After-Hours Triage for a Rural GP Practice Without Burning Out the Doctor

Rural and remote GPs face a challenge that urban clinics rarely encounter: they are often the only doctor within 50 kilometres, and their patients know it. Patients message or call after hours with everything from genuine emergencies to questions about whether a rash is worth worrying about. Without a system, the GP is either on call 24/7 or patients go without guidance.

OpenClaw provides a structured after-hours triage layer that handles incoming patient messages, separates true emergencies from routine concerns, and gives patients appropriate guidance. Without requiring the doctor to be personally available at 11 PM.

The Setup#

This GP uses Telegram because many of their rural patients already have it installed and find it more accessible than WhatsApp in areas with patchy mobile data.

curl -fsSL https://openclaw.ai/install.sh | bash
openclaw onboard --install-daemon

During onboarding, connect the clinic's Telegram bot. Patients message the bot's Telegram handle, which is published on the clinic website and appointment cards.

The After-Hours Triage Skill#

Skill: after-hours-triage
Trigger: incoming Telegram message (active 6 PM – 8 AM weekdays, all day weekends)
Prompt: "You are an after-hours message handler for [Clinic Name], a rural general practice. A patient has sent a message. Your job is:

1. Greet them warmly and acknowledge their concern
2. Assess their message for urgent symptoms, if they mention ANY of: chest pain, difficulty breathing, stroke symptoms (face drooping, arm weakness, speech difficulty), severe bleeding, loss of consciousness, severe allergic reaction, or thoughts of self-harm, immediately reply: 'This sounds like it may be a medical emergency. Please call 000 (emergency services) immediately or ask someone to drive you to the nearest emergency department. Do not wait.'
3. For non-urgent concerns: ask two clarifying questions to understand the situation better. Based on their answers, either: (a) reassure them and provide general guidance to monitor at home, or (b) advise them to call the clinic when it opens at [time] to book an urgent appointment.
4. Always end with: 'I am an automated after-hours assistant. For all medical decisions, please see Dr [Name] during clinic hours or call 000 in an emergency.'
5. Send a brief summary of the conversation to Dr [Name] via SMS at 7 AM each morning."

The Emergency Escalation Path#

The triage prompt is designed conservatively. Any mention of high-risk symptoms triggers an immediate emergency referral. The AI does not attempt to assess severity. This is intentional. Better to send a patient to emergency services unnecessarily than to miss a true emergency.

The Morning Summary#

Every morning at 7 AM, OpenClaw compiles all overnight messages and sends the GP a structured summary:

Skill: overnight-summary
Schedule: 0 7 * * *
Prompt: "Review all Telegram messages received overnight. Group them by urgency: (1) emergencies that were redirected to 000, (2) patients advised to call for an urgent appointment today, (3) patients given home monitoring advice. Format as a brief morning briefing for Dr [Name] and send via SMS."

This means the GP starts every morning knowing which patients need to be seen urgently, which had concerns addressed overnight, and whether any emergency situations arose.

Patient Feedback#

Patients in rural settings consistently report that receiving any response after hours, even from an automated system, significantly reduces their anxiety. The key is that the triage bot is honest about being automated and clear about when to call 000. It does not pretend to provide medical advice; it provides structured guidance and a clear pathway.

What the GP Cannot Delegate#

The bot handles routing. The GP still handles medicine. Any patient who is advised to call for an urgent appointment is seen the same day or next morning, and the GP reviews the overnight summary every day without exception. OpenClaw extends availability. It does not replace clinical responsibility.

The same pattern on a phone line rather than a chat thread is an AI receptionist: it answers the after-hours calls live, books what can be booked, and pages the on-call doctor for the rest. In the FrontDesk AI build that came to 71% of after-hours contacts handled without waking the doctor, with every emergency reaching a person in under a minute. If you want to know what the calls nobody answers are currently costing, the missed call calculator does that arithmetic.