Agent Platform (EAP)
Definition ¶
The Enterprise Agent Platform (EAP) is our governed agent execution plane. It runs reviewed AI agents that automate triage, evidence gathering, and guided remediation for operational workflows, taking inputs from Event Management System and other monitoring sources.
EAP replaces the previous n8n.io-based AIOps Studio workflow plane. The n8n model is retired; new automation belongs on EAP.
Why We Use EAP ¶
While Event Management System is excellent at consolidating and correlating alerts, a crucial next step is to act on that information. Manually responding to every incident is slow, error-prone, and doesn't scale. EAP addresses this with a governed agent runtime that provides:
- Automation of repetitive tasks — frees engineers from manual triage, evidence gathering, and ticket bookkeeping.
- Orchestration across approved systems — agents reach external systems only through reviewed MCP servers (e.g., emsapi, ServiceNow, Prometheus, Grafana, Intersight, Kubed).
- Reviewed decision-making — agent prompts, skills, and MCP scope go through Git-tracked review on two planes (project repo +
enterprise-agent-platform-registry) before activation. - Consistency and standardization — every approved agent runs under the same platform guardrails, sandbox, and audit trail, so the same input produces the same governed behavior.
Success Measurement ¶
The success of EAP is measured by its direct impact on operational efficiency and service reliability. Key metrics include:
- Reduction in Mean Time to Triage (MTTT) — average time from alert creation to triage for agent-assisted vs. manual responses.
- Agent-handled rate — percentage of incidents that reach a useful decision summary via an EAP agent before human takeover.
- Reduction in toil — engineering hours saved on repetitive triage and evidence-gathering steps.
- Project adoption — number of teams running governed agent projects on EAP.