Article 342: Serval Tripled Its Valuation to $3B on a Promise to Replace ServiceNow. The Numbers Are the Argument
Serval raised $150 million led by Sequoia at a $3 billion valuation, Reuters reported today. That triples the $1 billion it carried in December. Total raised: $277 million. Previous investors Elad Gil, First Round and General Catalyst came back.
The company builds AI agents for workplace processes, starting with IT operations: employee access, troubleshooting, software and hardware assets. Which is to say it is aimed directly at the category ServiceNow owns.
The sentence that is doing all the work
CEO Jake Stauch, to Reuters: "It's not an additional layer of AI spend on top of your existing system of record, it is a true alternative replacement for platforms like ServiceNow."
That is the entire pitch, and it is a much more aggressive claim than "AI copilot for IT". Every vendor selling an agent into the enterprise this year has promised to sit on top of the system of record. Bolt-on is safe: the incumbent keeps the data model, the audit trail, the approvals, and the renewal. Replacement means ripping out the workflow engine and the institutional memory encoded in it.
Stauch's supporting evidence is one customer: Lucid Motors migrated off ServiceNow in 45 days.
The part that makes replacement plausible
The obvious objection to "AI replaces your workflow platform" is that agents are unpredictable and workflows are compliance artefacts. Serval's answer is architectural, and it is the most interesting thing in the story.
Stauch says the company separates the agent administrators use to build tools from the agent employees use to access approved workflows. Permissions and approvals sit with IT and security. Human approval stays in the loop for password resets, application access, and changes to sensitive systems - because customers keep asking for it.
Read that carefully. The autonomy is in the authoring, not the execution. Someone builds the workflow with an agent, then it runs as a fixed, reviewed thing. That is not an autonomous agent replacing ServiceNow. That is a workflow platform where the workflow is generated rather than configured, and it is a far more defensible product than the headline suggests.
What $3B is buying
At roughly 12.9x a $232 million figure reported around the round, the multiple assumes a lot of ServiceNow's installed base moving. ServiceNow and Workday are not slow incumbents with no AI; they are platforms with deep integrations, decades of process data, and procurement inertia that is itself a moat.
The bet investors are making is a narrower one than it sounds: that IT operations - access provisioning, ticket triage, asset management - is the wedge where the workflow logic is standard enough to be reconstructed by a model, the pain is measurable in headcount, and the incumbent's licence cost is visible enough to compare against.
Key Takeaway: The "AI replaces the SaaS layer" thesis has been a slide in a thousand decks and a shipped product in almost none. Serval is the first company with a valuation that says otherwise. The reason to take it seriously is the split between builder-agent and runtime-agent - that is a real answer to the trust problem, not a marketing one. The reason to be cautious is that a $3B valuation on one 45-day migration is a bet that the second, tenth and hundredth migrations look like the first. Enterprise software is a business of reference customers, and right now they have exactly one.