Loading
Because the upside is modest and the downside is not. Answering opening hours saves minutes; answering anything that could be read as clinical advice is a serious failure, and a plausible-sounding model will attempt it unless it is stopped.
So the only usable assistant here is one whose limits are enforced rather than instructed.
Guardrails define the topics it may address and the topics on which it must hand over immediately, and it answers only from the documents and records you give it, citing them.
It holds no clinical record. Where a question touches symptoms, medication or advice, it stops and routes to a person — that is a configured boundary, not a hope.
It will not triage, advise, or discuss an individual's care in any form, and it must not be configured to.
And it does not make you compliant. It gives you enforced boundaries and a complete log; the policy and the lawful basis are yours.
Directions, opening hours, what to bring, how to register, how to request a repeat process, where a form is — the administrative tier that consumes reception time all day.
Every conversation is logged as AI-handled with what it cited, so a review of what it said is a query.
No. That is an enforced boundary with immediate handover, not a prompt instruction.
No clinical record. It answers from published information you provide.
Yes, logged with the sources it cited.
Half an hour on your own numbers is usually enough to say whether Pulse is the right place to start.