Patients message the clinic on WhatsApp and an AI agent asks structured triage questions, classifies urgency with OpenAI, and routes them to self-care advice, a same-day appointment, or an ER recommendation — with every conversation logged to the CRM for nurse review.
Typical results
Under 1 min
First response, 24/7
-65%
Front-desk call volume
100%
Transcripts nurse-reviewed
Figures reflect typical results for this kind of automation.
The challenge
The phone lines opened at 8 a.m. to a queue of the same question: should I come in? Receptionists with no clinical training gave inconsistent answers, worried patients showed up for issues a pharmacy could handle, and genuinely urgent cases waited on hold. After hours, callers got a voicemail and headed to the ER by default.
Before and after
Front-desk calls per day
−65%First response time
−96%Inquiries resolved without a call
The automation
We built an AI agent on the WhatsApp Business API that greets each inquiry with structured triage questions — symptoms, duration, age, red flags — and has OpenAI classify the urgency against protocols the clinic's medical director approved. Low-risk patients get clear self-care guidance, moderate cases book a same-day slot directly in the scheduling system, and red-flag answers trigger an immediate ER recommendation. Every conversation logs to the CRM for nurse review every morning.
How it works
- 01
Patient messages
A patient messages the clinic's WhatsApp number with "should I come in?" at any hour.
- 02
Structured triage
The bot asks approved triage questions — symptoms, duration, age, and red-flag checks.
- 03
AI classifies
OpenAI scores urgency against the medical director's protocols in seconds.
- 04
Routed to the right door
Self-care advice, a same-day slot booked in the calendar, or a clear ER recommendation.
- 05
Nurse safety net
Every transcript logs to the CRM, and a nurse reviews the classification queue each morning.
Starts from
- WhatsApp inquiries
- Missed calls diverted
Delivers
- Self-care advice
- Same-day slot
- CRM nurse log
The impact
Typically, around 68% of inquiries resolve entirely on WhatsApp without a phone call, and front-desk call volume drops from about 130 to 45 calls a day. Patients get an answer in under a minute at any hour, and the nurse review queue flags and corrects only a handful of classifications.
Questions about this automation
Who decides how the bot classifies urgency?
The clinical protocols do. The triage questions, the red-flag answers and the routing for each urgency level are written and approved by the clinic's medical director. OpenAI applies those protocols to the patient's answers, and any red-flag answer triggers an immediate ER recommendation regardless of the overall score.
Can patients reach a person instead of the bot?
Yes. Patients can ask for a person at any point, and conversations the AI can't classify with confidence are handed to staff during opening hours. After hours, uncertain cases get a conservative recommendation to seek care, and every transcript is logged to the CRM for the nurse's morning review.
What does a same-day booking look like in the chat?
For moderate cases, the bot reads open slots from the scheduling system and offers a few times in the chat. The patient taps one, the appointment is booked, and a confirmation with directions and what to bring is sent on WhatsApp. This is a multi-step build, usually rolled out in stages.