A familiar interruption.
A different way to handle it.
It is the start of Friday service. Guests are arriving, a table needs the bill, and the phone rings. The caller only wants a table for four, but answering means stepping away from the people already in the room. Letting it ring leaves the caller to try again or look elsewhere.
For this example, The Olive Room needs a way to handle straightforward reservations without asking an assistant to make decisions about allergies, private events or a complicated seating arrangement. The useful outcome is a checked booking with clear details, or a specific request for the team to follow up.
The details that make the difference.
A useful receptionist starts with information the business controls. Here is how this example is configured.
- Business knowledge
- Opening hours, directions, terrace policy and a current menu. Allergy questions go to a person.
- Booking rules
- A sample capacity of 32 guests and a 90-minute booking duration. Every overlapping reservation contributes to occupied capacity.
- Phone cover
- The existing provider forwards calls when the team is busy. Staff transfers go to a separate phone that does not forward back.
- Team workspace
- Bookings, call details and follow-up messages are available together. The owner controls access and call-content retention.
Understand the whole party
Tell asks for the date, time and number of guests. A reservation for four needs room for four people, rather than being counted as a single occupied place.
Check before confirming
Availability includes bookings that overlap the requested sitting. Tell also checks opening hours. The final reservation is checked again as it is saved, so a competing booking cannot silently take the same remaining capacity.
Leave something useful behind
After collecting the name and contact number, Tell confirms only a successfully saved booking. A preference such as a window seat is captured as a request for the team, not a promise about a particular table.
A good call has somewhere to go.
Switch between the scenarios to see how the assistant handles the request and what happens next.
One ordinary booking, taken care of.
- Tell
Hi, you’re speaking to an AI assistant for The Olive Room. How can I help?
- Caller
Could I book a table for four this Friday at seven?
- Tell
Of course. Let me check Friday at seven for four guests.
- Tell
That time is available. May I take your name and a contact number?
- Caller
Sophie Williams. I’ll give you my number. A window table would be lovely.
- Tell
Your booking is confirmed for four on Friday at seven. I’ve noted the window preference for the team, although I can’t guarantee that table.
Your team picks up a clear next step.
Sophie’s table is booked.
- When
- Friday · 19:00
- Party
- 4 guests
- Duration
- 90 minutes
- Note
- Window table requested
Illustrative booking record. A guest preference remains a request for the restaurant to review.
Calls, bookings and messages are kept in the workspace, with access based on the person’s role. Retained call content can help the team review what was said; recording and call-content retention are controlled by the owner.
Explore a sample workspace The public workspace demo uses The Olive Room sample data.The right person, at the right moment.
A person for the judgement calls
Private dining, allergy questions, complaints and unusual requests go to the team. The assistant does not invent a menu detail or approve a special arrangement.
Capacity, not a floor plan
Tell checks the configured guest capacity. It does not allocate a specific physical table or decide whether tables can be joined. Keep capacity conservative and route seating decisions to staff.
One place to keep availability current
These reservations live in Tell. If the restaurant also uses another booking system, agree how availability will be kept in sync before taking customer bookings through both.
Check the details.
Then decide what works.
These are evaluation steps for this example, not claims about achieved results. Start with test calls and review the records before forwarding customer calls.
Try the awkward times
Test a full sitting, a booking that overlaps another and a time near closing. Compare the answer with the bookings in the workspace.
Read the saved details
Check party size, date, contact details and notes after each test booking. Confirm that an unsuccessful reservation was never described as booked.
Test the human route
Ask an allergy question and request a person. Verify that the intended staff phone rings and decide how the team will cover it.
Good questions to ask.
Can it promise a particular table?
No. It can save a seating preference in the notes. Your team decides which physical table to allocate.
What if another caller takes the last space?
Capacity is checked again while the booking is saved. If the slot fills, the assistant needs to offer another checked time rather than confirm an unavailable reservation.
Can callers change or cancel a reservation?
The current voice booking flow creates new reservations. For changes and cancellations, Tell can collect a message or transfer the caller so your team can update the booking.
Does this replace our existing booking system?
The pilot stores reservations in Tell. An external booking-system connection must be agreed and verified separately; it is not assumed to be active.
