OrcaMonitor's AI Agents run entirely within your infrastructure using your configured model provider. They analyze incidents, correlate evidence, generate remediation proposals, and assist operators — but never execute without explicit human approval.