WhatsApp AI Business Dashboard

SONAQO Project - WhatsApp AI Assistant - Product Database - Admin Handover

This project is a prototype dashboard for helping a business reply to WhatsApp customers with AI. In very simple words: a customer sends a WhatsApp message, AI reads the question, AI checks the product data, and AI replies automatically. If the conversation needs a human, admin can take over the chat.

Project Goal

The goal is to make WhatsApp business communication cleaner. The business can store product data, view chat history, reply to customers, and control when AI should answer or when admin should step in. This helps admin avoid repeating the same answers from scratch.

The Simple Version

1. Customer chats with the business WhatsApp number.
2. The system receives the customer message.
3. AI reads the message and searches the product database.
4. If the question is simple, AI replies automatically.
5. If the customer needs human help, admin clicks the customer and replies manually.
6. Admin can return the chat back to AI after the issue is handled.

Problems Solved

• Customers often ask the same stock, price, and product alternative questions.
• Admin can reply late when many chats come in at the same time.
• Product data is often scattered in notes, old chats, or spreadsheets.
• The business needs a clear rule: when AI replies and when admin takes over.

Solution Built

The solution is a WhatsApp AI Business Dashboard with three main sections: Chat, Product, and Admin. Chat is used to view conversation history. Product is used to store product data. Admin is used to manually reply to customers when AI is not enough.

Customer WhatsApp View

This section shows the final experience from the customer's side. Customers do not need to see the dashboard. They only chat normally on WhatsApp, then AI replies with clear product information based on the product data added by admin.

How to Use the System

1. Add Product Data

1. Open the Product tab.
2. Enter the product name, such as MCB Chint 1P 6A.
3. Fill in category, price, stock, and product alternative.
4. Click Save Product.
5. After it is saved, AI can use this data to answer customer questions.

2. Customer Sends a Message

1. Customer asks through WhatsApp, for example: "Is the fan ready?"
2. The message enters the system through Cloud WhatsApp API integration.
3. The system stores the message in the database.
4. AI reads the question and searches for matching product data.
5. AI replies automatically if the chat is still in AI Active mode.

3. Admin Takes Over

1. Open the Admin tab.
2. Click the customer from the left panel.
3. Write a manual reply in the customer reply box.
4. Click Send.
5. When admin is active, the bot stays silent so the customer does not receive double replies.
6. After the issue is handled, click Return to AI.

JSON Database

For the prototype, data can be stored in a JSON file so it is easy to understand and fast to develop. Think of it as a digital notebook: it stores product notes, chat notes, customer status, and bot settings. A sample structure is available in frontend/data/wa-ai-assistant-database-example.json.

{
  "products": [
    {
      "id": "prd-001",
      "name": "MCB Chint 1P 6A",
      "category": "mcb",
      "price": 35000,
      "stock": 10,
      "alternative": "MCB Schneider 1P 6A"
    }
  ],
  "messages": [
    {
      "customer_phone": "6282123067587",
      "sender": "user",
      "message": "Is the fan ready?",
      "created_at": "2026-06-05T11:29:00+07:00"
    }
  ],
  "handover": {
    "customer_phone": "6282123067587",
    "mode": "admin"
  }
}

Can Grow to PostgreSQL / MySQL

When the system is ready for real usage, the JSON database can be moved to PostgreSQL or MySQL. The logic stays the same, but the storage becomes stronger. Products go into a product table, chats go into a messages table, customers go into a customers table, and AI/admin status goes into a handover table.

• JSON: good for prototype, demo, and quick testing.
• PostgreSQL/MySQL: good for production, larger data, multiple admins, backup, and reporting.
• Possible tables: products, customers, messages, bot_settings, admin_handover, and conversation_logs.

Cloud WhatsApp API Integration

The Cloud WhatsApp API integration is handled by the developer. The developer connects the WhatsApp business number, incoming message webhook, send-message endpoint, database, and AI responder.

1. Developer prepares Meta Developer App and Cloud WhatsApp API.
2. Developer sets the webhook so customer messages enter the backend.
3. Backend stores messages in JSON or database.
4. AI reads the message and product data.
5. Backend sends the reply back to the customer's WhatsApp.
6. If admin takes over, backend stops AI replies for that customer.

Main Features

• Customer chat dashboard and message history.
• Total chats, user messages, and AI reply counts.
• Product data: name, category, price, stock, and alternatives.
• Admin handover mode so admin can reply manually.
• Return-to-AI mode so the bot can become active again after admin is done.
• JSON database that can grow into PostgreSQL or MySQL.

Business Benefits

• Customers get faster first responses.
• Admin does not need to repeat the same answers all day.
• Product data becomes cleaner and usable by AI.
• Admin still has full control when a conversation needs a human.
• The system can grow from a simple prototype into a production system.

This System Can Grow Into

• WhatsApp AI customer service
• AI product assistant
• Lead CRM and follow-up
• Multi-admin dashboard
• Order tracking and invoice flow
• Chat reporting and AI performance monitoring