Serval Releases Catalyst Super Agent for Automated IT Workflows and Proactive Remediation

Enterprise service management startup Serval has announced the general availability of Catalyst, an administrative AI agent designed to inspect organizational ticket histories, standard operating procedures, and infrastructure telemetry to generate production IT automations. The release marks an architectural shift from reactive ticket-triage bots toward end-to-end automation synthesis, enabling organizations to draft executable TypeScript workflows, access policies, and onboarding journeys fro

2 min
Serval Releases Catalyst Super Agent for Automated IT Workflows and Proactive Remediation

Enterprise service management startup Serval has announced the general availability of Catalyst, an administrative AI agent designed to inspect organizational ticket histories, standard operating procedures, and infrastructure telemetry to generate production IT automations.

The release marks an architectural shift from reactive ticket-triage bots toward end-to-end automation synthesis, enabling organizations to draft executable TypeScript workflows, access policies, and onboarding journeys from natural language descriptions or existing help desk logs.

Proactive Background Agent Architecture

Automated Workflow Synthesis and Code-Backed Execution

Catalyst operates as an administrative layer across Serval's service platform. By indexing past support requests and documentation, the agent identifies high-volume, repetitive tasks (such as credential resets or identity provisioning) and stages executable workflows for administrator review.

Unlike legacy workflow designers that rely on proprietary rule builders and database tables, Catalyst generates standard TypeScript code. The system connects into enterprise identity systems including Okta, Google Workspace, and Microsoft Entra, constructing the required API interactions, user forms, and role-based access controls automatically.

Proactive Background Agents for Remediation

In addition to user-prompted workflow generation, Catalyst introduces roving background agents that monitor connected IT infrastructure on recurring schedules. These agents correlate multi-source signals (including network switch telemetry, DHCP records, and incident histories) to isolate operational drift before employees submit support tickets.

When an anomaly is detected, the background agent generates a draft remediation workflow and flags it for administrator approval. In customer production environments, this model has been used to isolate cross-office network misconfigurations and resolve just-in-time access provisioning requests without manual IT intervention.

Model-Agnostic Harness Architecture

Serval's technical architecture relies on a multi-model harness rather than proprietary foundation models:

  • Conversational Interaction and Tool Calling: Serval utilizes OpenAI GPT models to process end-user requests, parse ambiguous instructions, and manage parameter extraction.
  • Code Generation and Policy Synthesis: The platform routes workflow generation, TypeScript authoring, and schema drafting to Anthropic's Sonnet and Opus models.
  • Custom Endpoints and BYO Keys: Enterprise administrators can provide custom API keys or route requests to self-hosted endpoints, retaining data ownership and compliance boundaries.

The company positions this architecture as a direct challenge to established IT service management platforms such as ServiceNow, Atlassian Jira Service Management, and Freshworks. Founded in 2024 by former Verkada leaders Jake Stauch and Alex McLeod, Serval raised a $75 million Series B led by Sequoia Capital at a $1 billion valuation in late 2025, bringing its total funding to approximately $127 million.

Sources

Written by

More to read

  • LLM Autoscaling and Cold Starts in Kubernetes: Architecture, KEDA Metrics, Model Weight Caching, and Ephemeral GPU Provisioning

    Autoscaling large language model workloads on Kubernetes presents a fundamentally different engineering problem than traditional stateless microservices. While web APIs scale on CPU utilization or request rate within seconds, LLM inference instances require specialized GPU accelerators, massive container images, multi-gigabyte weight tensors, and intensive runtime compilation before serving a single token. Without proactive architectural design, a cold-starting LLM pod on Kubernetes often requi

    1 min
  • Adaptive Optimizers in Large Language Model Pre-Training: How AdamW, Adafactor, and Lion Scale Gradient Updates Across Billions of Parameters

    Adaptive Optimizers in Large Language Model Pre-Training: How AdamW, Adafactor, and Lion Scale Gradient Updates Across Billions of Parameters Large language model pre-training requires optimizing billions of parameters over trillions of tokens across distributed GPU clusters. Standard stochastic gradient descent (SGD) fails in this regime because Transformer loss landscapes are severely ill-conditioned, with gradient magnitudes differing by orders of magnitude across layers and token positions.

    1 min
  • Waymo Details Custom 5nm ASIC and Heterogeneous Compute Architecture for Robotaxis

    Alphabet subsidiary Waymo has disclosed the architectural specifications of the computing platform powering its sixth-generation autonomous driving system, revealing a custom 5-nanometer application-specific integrated circuit (ASIC) engineered specifically for front-end sensor ingestion and edge machine learning. The disclosure, authored by VP of Engineering Satish Jeyachandran and Compute Lead Daniel Rosenband, details how the company processes raw data from onboard sensor arrays in real time

    1 min