🐱 Winnie - WhatsApp AI Chatbot

🤖 AI Powered Chatbot using GPT-3

Chat on WhatsApp with an AI

Demo

🔨 Tech Stack

Server: Node.JS, Express

Database: Redis

API: WhatsApp Business Cloud API, OpenAI GPT-3

Deployment: Docker, Nginx, Linode

Misc: ElasticSearch, Kibana

⚙️ Project Structure

whatsapp-connector: Whatsapp connector is a microservice which connects a bot to WhatsApp API, handles messaging and incoming requests from WhatsApp

chatbot: Chatbot is a microservice which will connect to OpenAI GPT-3 model to answer user queries based on prompt

🥣 Meta Configuration

Multiple prerequisites and configuration needs to be done on the Meta Developers, please follow my below guide for details: https://blog.amitwani.dev/create-a-chatbot-using-whatsapp-cloud-api

🏃‍♂️ Running Application

• Run the application using docker-compose

  • Update Environment Variables

    • Fill necessary values in docker-compose-local.yml file inside docker folder for below environment variables:
      • VERIFY_TOKEN - Verify Token for WhatsApp Webhook
      • OPENAI_API_KEY - OpenAI API Key
      • WHATSAPP_TOKEN - WhatsApp Business Permanent/Temporary Authentication Token with whatsapp_business_messaging permission
      • WHATSAPP_SEND_MESSAGE_API - WhatsApp Cloud API Send Message API End Point with Phone Number ID
      • CHATBOT_URL - Chatbot microservice api url

  • Run command docker-compose -f docker-compose.local.yml up -d inside docker folder

• Run using npm

  • Install Dependencies

  Navigate to whatsapp-connector and chatbot to install their dependecies.

    npm install

  or

    yarn add

  • Update Environment Variables

    • Create .env file by copying .env.sample in both microservices
    • Fill necessary values for below environment variables:
      • VERIFY_TOKEN - Verify Token for WhatsApp Webhook
      • OPENAI_API_KEY - OpenAI API Key
      • WHATSAPP_TOKEN - WhatsApp Business Permanent/Temporary Authentication Token with whatsapp_business_messaging permission
      • WHATSAPP_SEND_MESSAGE_API - WhatsApp Cloud API Send Message API End Point with Phone Number ID
      • CHATBOT_URL - Chatbot microservice api url

  • Run the Server

      npm run start

🛳️ Deployment

Deploy with Docker easily using the Dockerfile provided in the respective services folder or the docker-compose.yml in the docker folder

👷 CI/CD

GitHub Actions is used to create a CI/CD workflow specified in the workflow.yml.

Workflow

• On every commit on main branch
• Build Docker Images of both microservices
• Push Docker Images of both microservices to docker repository
• Update docker-compose.yml by replacing environment varibles from GitHub secrets
• Copying docker-compose.yml and nginx.conf to the server
• Running docker-compose command to recreate containers
• Deployment is done

Acknowledgements

• Linode
• Hashnode

Feedback

If you have any feedback, please reach out to me at mtwn105@gmail.com