AI Workflow Automation Startup Serval Hits $3B Valuation

A $3 billion bet that AI agents can replace incumbent business software

Serval, a San Francisco company that sells enterprise service management software, has raised $150 million in a Series C round led by Sequoia. The round values the company at $3 billion.11 That is three times the $1 billion valuation Serval reached in December. Reuters reported that the jump reflects investor interest in AI startups going after established enterprise software vendors such as ServiceNow and Workday.18 The company has now raised $277 million in total. Returning backers Elad Gil, Evantic, First Round, General Catalyst and Redpoint took part, and Conviction and Headline joined as new investors.11

Serval has repriced itself very quickly. PitchBook data cited by Reuters put the company's value at $232 million in August 2025.6 It passed $1 billion in December and has now tripled again.15 The company was founded in 2024 and has raised three rounds in about 12 months.1 Crypto Briefing pointed out that this Series C alone is bigger than everything Serval raised before it.15 Dealroom placed the round in the 95th percentile of all-time US enterprise software Series C deals, based on a sample of 1,743 rounds.13

Read together, the coverage says something about AI workflow automation beyond one company's funding. Investors are no longer paying only for AI that answers questions. They want agents that act inside workflows and can take over from the systems those workflows currently run on.

What Serval actually builds

Serval's product is built around two AI agents with separate jobs. In the version TechCrunch described at the Series A stage, one agent writes internal automations for routine work such as approving software access or provisioning a laptop, under an IT manager's supervision. A second agent answers employee requests by calling only the tools that already exist.2 Stauch argued that this design lets the help desk agent turn down a dangerous request, such as deleting company data, simply because no tool exists for it.2 Because the tools are deterministic code rather than open-ended model output, they can carry detailed rules, for example requiring multi-factor authentication or restricting an action to certain hours.2

The current product, Catalyst, takes this further. SiliconANGLE reported that it launched in August. It scans support tickets for repetitive work and drafts automations, and an administrator reviews them before they go live.12 CityBiz described it as AI-assisted code generation that lets staff in IT, HR, finance, security and legal build and manage workflows without engineering skills.10 McLeod, the co-founder and CTO, said the goal is to give non-engineers "a powerful code generation tool" for automating their most tedious work.11

The no-code AI automation angle matters here, though the term fits loosely. Serval does not offer the drag-and-drop builders that defined earlier business process automation. Reuters reported in December that the platform generates code in the background, which Stauch said gives more flexibility than conventional workflow builders.6 Users describe a process in plain language and the agent writes, tests and publishes it. They can then refine it with more prompts or edit the code directly.4 In effect, Serval is selling the kind of AI coding that changed software engineering, packaged for operations staff who have never used a code editor.

The ServiceNow question

The coverage agrees most strongly on who Serval is targeting. Stauch told Reuters the product is "a true alternative replacement for platforms like ServiceNow" and not an extra layer of AI spending added to an existing system of record.16 SiliconANGLE reported that Serval sees ServiceNow as its main competitor.12

The main evidence is a single migration. Stauch named Lucid Motors as the carmaker that moved from ServiceNow to Serval in 45 days.16 According to the company, that switch covered IT service management, the configuration management database, and hardware and software asset tracking.1 The Next Web, citing Forbes, reported that Stauch called the 10,000-person company's move "a full rip and replace."

This is a change in strategy. In December, Reuters reported that Serval could either fully replace a customer's systems or sit on top of them as an "AI layer," an option meant for customers locked into multi-year contracts.6 The company had also built integrations with Jira Service Management and Freshservice, so customers could use its AI without replacing their ticketing tools.8 With this Series C, the layer option has mostly disappeared from the company's messaging, and the pitch is replacement.

ServiceNow's response, as reported, is defensive but solid. The company told Forbes it keeps a 98% renewal rate and that inbound interest from first-time customers is up 60%. Dealroom added that ServiceNow and Workday have both expanded agentic automation across IT and HR and are emphasizing enterprise controls.13 The timing was awkward for the incumbent. The Next Web reported that ServiceNow shares fell as much as 25% on the day of the announcement, their biggest intraday drop since its 2012 listing, after its bookings outlook missed forecasts. A startup funding round did not cause that drop. Still, the two events landing on the same day will feed the "SaaSpocalypse" argument that both the founders and Salesforce's Marc Benioff have weighed in on.

Sequoia's role adds weight to the story. The firm backed ServiceNow in 2009 and is now funding a company built to replace it.13 When it pre-empted the Series B, Sequoia partner Anas Biad said the firm had last heard customer feedback this strong 16 years earlier, when it partnered with ServiceNow.6

Growth claims, and what is missing

Serval says its annual recurring revenue has doubled every quarter since the Series B.11 It has not disclosed the actual figure, and several outlets noted that.1213 It says two Fortune 20 companies have signed up for organization-wide deployments. The press release says this happened "in the last few months," while The Next Web reported the founders as saying the past six months.11 Named customers include Fox, WPP, Live Nation and Ramp, and the company says most customers go live within 30 days.11

The customer base has changed noticeably. At the Series A, Serval's references were AI-native companies such as Perplexity, Mercor and Together AI.2 Perplexity's IT head said the company had automated more than half its IT tickets.3 The current list is weighted toward large media, advertising and automotive companies. That matters because it is harder to show success inside a legacy enterprise than inside a startup with an up-to-date tech stack.

The operational growth is also steep. In December, Serval had just under 30 employees and expected to pass 100 this year.6 It now says headcount has grown tenfold in 12 months and plans to hire more than 500 more people, opening offices in New York, London and Austin.12 One outlet appears to have misread the hiring plan. Crypto Briefing attributed the 500-plus hires to the ProServal services unit15, but the company's release and most other coverage say 500 is the company-wide target, with about 100 or more of those in ProServal.1110

Why the services arm is the real tell

The most revealing part of this round may be ProServal, more than the valuation. Serval plans to hire more than 100 "Automation Engineers" over the next year and place them inside large customers. About half will be recent graduates who go through a four-week training program.1 The company presents this as a break from legacy consulting, saying it hires for learning speed and not years spent on one platform.11 Stauch said nobody has ten years of experience in the role because the tools did not exist two years ago.

In this writer's reading, this is an admission. A pure product story would say that a plain-language interface lets departments automate their own work. Yet Serval itself says rolling out AI across an organization is still the biggest challenge in large enterprises.11 Building an implementation team suggests that AI agent workflows inside Fortune 20 companies still need people who understand how work actually moves between departments. ServiceNow built a large partner and consulting ecosystem for the same reason. Serval is hiring that capability in-house and staffing it with younger, AI-native workers.

Dealroom pointed to the main risk. The valuation assumes early deployments will hold up reliably across large organizations.13 Serval's architecture has a reasonable answer on governance. Reuters reported that IT and security teams control permissions and approvals, and that customers still generally want a human to sign off on sensitive actions such as password resets and access to applications.17 The open question is whether that human-in-the-loop model can scale across HR, finance and legal without simply recreating the ticket queues it was meant to remove.

The takeaway

What stands out across the reporting is how consistent it is. Every outlet ties the round to the same thesis: AI agents that write and run deterministic automations can become the system of record instead of a feature added to one. Investors are paying a large premium for that thesis while the revenue behind it remains undisclosed. One public migration and two unnamed Fortune 20 customers are a promising start, but they do not yet prove that incumbents can be displaced. The coming year, as Serval's new automation engineers go into some of the world's most complex companies, will show whether the $3 billion valuation reflects where the market is heading.