AI reception that's on call when yours isn't.

Sabiya is built specifically not to. In healthcare compliance mode, Sabiya refuses any question about diagnosis, dosing, treatment recommendations, or medication interactions — and routes the patient to a practitioner consultation. A deterministic safety guard catches the highest-risk patterns (drug dosing questions, self-harm signals) before they reach the language model, so refusals are reliable even under pressure. Reception sees the refusal in the chat history with a topic tag.

Sabiya's own appointment system runs in parallel today — patients book through the chat widget into Sabiya's calendar with confirmation and reminder emails. Direct sync to Cliniko, Halaxy, and Nookal is on the roadmap. If you'd rather keep using your existing system, Sabiya captures the patient's preferred time and details in the chat, and your reception confirms the booking in your PMS.

The morning briefing shows today's appointments live (queried fresh on dashboard load — bookings made at 7 AM appear at 7:34 AM), with the patient's reason, practitioner, and chat summary attached. Pending leads are tagged by topic — NDIS, WorkCover, DVA, general — so reception can prioritise the high-urgency follow-ups in their first 15 minutes. Synthesised insights show patterns across the week, like recurring questions you might want to add to the knowledge base.

Configure your cancellation window, late-cancel rules, and no-show policy once. Sabiya answers patients consistently and applies the rules at booking and rescheduling time. Patients get an instant confirmation and a day-before reminder by email — with self-serve cancel and reschedule links in each. SMS, tighter reminder timing, and deposits are on the roadmap.

Sabiya replies in your voice from day one. You set the tone, your terminology (do you say physio, physiotherapist, or PT?), and how your clinic talks about its work during onboarding. You can refine it any time as you see how Sabiya replies. No training period.

Patient data is hosted in Sydney, Australia (Supabase ap-southeast-2). Conversations are processed via AWS Bedrock with the Australian geographic profile — inference stays within Sydney and Melbourne data centres. Healthcare-aware mode redacts visitor PII before logs hit the database, and chat content is encrypted at rest. Sabiya is APP-compliant by design, and your patient data isn't used by third-party AI providers for training. Sabiya is intended for front-of-house tasks (bookings, intake, FAQs); clinical notes belong in your practice management system.