Z-API + OpenAI — WhatsApp Chatbot Example

Explore how to connect WhatsApp to OpenAI using Z-API and Node.js. This example receives messages through a webhook, generates AI responses, and sends them back to WhatsApp.

Compatibility Notice

This example uses the retired OpenAI model text-davinci-003. Update the OpenAI integration in index.js to a supported model and compatible API request format before testing AI responses.

See the OpenAI documentation for current integration guidance.

Prerequisites

• Node.js and npm installed.
• A Z-API instance connected to WhatsApp.
• An OpenAI API key.
• A public HTTPS URL for receiving webhooks. For local development, you can use ngrok.

Getting Started

1. Clone the repository

git clone https://github.com/Z-API/chat-gpt-example.git
cd chat-gpt-example

2. Install dependencies

npm install

3. Configure your environment

Create a .env file in the project root:

OPEN_AI_API_KEY=your_openai_api_key
Z_API_INSTANCE_ID=your_zapi_instance_id
Z_API_INSTANCE_TOKEN=your_zapi_instance_token

Get your instance credentials from your Z-API account and your API key from the OpenAI platform.

Keep your credentials private and do not commit your .env file.

4. Start the application

npm start

For development with automatic restarts:

npm run dev

The server listens on port 3000.

5. Configure the webhook

For local testing, install and configure ngrok, then expose port 3000:

ngrok http 3000

Copy the public HTTPS URL provided by ngrok and append /on-new-message:

https://YOUR-NGROK-DOMAIN/on-new-message

Set this address as the received-message webhook in your Z-API instance.

Keep the application and ngrok running while testing. If your tunnel URL changes, update the webhook address in Z-API.

6. Start a conversation

After updating the OpenAI integration and completing the setup:

1. Send !gpt from another WhatsApp account to the number connected to your Z-API instance.
2. Wait for the welcome message.
3. Send a text message to continue the conversation.

How It Works

1. Z-API forwards an incoming WhatsApp message to /on-new-message.
2. The application starts a chat session when it receives !gpt.
3. Subsequent text messages are sent to OpenAI.
4. The generated response is sent back through Z-API.

Development Notes

• Conversations are stored in memory and are lost when the application restarts.
• The welcome and error messages are in Portuguese and can be customized in index.js.
• This is a learning example. Review API compatibility, authentication, error handling, and persistent storage before adapting it for production.

Resources

• Z-API Documentation
• Z-API Website
• OpenAI Documentation
• ngrok