Script a healthcare chatbot's conversation flows so it routes and reassures without ever sounding like it's diagnosing

You are scripting the conversation flows for a patient-facing chatbot used by a healthcare practice or wellness service. The bot's job is to route, schedule, and answer logistics questions — it must never sound like it is assessing or diagnosing a patient's symptoms.

BOT'S INTENDED FUNCTIONS
Appointment scheduling and rescheduling, office hours and location info, prescription refill request routing.

COMMON REQUEST TYPES TO SCRIPT
Book a new appointment, reschedule an existing one, request a prescription refill, ask about office hours.

WHAT SHOULD ALWAYS ESCALATE TO A HUMAN
Any mention of symptoms, pain, medication side effects, or emotional distress; anything the bot doesn't have a matching flow for.

TONE
Friendly and efficient, not overly chatty, should feel like a helpful front-desk staffer.

EXISTING SYSTEM CONSTRAINTS
Bot can only see open appointment slots for the next 14 days; cannot access lab results or medical records.

RULES
For each request type, write the full flow: the bot's opening response, the follow-up questions it asks to route correctly, and the resolution (booked, information given, or handed to a human) — a flow that dead-ends without a clear resolution is incomplete. The instant a user's message contains anything that reads as a symptom description, a request for medical advice, or an emotional crisis, the flow must branch to the escalation path rather than attempting to be helpful about the medical content itself — write the exact bot line that performs this handoff, and make it warm rather than a cold form-letter refusal, since a curt refusal at a vulnerable moment is its own failure. Write the bot's refusal-and-redirect lines so they name what it can help with instead, rather than a bare "I can't help with that" that leaves the user stuck. For any flow where the user might reasonably follow up with a symptom-adjacent question mid-flow (for example, asking to reschedule because they feel too unwell to come in), write that specific branch explicitly rather than assuming it won't come up.

WHAT NOT TO DO
Do not write any bot line that names a possible cause for a symptom, suggests a self-care remedy, or implies reassurance about how serious something is ("that's probably nothing to worry about") — even a well-intentioned reassuring line here is a boundary violation for a bot whose actual job is routing, not clinical judgment.

MANDATORY DISCLAIMER
Write a short standing disclaimer line the bot shows at the start of every session, stating plainly that it can help with scheduling and general information but cannot assess symptoms or give medical advice, and that any health concern should be directed to their care team or, for emergencies, to call emergency services — and add an internal note that all scripted flows must be reviewed and approved by clinical and compliance staff before deployment.

OUTPUT FORMAT
1. The standing disclaimer shown at session start.
2. Each request type as a labeled flow: opening line, follow-up questions, resolution.
3. The escalation flow, written out in full, including the exact handoff line.
4. The internal review note.