Junior Exercise 5 - Incident Triage Warm-Up

Rating

• size: 5
• complexity: 3

Purpose

Practice using AI as a thinking partner during early incident triage without letting it create false certainty.

Scenario

After a deployment, confirmation emails are delayed.

Available signals:

• queue depth rose from 200 to 4,000
• email worker retries increased sharply
• order creation latency stayed normal
• release note mentions retry-policy cleanup

Worker pseudocode:

sendEmail(message)
if failure:
  retry in 30 seconds
  mark attempt count + 1
if attempt count > 3:
  move to dead-letter queue

Task

Use AI to help generate:

• possible causes
• missing evidence you would want next
• a safe next-step plan

Then decide what you would and would not conclude yet.

Expected output

• ranked hypothesis list
• next 3 verification steps
• one sentence on current confidence
• one escalation trigger

Good AI uses

• organizing partial evidence
• generating sensible next questions
• surfacing what is still unknown

Verification focus

• distinguish symptoms from causes
• do not assume the release note is the root cause
• identify what evidence would confirm or falsify each hypothesis

Anti-patterns to watch

• presenting a root cause too early
• copying an AI incident summary as if it were evidence
• confusing correlation with confirmation

Debrief prompts

• what did AI help you organize well
• where did AI make the situation look more certain than it was
• what did you deliberately refuse to conclude