Tutorial: Automate WhatsApp Replies with ChatGPT and WOZTELL on Make
Tutorial: Automate WhatsApp Replies with ChatGPT and WOZTELL on Make
This advanced, step-by-step guide will teach you how to build a Make scenario that connects WOZTELL with OpenAI’s ChatGPT, so you can automatically reply to WhatsApp messages using AI.
You’ll learn how to watch for new messages, generate smart responses, and send them back to users—no coding required.
Overview of the Flow
Your automation will consist of three modules:
1. Watch New WhatsApp Message – Detect incoming messages on your WOZTELL WhatsApp channel.
2. ChatGPT Completion – Send the message content to OpenAI for response generation.
3. Send WhatsApp Session Message – Deliver the AI-generated reply back to the original sender.
Visual flow:
Step 1: Watch New WhatsApp Message (WOZTELL)
1. Add the module WOZTELL | Unleash WhatsApp – Watch New WhatsApp Message in your Make scenario.
2. Connect your WOZTELL account.
3. Choose the WhatsApp channel you want to monitor (your inbox).
4. This module will trigger every time a new message is received.
Step 2: Generate a Reply with ChatGPT (OpenAI)
1. Add the module OpenAI (ChatGPT, Whisper, DALL-E) – Create a completion (prompt) after the WOZTELL trigger.
2. Connect your OpenAI account (API key required).
Model Selection: Recommendations
Choose the model best suited to your use case:
- For general support/FAQ and standard automated replies:
➤ gpt-3.5-turbo
Fast, cost-effective, ideal for FAQs and basic support.
- For advanced support or premium conversations:
➤ gpt-4-turbo
Better contextual understanding, higher quality replies, suitable for valuable customer interactions.
- For the highest quality, most advanced and context-aware responses:
➤ gpt-4o
OpenAI’s latest flagship model (2024), best for complex or high-stakes use cases.
You can select the model from the “Model” dropdown in the OpenAI module in Make, but be aware that each model have their own cost, so check them before choosing one.
3. Messages Section:
- Role: Set to User.
- Text Content: Map the incoming WhatsApp message data from the previous module.
- Example prompt:
Please compose a detailed reply to this message: 1. Message Data: Text Content - You can personalize the prompt for your business needs, but keep the mapped data variable so ChatGPT analyzes the actual WhatsApp message.
Screenshot example:
Step 3: Send the AI Response Back via WhatsApp (WOZTELL)
1. Add the module WOZTELL | Unleash WhatsApp – Send WhatsApp Session Message.
2. Connect with your WOZTELL account.
3. Channel: Choose the same WhatsApp channel as in Step 1.
4. WhatsApp Number: Map the sender's phone number from the trigger module (1. Sender).
5. Message Type: Set to Text.
6. Text: Map the output from ChatGPT (2. Result).
Screenshot example:
Final Steps
- Be sure that you had click Save on all modules.
- Test your scenario by sending a WhatsApp message to your channel.
- Activate the scenario to make it live.
Best Practices
- Customize the prompt in Step 2 for tailored, high-quality responses. A good prompt take a lot of time to be created, but it's worth it at the end.
- Secure your API keys and tokens and restrict access. You have to keep the OpenAI keys safe somewhere you can get them back, so be careful with it.
- Test each step independently to ensure correct data mapping.
- Use Make’s logs for troubleshooting if responses aren’t delivered.
Result
Once activated, your scenario will automatically:
- Watch for new WhatsApp messages,
- Generate smart replies using ChatGPT (with your chosen model and prompt),
- Send responses back to the sender via WhatsApp.
This setup streamlines your WhatsApp support and engagement with AI-powered automation!