AIRestAI Restaurant Receptionist
Menu
For restaurants

What AIRest does today

Everything on this page works in the product now. Planned work has its own section at the bottom, marked as planned.

On the phone

Answers the phone from your own facts

  • Greeting
    Picks up in your restaurant's name

    With the greeting you publish in the admin.

  • Questions
    Answers from facts you approved

    Menu, dietary and allergen questions, opening hours, address, parking and policies.

  • Unknown
    Offers a callback

    When your facts do not cover a question, Juno offers a callback.

  • A person
    Transfers the call

    When a guest asks to speak to someone, the call goes to your staff.

  • Language
    One language per restaurant

    You pick the language; speech recognition and the Amazon Polly voice follow it for the whole call.

  • Intents
    18 kinds of request

    Bookings, menu and allergen questions, hours, address, policies, private events, takeout, callbacks, complaints and requests for a person.

Bookings

Books only what is really free, only after a yes

  • Availability
    Checks your real tables

    Against tables, opening hours and party-size rules, before a time is offered.

  • Read-back
    Repeats the booking with the date resolved

    “Saturday, October 10 at 7:30 PM, for 4 people, under the name…”

  • The gate
    Writes the booking after the guest says yes

    A confirmation whose details differ from the read-back is rejected.

    Yes first
  • One table
    No double-booking

    A Postgres constraint makes two overlapping bookings on one table impossible.

  • House rules
    Large parties, opening hours, notice

    Parties above your limit go to staff with a callback offer, times outside your hours are refused, and you set the minimum notice and how far ahead guests can book.

  • Confirmation
    SMS and email to the guest

    Sent when the booking is created, on a connected Twilio line and email account.

Your team

An admin your staff can read at a glance

Dashboard, calls, bookings, usage and costs, staff and users.

  • Calls
    Transcript and outcome of every call
  • Bookings
    The booking ledger and tonight's tables
  • Facts
    Taken from your website or a document

    The platform extracts the facts; you approve or reject each one before Juno may use it.

  • Profile
    Drafts, publishing and rollback

    Change the greeting, rules and facts in a draft, publish when ready, roll back if needed.

  • Roles
    Owner, manager, host, viewer

    Each person sees and changes only what their role allows.

  • Locations
    Several restaurants, one account

    Each location keeps its own profile, phone numbers and usage limits.

Connections

Phone numbers, events and usage

  • Numbers
    Twilio Voice numbers from the admin

    Provisioned per restaurant. Webhooks are signed and fail closed in production.

  • Events
    Seven n8n webhook events

    Call completed, call unrouted, reservation created, booking created, booking review requested, action requested and callback requested.

  • Usage
    Metered per restaurant

    Call minutes and model tokens, with usage limits.

  • Model
    Any OpenAI-compatible endpoint, or Ollama

    The receptionist runs on the model the deployment is configured with.

On the roadmap

What we plan to build next

Everything in this box is future work, listed so you can see where the product is going.

  • Voice
    Natural voices from ElevenLabs

    Flash v2.5 speech streamed over Twilio Media Streams, and one branded voice per restaurant.

    Planned
  • Speed
    Streaming speech for faster turns
    Planned
  • Languages
    English and French in one call

    For Ottawa–Gatineau first, then Ukrainian for community restaurants.

    Planned
  • Staff
    Staff alerts, reminders and call summaries
    Planned
  • Systems
    Two-way sync with reservation systems, and a waitlist
    Planned
  • Demo
    A real phone number for the public demo
    Planned