Crisis Protocol
Your clinicians write the protocol. The system recognises the words you listed and performs the action you defined, without assessing anyone's risk.
Many practices already have a crisis policy written down. On the phone it is applied by whoever happens to pick up, and it varies with training, tenure and the hour. Encoding it means the same words produce the same action at three in the afternoon and three on a Sunday morning. The system contributes execution, never judgment.
The failure people reasonably fear is software deciding whether someone is in crisis. That decision does not exist anywhere in this design. There is a list your clinicians wrote, an action they attached to it, and a default route to a person for everything else. There is no risk model and no severity score.
A caller in distress who reaches hold music while the front desk is on the phone with a pharmacy has met the practice's phone system, not the practice. Every line is answered at once, so the routing step your protocol defines happens immediately rather than after a queue.
Because triggered calls are captured with the caller's wording intact, the practice can review the transcript set, see which phrases fired and which did not, and revise the list. A crisis protocol becomes something the practice maintains rather than something it assumes is working.
Under 1 second to answer
No hold queue sits between a caller and the routing step your protocol defines
24/7 coverage
The protocol runs identically overnight, on weekends and on holidays
HIPAA BAA included
Call records and escalation logs encrypted with AES-256 at rest and TLS in transit
No, and it is worth being blunt about it. There is no assessment step, no screening question and no scoring. It matches words your practice supplied to actions your practice defined. Any judgment about a caller's state is made by a person, after the routing your protocol specified has already happened.
Whatever you wrote down. That is usually a transfer to a named destination, your own message read word for word, or both in a sequence you set. There is no paraphrase, no summarising and no reassurance the system invented for the occasion.
None by default. You configure the destination, because services differ across the United States, Canada and Australia, and because practices differ on whether the phone should transfer, read a number, or hand the call straight to a person. We do not assert what your protocol should say.
It follows your default route, which many practices set to a person. There is no branch that resolves an unclear call on its own. Escalating a call that did not need escalating is a recoverable error; a call that quietly ends in voicemail is not.
Third-party callers follow the branch your protocol defines for them, which is usually different. The system does not confirm whether the person being described is a patient, and it does not evaluate what the caller is reporting. It captures the wording and routes as configured.
No. The clinician your protocol names is still the clinician, and a human service can remain part of the chain. What changes is that every call reaches the routing step the same way, is answered immediately, and leaves a record of what happened.
Related
After-Hours Call Coverage for Psychiatry Practices
Who gets reached overnight, in what order, and what happens when the first destination does not answer.
Related
Between-Session Calls to a Therapist
What may interrupt a fifty-minute session, and what waits for the gap between them.
Related
Family and Third-Party Calls in Behavioral Health
The call from a relative who wants someone to know something, handled without confirming anything.
Pillar guide
AI Receptionist for Behavioral Health
The full behavioral health front desk: intake, scheduling, medication calls, coverage and escalation.
See how MedReception AI handles after-hours calls, scheduling, intake, and patient communication for medical practices like yours.
