When a call is urgent, a person takes it.

You give us one number: a mobile or a desk that is answered. If a caller says any word or phrase on your red-flag list, the agent stops booking and puts the call through to that number straight away.

We write the list with you. It usually covers things such as chest pain, trouble breathing, heavy bleeding, severe swelling, a very young or elderly patient in distress, and the words “emergency” or “urgent”. Each clinic decides what belongs on it.

If nobody answers that number, the agent tells the caller so, tells them to call the local emergency number, and sends your team the details at once.

If a caller asks to speak to a person, in Arabic or English, the agent does not argue, persuade or ask why. It puts the call through to the same number.

The agent does not say what a symptom means, what to take, whether to wait, or whether something is serious. It does not diagnose and it does not reassure. It books, takes a message, or hands the call to a person.

If the agent is not sure of an answer, it does not guess. It says so, takes the caller’s name, number and reason, and sends your team a message. It always says that it is an AI assistant.

• The red-flag list and the handoff number are yours. We change them only when you ask.
• You ring the agent yourself and test the rule before anything goes live.
• Every month a person reads a sample of your calls, and the report says what we fixed.

What is stored from each call is in the data note. The rules every agent follows are in how we use AI voices. An agent is not an emergency service.