An AI receptionist that books appointments

It is 23:47 and the clinic closed at 21:00. A patient opens the clinic's website on a phone and asks Noor for a cleaning on Saturday. Noor offers the free times, the patient picks 10:00 and gives a name and phone number, and the booking is confirmed with a reference. Beside the phone, the same booking appears in the clinic's admin while nobody is on shift. The next morning it is waiting at reception. The video has no sound.

Oravie is a concept brand we designed and built to show this kind of product. The clinic and its patients are fictional.

Clinics lose patients who want to book outside opening hours or can't get through on the phone. The reception desk is busy during the day and closed at night, which is exactly when many people sit down to book.

• Grounded in the clinic's own information

  Noor answers questions about treatments, prices and insurance using what the clinic has published, and cites the source for each answer.

• Only real availability

  She reads the calendar and offers only times that are genuinely free, presented as buttons the patient can tap.

• Books and cancels

  A booking takes a name and a phone number, returns a reference, and can be cancelled in the same chat.

• Handles emergencies

  Severe pain or swelling receives the clinic's emergency instructions immediately, not a booking flow.

• Every booking appears in the clinic admin

  The team sees the day's appointments and which ones Noor booked overnight.

• Next.js
• Vercel AI SDK
• Claude
• Supabase

The assistant can't invent a time or a price — it reads slots and writes bookings only through a small set of server tools, and is restricted from offering medical advice.

• →We can put an AI assistant in front of real business data: calendars, price lists, bookings.
• →We build it with safe limits: no medical advice, sources shown, no invented availability.
• →What the AI does appears where your team already looks, not in a separate inbox.