Serval, a San Francisco-based startup, is attempting to redefine enterprise IT operations with the launch of Catalyst, a platform that moves beyond the reactive, ticket-based workflows that have long defined the industry. While most tools in the era of agentic AI have focused on accelerating existing processes—such as helping chatbots summarize logs or answer basic questions—Catalyst aims to automate the creation of the workflows themselves.
Catalyst, which debuted on August 20, 2026, isn’t just another chatbot designed to deflect support requests. Instead, it is an AI agent that builds the automations themselves. By analyzing help desk ticket history, Catalyst identifies repetitive patterns and then drafts the actual workflows, skills, forms, access policies, and dashboards needed to handle them. It even creates background agents that continuously inspect connected systems to propose fixes before an employee ever realizes there is a problem.
Jake Stauch, CEO of Serval, frames this as a fundamental shift in how companies manage their internal operations.
Most AI agents today wait for an employee to ask a question or submit a ticket. We believe the future is AI that acts before an employee ever submits a request.
Stauch describes the goal as “automating automation itself,” moving the industry from reactive support to proactive intelligence.
The technical approach here is notable. Rather than relying on the low-code, drag-and-drop interfaces that have defined the last decade of enterprise software, Catalyst generates readable, versionable TypeScript workflows. This suggests a future where the IT admin’s role shifts from being a manual builder of workflows to an architect and reviewer. To keep things safe, all Catalyst builds are staged as drafts for human review before they are ever published, ensuring that the AI operates within enterprise guardrails.
The early results from customers suggest this approach has legs. During its beta phase, over 90% of Serval customers used Catalyst as their starting point for building new automations. Ramp, for instance, reported building workflows 50% faster and expanding Serval’s footprint across roughly 10 different teams, including finance, legal, and business operations. Meanwhile, Mercor used the platform to onboard more than 4,000 external experts, scaling its operations across seven teams.
This puts Serval in a position of competing directly with ServiceNow. ServiceNow is a titan in this space, reporting $3.88 billion in subscription revenue for the second quarter of 2026 alone, with $1 billion in AI contract value. The company is moving aggressively, having closed its $2.85 billion acquisition of Moveworks in December 2025 to “supercharge enterprise-wide AI adoption,” according to ServiceNow’s Amit Zavery. ServiceNow has also launched its own Level 1 Service Desk AI Specialist to autonomously diagnose and resolve common requests.
The competitive dynamic is stark. Serval, which has raised $127 million in total funding and reached a $1 billion valuation following a Sequoia-led Series B in late 2025, is positioning itself as an AI-native alternative to the incumbents. Serval has claimed that less than 10% of the AI products customers buy from ServiceNow actually get deployed, a figure ServiceNow disputes. While Serval’s agility is clear, it faces the massive challenge of displacing a platform that is deeply embedded in the world’s largest enterprises, including companies like Fox, Spotify, and Notion.
DataM Intelligence projects the enterprise AI agent market to grow from $6.65 billion in 2025 to $142 billion by 2035. As that market matures, the question for IT leaders isn’t just which tool is faster, but which philosophy will win. Is the future of IT a massive, all-encompassing platform that adds AI features to its existing stack, or is it a new generation of AI-native tools that rewrite the rules of how systems are built from the ground up? For now, the industry is split between these two paths, with the ultimate winner likely determined by whether enterprises prioritize the convenience of integrated platforms or the specialized efficiency of AI-native architectures.