Built since 2020 with 1,500 hoteliers. It answers guests on WhatsApp, drafts the email reply for the front desk to approve, flags the one line in the night audit worth a look, and suggests tomorrow's rate. A person always decides.
Each card is a real situation at Westbrook Hotel, redrawn in code. The clock is yours. Press the buttons: nothing is sent anywhere.
Sarah Mitchell, room 204, asks for a late breakfast at the time you see in the chat. Zurba extends it to 11:30, writes it on the reservation, and tells the kitchen.
AC in 204. Task to Jake Thompson on Telegram, 10-minute timer, then Dave Collins.
03:00. 104 folios posted. One line flagged: room 311 marked clean, camera saw it occupied.
Tomorrow is 91% booked. Zurba suggests +$6 on the standard room; Emily Carter taps Apply or Ignore.
Same rule as above: Zurba does the routine part and a person decides the rest.
Zurba apologises, logs the complaint on the reservation, opens a task for the right person and offers a gesture from the list the hotel set, up to the limit the hotel set. If the guest is still unhappy, it sends the case to the duty manager, Dave Collins, with the whole thread.
Emily Carter asks in Telegram how yesterday went. Zurba answers with occupancy, revenue, incidents, open tasks and pending approvals, and sends a PDF if she asks for one.
No. If a guest needs to pay, Zurba sends a payment link instead. The number never goes through the chat.
Only from the hotel's own list of small gestures. Bigger ones wait for a manager on Telegram.
Small things, like a late breakfast, yes. Anything above a small gesture needs a person to decide.
The hotel sets a monthly AI budget. After it is used, chats go to reception.
Zurba has been built with the people who answer guests, run the desk and close the night audit.
61 guest intents and 464 example questions, in English and Arabic.
The hotel answers 31 required questions before Zurba may talk to a guest, per language.
Tell us about the property and we will get back to you as soon as possible.