical project ^[3][4].

The founding anecdote is the argument. Stauch says a CFO handed an IT leader a straightforward rule for approving expenses, and turning that rule into a working workflow took two months and hundreds of steps in Okta Workflows ^[1]. If each rule costs a project, automation coverage scales with the number of people you can put on building it. That is the cost structure Catalyst is aimed at.

Stauch describes the inversion plainly: Catalyst "focuses on the problem, not the solution," analysing the tickets users raise, what is possible through available APIs and what the platform can do, rather than reproducing the steps an IT team used to close a ticket ^[4]. Nothing publishes itself. According to Stauch, administrators can examine each step, add permission checks so an automation runs only for designated users, require approval from named individuals, groups or workflows, and decide what gets deployed ^[5]. The human is still in the loop, but at review rather than at build, which is where the headcount argument lives or dies.

The funding says investors buy it. Serval raised a $47 million Series A in October 2025 and a $75 million Series B led by Sequoia, taking total funding to $127 million at a $1 billion valuation ^[6][7][8]. Those two named rounds sum to $122 million, $5 million short of the stated total, so earlier capital exists that the source does not detail ^[1].

The incumbent numbers cut the other way. ServiceNow reported second-quarter 2026 subscription revenue of $3.88 billion, up 24.5% year over year, with its AI business past $1 billion in annual contract value and customers running agentic AI in production up ninefold in nine months ^[9][10][11]. Stauch claims that among customers Serval talks to, less than 10% of the ServiceNow AI products they bought have been deployed, and that ServiceNow is pushing them to reprioritise budget so it can show AI revenue ^[12]. ServiceNow did not respond to a request for comment on that claim ^[13]. Mike Leone of Moor Insights & Strategy calls the criticism fair, saying nearly every enterprise vendor sold AI into 2025 budgets faster than customers could staff the rollouts ^[14].

Leone also supplies the discipline. He says Serval's TypeScript approach produces automation you can read, version and hand to an auditor, built in hours rather than through a services engagement, and that ServiceNow moves more slowly there ^[15]. But he warns against reading that as a moat: code generation is becoming a standard way to build enterprise tooling, ServiceNow already has a code agent, and an agent acting on your behalf has to know your processes, your approvers and which systems break when it touches them ^[16][17].

Watch the acceptance rate: what share of what Catalyst proposes an administrator actually deploys, and whether review becomes the new queue. Watch whether Serval accounts add automations without adding implementation staff, since that is the only proof that spend has decoupled from workflow headcount. And watch ServiceNow's deployed-versus-sold ratio on its AI SKUs, which is the number Stauch is betting stays low ^[12].