• AWS-native AI integration · ships in 6–10 weeks

Vet clinic adds after-hours triage with zero misroutes

  • Voice AI · Veterinary
  • Independent small-animal veterinary clinic — 4 doctors, ~$3.2M revenue, ezyVet

The problem

Two clear categories of lost revenue: routine after-hours bookings going to a 24-hour competitor, and anxious pet-owner calls during business hours where front desk was unavailable. The lead DVM also wanted a way to triage "is this an emergency?" calls overnight.

Our approach

What we built

  • Inbound voice agent handling routine appointment booking, Rx refill requests, and after-hours emergency triage.
  • Triage rubric co-designed with the lead DVM: keywords like “bloat,” “blocked cat,” “seizure,” “hit by car,” “ate chocolate/grapes/xylitol” route directly to the on-call DVM’s mobile. Lower-acuity cases get next-morning callback. Rule-based, not LLM-judged.
  • SMS follow-up sending pre-arrival paperwork link and arrival instructions.
  • Monthly transcript review with the lead DVM to recalibrate the triage rubric — three triage decisions were re-categorized in month two based on her review.

Caveats

True veterinary emergencies require live human triage in minutes, not 24-hour callback — the rule-based escalation is non-negotiable and we wrote it into the contract. State medical-records release requirements vary; we route any release request to the practice manager rather than auto-fulfilling.

Stack. Vapi · AWS Bedrock Claude · ezyVet REST · Twilio SMS · custom emergency-triage rubric reviewed monthly by the lead DVM.

Outcome

~6 routine appointments per week now booked while the clinic is closed. 100% of true emergencies escalated correctly (n=11 in the 90-day window; 0 misroutes). Post-call NPS 8.6 vs 7.2 for the prior human answering service.

100%
true emergencies escalated correctly
6/wk
after-hours bookings captured
8.6
NPS (was 7.2)