Plain-language investigation
Start from the symptom and attach a service, time range, trace, dashboard, alert, or on-call context.
Ask about a service, anomaly, trace, alert, or on-call owner in plain language, let governed tools gather the relevant telemetry, review the supporting evidence, and require approval before sensitive operations execute.

The investigation remains reviewable because the question, context, tools, evidence, proposed action, approval, and result stay visible.
Turn recurring diagnosis or response steps into governed automations instead of rewriting the same prompt.
Separate advice from action, require approval where policy demands it, and retain the execution outcome.
workflow
AI shortens the path through the evidence while policy remains in control of operational action.
Attach the service, environment, time range, alert, trace, or dashboard that anchors the investigation.
Query the relevant telemetry and ownership data through the configured tool and network policy.
Verify the supporting results and send sensitive operations through the required approval path.
Keep the operation, approver, result, and execution details available for later review.
proof / shipped
Current product surfaces include chat, preserved investigations, automations, approvals, executions, prompts, providers, MCP tools, and network policy.
Ask about a service, anomaly, trace, alert, or on-call owner in plain language, let governed tools gather the relevant telemetry, review the supporting evidence, and require approval before sensitive operations execute.