Gourmet Chef AI Manager

An AI restaurant manager that takes orders, books tables, tracks deliveries, and serves customers across Telegram, Email, and Web.

Comment

🍽️ Gourmet Chef AI

An AI-powered multi-channel restaurant manager that can communicate, understand, and take real actions through Telegram, Email, and Web.

Gourmet Chef AI is an intelligent restaurant management system that goes beyond a traditional chatbot.

It allows customers to interact with a restaurant through Telegram, Email, and Web, while the AI agent connects to real restaurant data, performs actions such as food ordering and table reservations, and keeps customers and restaurant owners updated.

💡 Inspiration

Most restaurant websites and AI assistants are passive.

A customer usually has to open a website, find the right page, fill out forms, or manually contact the restaurant.

We wanted to build something different.

Instead of making another chatbot that simply answers questions, we wanted to create an AI restaurant employee that can actually take action.

The idea behind Gourmet Chef AI was simple:

What if a restaurant had an AI assistant that customers could reach wherever they already communicate?

A customer should be able to message the restaurant through Telegram, Email, or the Web and simply say:

"Book a table for 4 tomorrow at 8 PM."

The AI should understand the request, check the restaurant's real data, create the reservation, confirm it with the customer, and notify the restaurant.

That became the foundation of Gourmet Chef AI.

🤖 Why Caspian?

The most important part of this project is that the AI is not trapped inside a single chat interface.

We use the Caspian SDK as the communication layer between customers and our restaurant agent.

Multiple channels communicate with the same underlying AI handler.

                         CUSTOMER
                            │
              ┌─────────────┼─────────────┐
              │             │             │
           Telegram       Email           Web
              │             │             │
              └─────────────┼─────────────┘
                            │
                       Caspian SDK
                            │
                    Restaurant AI Handler
                            │
              ┌─────────────┴─────────────┐
              │                           │
        Featherless AI              Restaurant Logic
                                          │
                                   ┌──────┴──────┐
                                   │             │
                                MongoDB      Admin Panel

This means we don't need completely separate restaurant logic for Telegram and Email.

The agent has one core brain and multiple ways for customers to reach it.

The communication channel should not define the intelligence of the agent.

✨ What It Does

👤 Customer Features

Customers can interact with Gourmet Chef AI through:

  • 📱 Telegram
  • 📧 Email
  • 🌐 Web

Customers can:

  • 🍽️ Browse the complete restaurant menu
  • 🔎 Search for specific dishes
  • 💬 Ask natural-language menu questions
  • 🤖 Get personalized food recommendations
  • 🛒 Place food orders
  • 📦 Track orders
  • 🪑 Reserve tables
  • 📅 Check reservation status
  • ❌ Cancel or manage reservations
  • ⭐ Submit reviews
  • ❓ Ask restaurant-related questions
  • 🔔 Receive order and reservation updates

👨‍💼 Admin Features

The restaurant owner gets a dedicated Admin Dashboard.

The admin can:

  • ➕ Add menu items
  • ✏️ Edit menu items
  • 🗑️ Remove menu items
  • 🟢 Manage menu availability
  • 📦 View incoming orders
  • 🔄 Update order status
  • 🪑 View reservations
  • 🔄 Update reservation status
  • 👥 Manage customers
  • ⭐ Review customer feedback
  • ⚙️ Manage restaurant information
  • 📊 Monitor restaurant operations

🧠 AI Intelligence

We use Featherless AI as the inference layer for natural-language understanding and responses.

The AI can understand requests such as:

"I want something spicy under ₹500."

Instead of simply generating a random answer, the system retrieves the actual restaurant menu from MongoDB and uses that data to provide a relevant response.

🔄 AI Decision Flow

Customer Request
       ↓
Understand Intent
       ↓
Extract Requirements
       ↓
Query Restaurant Data
       ↓
MongoDB
       ↓
Relevant Menu Items
       ↓
Featherless AI
       ↓
Natural-Language Response

This approach helps prevent the AI from inventing:

  • ❌ Dishes
  • ❌ Prices
  • ❌ Availability

MongoDB remains the source of truth for restaurant data.

🍕 Smart Menu Recommendations

One of the challenges we specifically addressed was making the AI understand what the restaurant actually offers.

Instead of hardcoding recommendations, menu queries are dynamically connected to the restaurant database.

A customer can ask:

"Show me all pizzas."

or:

"I want vegetarian food under ₹500."

or:

"Recommend something spicy."

The system searches the actual available menu and responds according to the user's requirements.

If a requested item is unavailable, the agent does not silently replace it with a random dish.

Instead, it politely informs the customer and can suggest relevant available alternatives.

This ensures that the AI recommends what the restaurant actually has.

🛒 Smart Food Ordering

Customers can place orders through the supported channels.

For example:

"I want 2 Margherita pizzas and one Coke."

The system checks the actual menu data and creates the order using the corresponding database items.

📋 Order Information

Each order stores important information such as:

Field Description
🆔 Order ID Unique order identifier
👤 Customer Customer information
🍕 Items Ordered food items
🔢 Quantities Quantity of each item
💰 Prices Price of each item
💵 Total Amount Total order value
📊 Order Status Current order status
📱 Communication Channel Channel used by customer

The requested food is validated against the actual menu before it is added to the order.

If an item is unavailable, the system politely informs the customer instead of automatically replacing it with another unrelated dish.

🪑 Table Reservations

Customers can reserve tables through Telegram, Email, or Web.

For example:

"Book a table for 4 people tomorrow at 8 PM."

The system processes the request and creates a reservation containing:

  • 👤 Customer Details
  • 📅 Date
  • 🕐 Time
  • 👥 Party Size
  • 📝 Special Requests
  • 🆔 Reservation ID
  • 📊 Reservation Status

The customer's confirmation is sent through the available communication channel.

If the customer provides an email address, an email confirmation can also be sent.

If no email is provided, the system does not invent or use a default customer email.

🔔 Real-Time Notifications

Restaurant operations do not end when an order or reservation is created.

The system also handles status updates.

For example, an order can move through:

🟡 Confirmed
      ↓
🟠 Preparing
      ↓
🔵 Ready
      ↓
🟣 Out for Delivery
      ↓
🟢 Delivered

When an admin changes an order status, the customer can receive an update through the communication channel available for them.

Similarly, reservation status changes can be communicated to customers.

This turns the AI from a simple chatbot into a real operational assistant.

👨‍💼 Admin Dashboard

The restaurant owner has a dedicated Admin Panel.

From the dashboard, the owner can manage restaurant operations without directly interacting with the database.

🛠️ Restaurant Management

The admin can:

  • ➕ Add menu items
  • ✏️ Edit menu items
  • 🗑️ Remove menu items
  • 🟢 Manage availability
  • 📦 View incoming orders
  • 🔄 Update order status
  • 🪑 View reservations
  • 🔄 Update reservation status
  • 👥 Manage customers
  • ⭐ Review customer feedback
  • ⚙️ Manage restaurant information
  • 📊 Monitor restaurant activity

Changes made by the admin are reflected in the system's operational data.

📧 Email Experience

We wanted the email experience to feel like a real restaurant service rather than an automated developer notification.

Reservation confirmations, order confirmations, and notifications use structured and responsive email templates.

Important information such as:

  • 🆔 Order ID
  • 🪑 Reservation ID
  • 📅 Date
  • 🕐 Time
  • 🍽️ Items
  • 🔢 Quantity
  • 💰 Total
  • 📊 Status

is presented clearly so customers and restaurant owners can understand the information quickly.

The goal is to make every email feel professional, readable, and restaurant-ready.

🏗️ How We Built It

The project uses a full-stack architecture.

🎨 Frontend

  • React
  • Modern responsive UI
  • Customer-facing restaurant interface
  • Admin dashboard

⚙️ Backend

  • Node.js
  • Express
  • REST APIs
  • Centralized business logic

🗄️ Database

  • MongoDB

MongoDB stores:

  • Menu
  • Users
  • Orders
  • Reservations
  • Reviews
  • Restaurant information

🧠 AI

  • Featherless AI

Used for:

  • Natural-language understanding
  • Recommendations
  • Restaurant conversations
  • Intent processing
  • Context-aware responses

🌐 Communication

  • Caspian SDK
  • Telegram
  • Email
  • Web

🔄 Complete System Architecture

                         CUSTOMER
                            │
              ┌─────────────┼─────────────┐
              │             │             │
           Telegram       Email           Web
              │             │             │
              └─────────────┼─────────────┘
                            │
                       Caspian SDK
                            │
                    Single AI Handler
                            │
              ┌─────────────┴─────────────┐
              │                           │
        Featherless AI              Restaurant Logic
                                          │
                                  ┌───────┴───────┐
                                  │               │
                               MongoDB        Admin Panel
                                  │               │
                                  └───────┬───────┘
                                          │
                              ┌───────────┴───────────┐
                              │                       │
                           Orders                Reservations
                              │                       │
                              └───────────┬───────────┘
                                          │
                                          ↓
                                    Notifications
                                          │
                                          ↓
                                      CUSTOMER

🔄 How It Works

The complete system can be summarized as:

Customer
   ↓
Telegram / Email / Web
   ↓
Caspian SDK
   ↓
AI Handler
   ↓
Intent Understanding
   ↓
Restaurant Business Logic
   ↓
MongoDB
   ↓
Real Restaurant Data
   ↓
Action
   ↓
Notification
   ↓
Customer

In simple words:

Customer → Channel → Caspian → AI → Restaurant Logic → MongoDB → Action → Notification → Customer

The AI understands what the customer wants, the application checks the real restaurant data, performs the required action, and communicates the result back to the customer.

🛠️ Technology Stack

Technology Purpose
React Customer interface and Admin Dashboard
Node.js Backend runtime
Express.js REST API and backend services
MongoDB Restaurant and operational data
Caspian SDK Multi-channel AI communication
Featherless AI AI inference and natural-language understanding
Telegram Customer communication
Email Customer and owner communication

🚀 Future Improvements

The project can be extended with:

  • 📱 WhatsApp integration
  • 💳 Online payments
  • 🚚 Delivery partner integration
  • 📦 Inventory management
  • 📊 Advanced restaurant analytics
  • 👥 Staff management
  • 🎁 Customer loyalty programs
  • 📢 Automated marketing campaigns
  • 🎙️ Voice-based restaurant ordering
  • 🌍 Multi-language support
  • 🏪 Multi-restaurant SaaS support

The long-term vision is to allow multiple restaurants to use the same platform while keeping each restaurant's:

  • Menu
  • Customers
  • Orders
  • Reservations
  • AI configuration
  • Branding
  • Admin panel

completely separate.

🎯 Vision

We don't want Gourmet Chef AI to be just another restaurant chatbot.

We want it to feel like a real AI employee for a restaurant:

  • 🌐 Available across channels
  • 🧠 Connected to real restaurant data
  • ⚡ Able to take real actions
  • 💬 Able to communicate with customers
  • 📊 Able to help restaurant owners run their business

Gourmet Chef AI doesn't just answer. It understands, acts, and communicates.

🏆 Built for the Caspian Buildathon

Gourmet Chef AI was built around the core idea of the Caspian Buildathon:

Build an AI agent that can reach anyone.

Instead of creating an AI that waits for a user to open a chat window, we gave the restaurant agent multiple communication channels.

The same restaurant intelligence can communicate through Telegram, Email, and Web while interacting with real restaurant data and performing real operations.

It doesn't just answer. It acts.

❤️ Final Thought

Gourmet Chef AI combines AI, real restaurant data, automation, and multi-channel communication into one system.

Our goal is simple:

Give every restaurant an AI employee that can communicate, understand, act, and keep the restaurant team informed.

🍽️ Gourmet Chef AI — Your Restaurant's AI Employee

Built With

Share this project:

Updates

Submission history