HiNosh: an AI phone agent that takes restaurant orders

We designed and built an AI phone agent that answers restaurant calls 24/7, takes orders from the real menu and sends them straight to the kitchen.

Client
HiNosh (our own product)
Industry
Restaurants
Status
Live in staging · onboarding first pilot restaurants in Canada
HiNosh
The challenge

What needed solving

What we built

The solution

A natural phone agent

Answers instantly with the restaurant's name, understands real ordering talk like "large pep", handles options, deals and delivery, and confirms the total.

AI menu import

Owners upload photos or a PDF of their menu; AI reads it and they review it side by side with the original before publishing.

Payments by text

The agent texts a secure Stripe payment link; money goes to the restaurant's own Stripe account and the ticket prints "PAID".

Straight to the kitchen

Orders print on cloud kitchen printers and appear on a live board, and the caller gets a text when the order is ready.

Owner dashboard

Every call with a summary and transcript, orders, complaints, analytics, menu editor, multi-branch and team roles.

How it works

Our approach

  • The AI handles the conversation; our own system calculates every price and total, so the agent never guesses a number.
  • A cascaded voice pipeline (speech-to-text, language model, text-to-speech) so we can pick the best component for each language and keep orders exact in noisy kitchens.
  • Each restaurant's data is isolated, with role-based permissions and an audit log for sensitive actions.
  • Restaurants keep their number and choose whether the agent answers every call or only when staff are busy.
Technology
  • LiveKit
  • Deepgram
  • Azure OpenAI
  • Twilio
  • Stripe
  • .NET
  • Python
  • Microsoft Azure
Services
ExploreSee the HiNosh product page

Your project could be next

Tell us what you want to build or automate, and we'll show you how we'd approach it.

Start a project

Tell us what you're trying to build.

Share the problem or the idea. We reply within one business day.