Dvara welcomes every guest, books more tables, protects peak-time revenue and turns cancellations into new bookings. It answers on WhatsApp, books real availability, writes the reservation to your Google Calendar and reminds the guest so they turn up. Day or night.
Dutch & English · restaurants first · governed AI · EU-hosted · built by C-Drivelab
Hoi! Is er zaterdag om 19:30 nog plek voor 4 personen?
Zeker — zaterdag 19:30, tafel voor 4 is vrij. Zal ik hem vastleggen?Je spreekt met de AI-assistent van Bistro Hugo. Liever een medewerker? Typ mens.
Ja graag 🙏
Genoteerd. Zaterdag 19:30, 4 personen. Donderdag stuur ik een herinnering — één tik om te bevestigen of wijzigen.
Typical first reply
Modelled reference week, not customer data
The door that never closes — 22:41 on a Saturday included.
No rip-and-replace, and nothing to learn. Dvara handles the conversation and puts the confirmed booking where your staff already look.
Each agent speaks your trade, follows your rules and hands over to a person the moment it isn't sure. Pick a business and watch it work.
Restaurants are live today. Salon, clinic and garage templates run on the same booking core and are being validated one at a time through 2027 — shown so you can see where this goes, not sold as available.
Answers table requests on WhatsApp and web chat, checks real availability from your rules, books, reminds, and refills cancellations from a waitlist.
Table for 4 on Saturday at 19:30?
Saturday 19:30 for 4 is free. Shall I hold it?
Yes please, one guest is vegetarian.
Done, noted the vegetarian guest. I will send a reminder on Thursday.
Books by stylist, service duration and buffer time, handles rebooking, and keeps the day packed without double-handling messages.
Balayage with Maya on Friday afternoon?
Maya has 14:00 or 16:30. Balayage takes about 3 hours, so 14:00 fits best.
14:00 please!
Booked with Maya, Friday 14:00. See you then.
Collects the reason for the visit, basic details and preferred times, books routine appointments, and routes anything urgent to staff. It never gives medical advice.
I have had a toothache since last night.
Sorry to hear that. I will flag this as urgent for the team. Can you come in today at 16:10?
Yes, I can.
Booked for 16:10 and the team has been alerted. Please call us if it gets worse.
Takes the plate and the problem, offers a slot that fits the job length, creates the job card and keeps the customer updated as work moves.
Brake noise on my Golf, can I come Tuesday?
Tuesday 09:00 is open. A brake check takes about 2 hours. What is your plate?
Plate is on its way, book 09:00.
Held for Tuesday 09:00. Send the plate and I will complete the job card.
Your rules decide what is bookable; Dvara holds the conversation and keeps it in one shared inbox. Today it writes confirmed bookings to Google Calendar. Outlook, reservation-system, CRM and payment connectors are on the roadmap, not in the product.
Understand · check your rules · book · remind · follow up · report. Every step logged.
The model holds the conversation. Your rules hold the calendar — so availability, capacity and pricing are never a guess.
A message lands and is matched to the business that owns the number.
Intent, party size, date and language, read from plain Dutch or English.
Your rules decide what is bookable. Nothing free? A waitlist is offered.
Commits once, under a uniqueness constraint. No double bookings.
A nudge before the visit, with one tap to confirm, move or cancel.
Cancellations go to the waitlist within a minute. Releasing unconfirmed peak-night tables for refill — Saturday Protection Mode — is on the roadmap.
Every gate emits an event, so reports come from records, not memory.
Reference flow for a single location — not customer data.
One shared inbox, a separate set of rules per site, and one view across every location. Add a branch without rebuilding anything.
Sample locations
Analytics ships today. On the roadmap: the same conversation that books the visit takes the deposit, asks for the review and brings quiet slots back to life. Not sold as available yet.
Illustrative week
Guest data, allergies, deposits and private-event requests are sensitive. Dvara is built so you can show how the AI behaved — not just say so.
Guests are told they are talking to an AI at the top of every conversation, and the disclosure is stored with it — not a toggle someone can quietly switch off.
Staff take over any conversation mid-sentence from the inbox; a guest reaches a person by replying with one word. High-value and sensitive requests — large groups, private events, allergies, payments — wait for a human.
Every AI message is marked, every tool call it made is kept, every booking it created says so. Append-only, exportable, yours.
The model phrases the reply. Your availability rules and a database constraint decide what can be booked. It cannot invent a table.
One European region, AI inference inside the EU, encrypted at rest and in transit. Retention rules per restaurant, automatic deletion, and an export on demand.
Reminders, waitlist offers and allergy notes each have the guest's recorded yes. Your conversations are never used to train models unless you decide otherwise.
For a single restaurant this simply means fewer surprises. For groups, franchises, hotels and multi-location operators it is the list procurement asks for — and we can hand it over.
Plus a one-time €99 setup. Onboarded within 48 hours, including your menu, opening hours and policies.
Still unsure? Tell us your setup and we will say plainly whether Dvara fits.
No. Availability is answered by your rules and the booking commits under a database uniqueness constraint. The model phrases the reply; it never decides what is bookable.
It hands over. Low confidence, complaints, large groups and payment questions go to a human queue, and staff can take over any conversation mid-sentence.
Always. The disclosure appears at the start of every conversation and is recorded with it, and a guest can reach a person by replying with a single word.
In one European region, encrypted at rest and in transit. You configure retention, you can export everything, and deletion is demonstrable rather than asserted.
No, and not by default ever. Our model provider is contractually barred from training on or retaining your conversations, and our own quality reviews use anonymised examples. If that ever changed it would be your explicit, written decision — never a setting we flip.
No. Dvara answers in front and writes the confirmed booking to your Google Calendar; your floor keeps working the way it does. To be precise about today: Dvara is the booking source of truth and the calendar is a one-way mirror. Direct connectors to reservation systems and CRMs are on the roadmap, not shipped.
Enquiries answered inside a minute, covers booked, or no-shows avoided — we pick the metric together and report it honestly. If it did not move, you walk away.