🍽️ AI Restaurant Operating System

WhatsApp Ordering • AI Routing • Kitchen Operations • Waiter Dashboard • Reports

A working restaurant automation prototype built with n8n, LLMs, Kapso WhatsApp API, Supabase, JavaScript, and Netlify.

Overview

This project connects a customer-facing WhatsApp ordering experience with restaurant operations.

Customers can browse an interactive menu, submit orders through WhatsApp, receive AI-assisted responses, confirm or modify orders, request a waiter, ask for the bill, and complete the order lifecycle. Staff use dedicated kitchen and waiter dashboards, while managers can review operational reports.

This repository contains a sanitized public portfolio version. Credentials, secrets, customer data, and private infrastructure identifiers have been removed.

Business Problem

Restaurant teams often manage digital enquiries, orders, kitchen updates, waiter requests, and payment closure across disconnected manual steps. This creates duplicated work, slow handoffs, inconsistent customer responses, and limited operational visibility.

Solution and Business Value

This system connects the customer conversation directly to restaurant operations. It reduces repetitive message handling, structures incoming orders, routes tasks to the correct team, and gives staff a shared view of order status.

Faster handoff from customer order to kitchen
Consistent order capture and confirmation
Clear separation of kitchen, waiter, and cashier responsibilities
Reduced manual follow-up for routine requests
Better operational visibility through live dashboards and reports
Human handover remains available when automation is not appropriate

Main Capabilities

Area	Capability
Customer experience	Interactive digital menu and WhatsApp order handoff
AI layer	Intent classification, FAQ handling, and structured order parsing
Session management	Customer states such as waiting order, confirmation, modification, cancellation, and active order
Kitchen operations	Pending, preparing, and ready order lifecycle
Waiter / cashier	Serve order, request bill, payment closure, and service alerts
Database	Supabase order storage, order events, and customer session data
Reporting	Revenue, order counts, active orders, cancellations, dine-in, takeaway, and average order value
Realtime	Dashboard refresh through Supabase realtime events
Session closure	Resets the customer to `waiting_order` after payment

System Architecture

Repository Structure

AI-Restaurant-Operating-System/
├── README.md
├── NOTICE.md
├── .gitignore
├── docs/
│   ├── ARCHITECTURE.md
│   ├── BUSINESS_FLOW.md
│   ├── DATABASE_SCHEMA.md
│   ├── FINAL_REVIEW.md
│   ├── SECURITY.md
│   └── SETUP.md
├── images/
│   ├── interactive-menu.png
│   ├── kitchen-dashboard.png
│   ├── waiter-dashboard.png
│   ├── reports-dashboard.png
│   ├── main-n8n-workflow.png
│   ├── close-session-workflow.png
│   ├── system-architecture.svg
│   └── workflow-overview.svg
├── screenshots/
│   └── README.md
├── workflows/
│   ├── ai-restaurant-operating-system-public.json
│   ├── close-order-session-public.json
│   └── README.md
└── web-app/
    ├── index.html
    ├── app.js
    ├── styles.css
    ├── _redirects
    └── README.md

Core Workflow

Kapso receives an incoming WhatsApp message.
n8n normalizes the payload and rejects outbound/self-generated messages.
The customer record and current session status are retrieved.
The message is routed by session state and intent.
LLM workflows parse orders, classify requests, or answer menu FAQs.
Confirmed orders are written to Supabase.
Kitchen and waiter dashboards manage the order lifecycle.
After payment, the close-session workflow returns the customer to waiting_order.

Engineering Highlights

State-based conversation routing instead of relying on a single unrestricted AI prompt
Structured LLM outputs for reliable order parsing and validation
Idempotency and self-message filtering to reduce duplicate processing
Separate customer, kitchen, waiter, cashier, and reporting concerns
Database-backed order lifecycle with realtime dashboard updates
Sanitized public workflows with credentials and private identifiers removed

Technology Stack

Automation: n8n
AI / LLM: OpenAI-compatible models and Gemini through OpenRouter
WhatsApp: Kapso WhatsApp API
Database: Supabase / PostgreSQL
Frontend: HTML, CSS, JavaScript
Hosting: Netlify
Integration: REST APIs, webhooks, JSON, realtime events

Included Workflows

1. AI Restaurant Operating System

workflows/ai-restaurant-operating-system-public.json

Handles:

Message normalization
New and returning customers
Session routing
AI order parsing
FAQ responses
Order confirmation
Modification and cancellation
Waiter alerts
Bill handling
Supabase order creation
Conversation logging

2. Close Order Session

workflows/close-order-session-public.json

Triggered after payment to reset the customer session for a new order.

Web Application

The web-app folder contains:

Interactive product menu
Dine-in and takeaway selection
WhatsApp order generation
Kitchen dashboard
Waiter / cashier dashboard
Reports dashboard
Supabase realtime updates
Paid-and-close webhook integration

Configuration placeholders are documented in web-app/README.md.

Screenshots

Interactive Menu

Kitchen Dashboard

Waiter / Cashier Dashboard

Reports Dashboard

Main n8n Workflow

Close Order Session Workflow

Security Notice

This public version excludes:

API keys
Credentials
Production webhook secrets
Customer phone numbers
Private Supabase identifiers
Internal deployment details

See docs/SECURITY.md before using any workflow in another environment.

Project Status

Functional prototype / portfolio project

The system has been developed and tested as a functional restaurant automation demo. Production deployment would require authentication, row-level security, monitoring, rate limiting, backup policies, and environment-based secret management.

Author

Badreldin Mohamed Awad
AI Automation Engineer | n8n Workflow Architect | Applied AI Systems Builder
Al Ain, UAE
Email: badrna3om@gmail.com

Navigation Menu

Repository files navigation