Why severity scoring, not just ticket creation

Most maintenance platforms stop at "a ticket was created." That leaves someone — a landlord, an overnight manager, a maintenance coordinator — to read through a transcript or voicemail and decide how urgent it actually is. That's the exact step that causes both wasted trips and missed emergencies.

AmenitySOS's triage engine does that classification automatically, during the call itself, using the same reasoning a trained dispatcher would apply: what's the issue, how fast is it getting worse, and what's the realistic worst case if it waits until tomorrow.

The five-level scale

5
Life-safety — auto-escalated
4
Urgent — same-day
3
Moderate — next business day
2
Minor — scheduled
1
Cosmetic — routine

Built to be conservative, on purpose

  • Emergency keywords trigger instant escalation. Flood, fire, gas, sparking, and sewage backup all auto-classify at Severity 5 the moment they're mentioned.
  • Ambiguity rounds up, not down. If the model isn't confident an issue is minor, it defaults to the higher severity tier rather than risk missing a real emergency.
  • Photos refine, not just confirm. A single image can move a ticket from Severity 2 to Severity 5 once the AI sees active flooding instead of a slow drip.
  • Every score comes with a confidence rating, so staff know exactly when to trust the classification and when to take a closer look themselves.
How it improves over time

Every resolved ticket feeds the model: override reasons when staff disagree with a score, actual time-to-resolve, and contractor outcome data. Over weeks, the triage engine adapts to a property's specific quirks — an older HVAC system, a building with frequent minor plumbing issues — without anyone retraining it manually.