Dynamic WhatsApp Chatbot with Twilio, OpenAI & Pinecone 🚀

This is a fully customizable WhatsApp chatbot built using Twilio, OpenAI, and Pinecone, allowing users to configure conversation flows, FAQs, and API integrations via a simple JSON-based setup.

🔹 Features

✅ Customizable Flow: Define your chatbot’s conversation flow in flow.json.
✅ AI-Powered FAQ Search: Uses Pinecone vector search for fast, relevant responses.
✅ Real-time API Calls: Fetch dynamic data from external APIs, configurable in flow.json.
✅ OpenAI Fallback: If no FAQ match is found, AI generates a response.
✅ WhatsApp-Ready: Built for Twilio WhatsApp API, instantly deployable.
✅ State Management: Uses Redis for seamless conversation tracking.
✅ Zero Code Changes Needed: Just update JSON files & run the bot!

---

📌 Installation

1️⃣ Clone the Repository

git clone https://github.com/your-username/whatsapp-chatbot.git
cd whatsapp-chatbot

2️⃣ Install Dependencies

npm install

3️⃣ Set Up Environment Variables

Create a .env file with:

TWILIO_ACCOUNT_SID=your_twilio_sid
TWILIO_AUTH_TOKEN=your_twilio_auth_token
TWILIO_PHONE_NUMBER=your_twilio_phone
OPENAI_API_KEY=your_openai_api_key
PINECONE_API_KEY=your_pinecone_api_key
PINECONE_INDEX_NAME=faqs
REDIS_URL=your_redis_url
PORT=3000

---

📌 Configuration

Define Chatbot Flow (flow.json)

Modify flow.json to customize the chatbot:

{
  "welcome": {
    "type": "plain",
    "message": "Welcome to our chatbot! How can I help?",
    "next": "main"
  },
  "main": {
    "type": "branch",
    "branches": [
      { "condition": "1", "next": "propertySearch" },
      { "condition": "2", "next": "faqNode" },
      { "condition": "3", "next": "support" }
    ]
  },
  "propertySearch": {
    "type": "api",
    "message": "Searching for properties...",
    "api": {
      "url": "https://api.example.com/properties",
      "method": "GET",
      "params": ["city", "budget"]
    },
    "next": "closingNode"
  },
  "faqNode": {
    "type": "faq",
    "next": "main"
  },
  "support": {
    "type": "plain",
    "message": "Contact support at support@example.com",
    "next": "main"
  }
}

Add FAQs (faqs.json)

[
  {
    "id": "faq_1",
    "question": "What is ZeroDeposit Rental?",
    "answer": "ZeroDeposit replaces traditional security deposits with a financial guarantee."
  },
  {
    "id": "faq_2",
    "question": "How much does the Rental Guarantee cost?",
    "answer": "The guarantee fee is 9% of the deposit amount."
  }
]

---

📌 Upload FAQs to Pinecone

node upload_faqs.js

---

📌 Run the Chatbot

node bot.js

---

📌 Webhook Setup for Twilio

1. Go to Twilio Console → Messaging → WhatsApp Senders
2. Set Webhook URL:

   https://your-server.com/webhook
   

3. Method: POST
4. Save and Test by sending a message to your Twilio number.

---

🚀 Deployment

Railway

1. Deploy on Railway.app
2. Add Environment Variables from .env
3. Deploy & Start the Bot!

Render

1. Push Code to GitHub
2. Deploy on Render.com
3. Expose Webhook

Self-Hosting

Run using PM2:

npm install -g pm2
pm2 start bot.js --name whatsapp-bot
pm2 save

---

📌 Next Steps

• ✅ Multi-language Support (Use OpenAI translation)
• ✅ Analytics Dashboard (Track chatbot performance)
• ✅ Webhook Event Logging (Monitor responses)

---

👨‍💻 Contributing

Feel free to fork, modify, and submit PRs! Let's build smarter chatbots together. 🎉

---

🚀 Built with ❤️ using OpenAI, Twilio & Pinecone. 🚀