AI Querier

Ask questions about an incident in plain language

Alert payloads are long. A single Sentry or monitoring alert can run to hundreds of lines, and the detail a responder needs — which region, which cluster, which component — is somewhere inside it.

The AI Querier answers questions about that data in plain language. Ask which region was affected or what triggered an alert, and the Querier extracts the answer from the incident's payload and context.

Questions can be asked in the incident's Slack or Microsoft Teams channel, so responders get answers without leaving the channel they are working in or opening a monitoring tool.

The Querier retrieves facts that are present in the data. It does not infer cause. To investigate cause, use the Root Cause Agent.

Before you begin

Requirements

• The Slack or Microsoft Teams integration must be configured for the team.
• The incident must have an alert payload. The Querier reads what your monitoring tool sent.

Important

• The Querier can only report what the payload contains. If your monitoring tool does not send the affected region, no question about the region will produce it.
• Richer alert payloads produce better answers. If responses are consistently thin, the integration sending sparse alerts is the thing to fix.
• Confirm any answer that will drive a mitigation decision against the payload itself.

Ask a question

In Slack or Microsoft Teams

1. Open the incident's channel.
2. Ask your question in plain language, for example Which region was impacted?
3. The answer is returned in the channel.

Tip: Ask one thing at a time. A single specific question — which cluster is affected — returns a cleaner answer than a compound one asking for the cluster, the region, and the trigger together.

What the Querier can answer

AI Querier or Root Cause Agent?

Both read the same incident. The Querier retrieves; the Root Cause Agent reasons.

Use the Querier when you know which fact you need. Use the Root Cause Agent when you do not yet know what to ask.