Serval Catalyst Launches Proactive AI Agents to Fix IT Before Tickets
Serval launches Catalyst, an AI super agent that finds IT automation opportunities, builds governed workflows and fixes problems before tickets are filed.
Summary
Serval made Catalyst generally available and enabled it by default on Aug. 20, 2026, after more than 90% of customers adopted it as their beta automation starting point. The admin facing super agent scans tickets, SOPs, spreadsheets and instructions, then drafts code backed workflows, skills, onboarding and offboarding journeys, access policies, dashboards and scheduled agents that propose fixes before tickets. A demo generated TypeScript password reset automation across Okta, Google Workspace and Microsoft Entra; a customer agent used switch telemetry, DHCP data and ticket history to trace two offices’ network failures to configuration drift.
Serval evaluates swappable models rather than building one: CEO Jake Stauch said OpenAI GPT led end user interactions and tool calling in May 2026, while Anthropic Sonnet and Opus led code generation. Customers can provide OpenAI or Anthropic keys or compatible endpoints. Drafts inherit user and workspace permissions and can require review. Customers own materials and outputs, which Serval says are not used for training, although Serval and subprocessors process them. Options are SaaS, customer owned AWS without persistent Serval IAM access, or Kubernetes in any cloud or on premises.
Ramp says Catalyst cut workflow building time 50% across about 10 teams; Serval says 600 laptop replacements saved 150 hours. Mercor onboarded more than 4,000 experts across seven teams, Together AI automates 95% of just in time access, and Perplexity handles over half of IT requests and all onboarding. Pricing is private; Stauch says licenses resemble ServiceNow but ownership costs are usually half, sometimes 10% to 20%. Founded in 2024 by Stauch and Alex McLeod, Serval raised $47 million in October 2025 and $75 million in December at a $1 billion valuation, about $127 million total, after 500% revenue growth since August 2025. Its test is whether one conversational layer can outperform rival tool suites in customized enterprises.
Positives
- More than 90% of Serval customers adopted Catalyst as their automation starting point during beta.
- Ramp reports 50% faster workflow building across about 10 teams, while 600 automated laptop replacements saved 150 hours.
- Together AI automates 95% of just in time infrastructure access requests while retaining approval and audit controls.
- Perplexity automatically handles more than half of incoming IT requests and all employee onboarding through Serval.
- Customer materials and generated outputs remain customer owned and are not used to train Serval or third party models.
Risks & concerns
- Catalyst generates code and can initiate production changes, creating risks that require permissions, draft staging, review and publishing controls.
- ServiceNow, Atlassian and Freshworks already offer natural language workflow or agent creation, narrowing Catalyst’s competitive distinction.
- Serval does not publish pricing, and its software license can cost about as much as ServiceNow despite claimed ownership savings.
- Catalyst’s effectiveness remains unproven in large, highly customized enterprise environments with complex legacy configurations.
- Ramp and other performance metrics come from Serval’s own case studies rather than independent assessments.