Every guest, a regular.
Regular brings every email, text, WhatsApp, iMessage and call into one work list, attached to the guest and their booking. It answers the easy questions, drafts the rest for your approval, and hands you the ones that need a person.
-
Sarah, by textTonight 7:30, 4 guests
“Stuck on the bridge, about 20 min late, so sorry!”
AnsweredReplied, and noted on her booking.
-
Marco, by emailSaturday 8:00, 4 guests
“Any chance we could bring two more on Saturday?”
ApproveNo table for six at 8. The draft offers 6:15 or 9:30.
-
Priya, on WhatsAppFriday 7:00, 2 guests
“It’s our tenth anniversary, and my wife has had a hard year…”
YoursA personal note. Her booking and history are ready for you.
The guest is waiting. The answer is in four places.
A question about Saturday arrives by text. The booking is in one system, last week’s email is in another inbox, and the person who knows the guest is on the floor. By the time someone pieces it together, the guest has waited an hour, or called twice.
Regular puts every conversation in one list, sorted by what needs you first, with the guest, their booking and everything already said attached. It works from a phone, between tables.
House rules for the AI
You decide what the agent may do, for each restaurant and each kind of request. The defaults:
These rules are enforced underneath the AI, not just suggested to it. An outside assistant connected over MCP plays by the same rules: it can’t change a booking without your approval either.
When the phone rings, you already know who it is.
Regular matches the number to the guest and puts a card on your phone: their name, their booking, and the likely reason they’re calling. The card arrives at once; a short AI note follows when it’s ready.
It works alongside your phone system. It doesn’t replace it.
Junk filtered out. Guests never lost.
Shared inboxes fill up with sales pitches and scams. Regular sets them aside, off the work list and without a notification, and keeps them recoverable for 30 days. One tap on “Not junk” brings a message back and remembers the sender.
A guest telling you a fake booking site scammed them is a guest, not junk.
Never filtered, whatever the AI thinks
- Anyone you already know as a guest
- Anyone with a booking
- A conversation your team has replied to
- Anything the agent isn’t confident about: that is flagged for you instead
Works with what you already run
Regular connects to your systems through open, documented connectors. Nothing needs replacing.
Bookings
Bookings, notes and availability come from the system you use today, so the agent always answers with the booking in front of it.
Messages
Email, SMS, WhatsApp, iMessage, RCS, Instagram, Messenger and Viber, each reply going back on the channel it came from.
Phones
Call alerts from your phone system or carrier, so the guest’s card is on screen before you pick up.
Your AI
Choose the AI provider for each restaurant: OpenRouter, Anthropic, any OpenAI-compatible service, or a model you run yourself.
Any MCP client, such as Claude Code, can work the list too, under the same house rules. Connectors and openness
Made inside a working restaurant group
Regular is being built inside a restaurant group, for its own teams first: real inboxes, real bookings, real phones ringing during service.
Replies should sound like your front of house on a good night, never like a robot. We’re now opening it to a few other groups.
Join the waitlist
We’re bringing restaurant groups on a few at a time. Tell us about yours and we’ll be in touch.
We use these details only to reply to you about Regular. See the privacy notes.
You’re on the waitlist.
Thank you. We’ll write to you at the address you gave.