Restaurant system
Customers order on WhatsApp in their own words, and the order shows up in the kitchen.
Our own product, in development. A multi-venue SaaS for restaurants that takes orders over WhatsApp.
At a glance
- Multi-venue
- each restaurant isolated within the same platform
- text orders, interpreted with AI
- CI
- every change runs through GitHub Actions
The problem
A small restaurant uses one tool for tables, another for delivery, a spreadsheet for the cash register and a phone for WhatsApp orders. We wanted a product that brought all of that together and that any venue could sign up for with nothing to install.
How we approached it
The system is a monorepo with a NestJS API and a Next.js web app, on Supabase (Postgres, Auth and Realtime) with Drizzle. The API is organized by domain: orders, menu, tables, cash register, delivery, customers, staff, reports, invoicing and payments, among others. Each restaurant lives isolated within the same platform.
The core of it is the WhatsApp bot: customers write their order however they like, and a language model turns it into a real order that shows up at the venue in real time. Restaurant subscriptions are billed through Mercado Pago, files go to Cloudflare R2, and every change runs through continuous integration on GitHub Actions.
Before launch we ran a production-readiness audit and wrote a pilot plan for the bot.
Where it stands
It’s our own product and it’s still in development. No restaurants are live yet.