Analyze incidents with AI SRE to understand the root cause and get a summary of what happened.

Analyze an incident

Start an investigation by bringing an incident into the chat in one of three ways in AI SRE → New chat:

• Click on an ongoing incident from the welcome screen.
• Type @ in the message box to search for and select an incident.
• Ask AI SRE to find ongoing or past incidents, e.g., "show me incidents from last Tuesday".

Start an investigation directly from an incident

Open an ongoing incident in Incidents and click Investigate with AI SRE in the incident timeline, or Use AI SRE near the top of the page, to launch an investigation of that incident. To investigate every new incident automatically, set up the AI SRE integration in Incidents.

Once an incident is in the chat, AI SRE pulls the full record, including:

• The incident timeline, including status changes, acknowledgements, and resolutions.
• On‑call users involved.
• Associated comments and linked Linear issues.

The agent then typically queries related logs and metrics from the time of the incident to identify the likely cause and provide a summary of its findings.

Query the on-call schedule

Ask questions like:

• "Who is on‑call right now?"
• "Which schedules does the web team have, and who is on‑call for each?"

Curious about on-call schedules?

To learn more about on‑call schedules, see the getting started with on‑call guide.

Escalate an incident

Ask AI SRE to escalate an incident when it needs a human on it, for example "escalate this to the on‑call engineer". AI SRE never pages anyone on its own. It proposes the escalation and waits for your approval first.

Before anyone is paged, AI SRE shows an Escalate incident? card in the chat with the incident, who will be paged, and how they will be notified. The target can be a team member, an escalation policy, or an on‑call schedule. Approve the card to page them, or reject it.