WhatsApp AI Customer Support Bot Setup Guide

What This Agent Does

This workflow creates an intelligent customer support bot that automatically responds to WhatsApp messages with AI-powered assistance. When customers message your business, the bot instantly classifies their inquiry (product questions, contact requests, or general support), routes them to the appropriate response handler, and delivers personalized answers—all without manual intervention.

Key benefits include:

• 24/7 automated support that responds instantly to customer inquiries
• Reduced support team workload by handling routine questions automatically
• Improved customer satisfaction with immediate, intelligent responses
• Scalable customer engagement without hiring additional support staff
• Seamless AI integration that learns from your business context

Target use cases: E-commerce businesses, service providers, SaaS companies, restaurants, and any organization receiving high volumes of WhatsApp customer inquiries.

Who Is It For

This workflow is ideal for:

• Small to medium businesses looking to automate customer support
• E-commerce stores managing product inquiries and order questions
• Service-based companies handling appointment and contact requests
• Support teams wanting to reduce repetitive question handling
• Growth-focused organizations needing scalable customer communication

No coding experience required—TaskAGI handles all the technical complexity.

Required Integrations

WhatsApp Business

Why it's needed: This integration enables your bot to receive incoming WhatsApp messages and send responses directly through WhatsApp Business API, ensuring your customers receive support on their preferred messaging platform.

Setup steps:

1. Access WhatsApp Business Platform

   • Visit business.facebook.com
   • Log in with your Facebook Business account (create one if needed)

2. Create or Link a WhatsApp Business Account

   • Navigate to "WhatsApp" in the left sidebar
   • Click "Get Started" or select your existing business account
   • Verify your phone number (this becomes your business WhatsApp number)

3. Generate API Credentials

   • Go to Settings → API Setup
   • Create a new system user with "Admin" access
   • Generate a permanent access token (save this securely)
   • Note your Phone Number ID and Business Account ID

4. Configure Webhook in WhatsApp Settings

   • In WhatsApp settings, go to Configuration → Webhooks
   • Set your webhook URL (TaskAGI will provide this after creating the workflow)
   • Subscribe to messages and message_status events
   • Verify the webhook token

5. Connect to TaskAGI

   • In TaskAGI, navigate to Integrations → WhatsApp Business
   • Enter your Access Token
   • Enter your Phone Number ID
   • Enter your Business Account ID
   • Click "Test Connection" to verify
   • Save the integration

How to obtain API keys/credentials:

• Access Token: Generated in Facebook Business Manager under System Users
• Phone Number ID: Found in WhatsApp Settings → Phone Numbers
• Business Account ID: Located in WhatsApp Settings → Business Account Info

Configuration Steps

Step 1: WhatsApp Webhook (Node 6220)

This is your trigger node—it listens for incoming WhatsApp messages.

Configuration:

• Webhook Type: Select "WhatsApp Business"
• Integration: Choose the WhatsApp Business integration you just configured
• Events to Listen: Ensure "Incoming Messages" is enabled
• No additional parameters needed—this node automatically captures incoming message data

What it captures:

• Sender's phone number (chatId)
• Message text content
• Message timestamp
• Sender's profile information

Step 2: Parse WhatsApp Message (Node 6221)

This node extracts and structures the raw WhatsApp data into usable information.

Configuration:

• Input: Connect from WhatsApp Webhook (Node 6220)
• Function Type: Select "Parse Message"
• Extract Fields:
  • messageText - The customer's message content
  • senderPhone - Customer's phone number
  • chatId - Unique conversation identifier
  • timestamp - When the message was received

Example output structure:

{
  "messageText": "Do you have the blue shirt in size M?",
  "senderPhone": "+1234567890",
  "chatId": "1234567890",
  "timestamp": "2024-01-15T10:30:00Z"
}

Step 3: Classify User Intent (Node 6222)

This node analyzes the customer message to determine what they're asking about.

Configuration:

• Input: Connect from Parse WhatsApp Message (Node 6221)
• Analysis Type: Select "Intent Classification"
• Intent Categories: Define three main categories:
  • product - Questions about products, availability, pricing
  • contact - Requests for contact info, appointments, support
  • general - Other inquiries or greetings

Example classification:

Input: "What's your phone number?"
Output: { intent: "contact", confidence: 0.95 }

Step 4: Route by Intent (Node 6223)

This switch node directs the message to the appropriate response handler based on classified intent.

Configuration:

• Input: Connect from Classify User Intent (Node 6222)
• Switch Field: Select intent
• Case 1 - Product Intent:
  • Condition: intent == "product"
  • Route to: Generate Product Response (Node 6224)
• Case 2 - Contact Intent:
  • Condition: intent == "contact"
  • Route to: Generate Contact Response (Node 6225)
• Default Case:
  • Route to: Generate Default Response (Node 6226)

Step 5: Generate Product Response (Node 6224)

Handles product-related inquiries with templated responses.

Configuration:

• Input: From Route by Intent (Node 6223) - Product case
• Response Template: Create a template like:

  Thank you for your interest! 
  [PRODUCT_INFO]
  For more details, visit our website or reply with your specific question.
  

• Data Source: Connect to your product database or knowledge base
• Output: Structured response ready for AI enhancement

Step 6: Generate Contact Response (Node 6225)

Handles contact and support requests with business information.

Configuration:

• Input: From Route by Intent (Node 6223) - Contact case
• Response Template: Include:
  • Business hours
  • Phone number
  • Email address
  • Website URL
  • Appointment booking link (if applicable)
• Example:

  We'd love to help! Here's how to reach us:
  📞 +1-800-SUPPORT
  📧 support@company.com
  🕐 Hours: Mon-Fri 9AM-6PM EST
  

Step 7: Generate Default Response (Node 6226)

Handles unclassified or general inquiries.

Configuration:

• Input: From Route by Intent (Node 6223) - Default case
• Response Template:

  Thanks for reaching out! I'm here to help.
  Could you provide more details about your question?
  

• Tone: Friendly and open-ended to encourage clarification

Step 8: Build AI System Prompt (Node 6227)

Creates the AI context that powers intelligent responses.

Configuration:

• Input: Receives outputs from all three response generators (6224, 6225, 6226)
• System Prompt Content: Define your AI's personality and guidelines:

  You are a helpful customer support assistant for [COMPANY_NAME].
  - Be friendly and professional
  - Provide accurate information
  - Offer solutions, not just answers
  - Keep responses concise (under 160 characters for WhatsApp)
  - If unsure, offer to escalate to a human agent
  

• Context Variables: Include:
  • Customer's original message
  • Classified intent
  • Pre-generated response template
  • Business information

Step 9: AI Agent (Node 6228)

The intelligent core that generates personalized responses.

Configuration:

• Input: Connect from Build AI System Prompt (Node 6227)
• AI Model: Select "GPT-4" or your preferred model
• Temperature: Set to 0.7 (balanced between creative and consistent)
• Max Tokens: Set to 150 (keeps WhatsApp responses concise)
• Instructions:
  • Use the system prompt provided
  • Reference the customer's original message
  • Incorporate the pre-generated response template
  • Maintain brand voice and tone

Step 10: Format AI Response (Node 6229)

Optimizes the AI response specifically for WhatsApp.

Configuration:

• Input: Connect from AI Agent (Node 6228)
• Formatting Rules:
  • Maximum 160 characters per message (WhatsApp standard)
  • Remove markdown formatting
  • Convert emojis to text if needed
  • Add line breaks for readability
  • Include call-to-action buttons if supported

Example transformation:

Before: "We have **blue shirts** in sizes S, M, L, XL. Price: $29.99"
After: "We have blue shirts in S, M, L, XL. Price: $29.99"

Step 11: Send WhatsApp Response (Node 6230)

Delivers the final response back to the customer.

Configuration:

• Input: Connect from Format AI Response (Node 6229)
• Integration: Select your WhatsApp Business integration
• Recipient Field: [[nodes.6229.chatId]] (automatically uses customer's chat ID)
• Message Type: Select "Text Message"
• Message Content: Use the formatted response from Node 6229
• Enable Delivery Notifications: Toggle ON to track message delivery

Testing Your Agent

Test Execution Steps

1. Prepare Your Test Environment

• Ensure WhatsApp Business integration is active
• Have a test WhatsApp number ready (your personal number)
• Verify webhook is receiving messages (check TaskAGI logs)

2. Send Test Messages

Send these test messages from your WhatsApp to your business number:

Test 1 - Product Intent:

"Do you have the red jacket in size L?"

✅ Expected result: Product-focused response with availability info

Test 2 - Contact Intent:

"What are your business hours?"

✅ Expected result: Contact information and hours displayed

Test 3 - General Intent:

"Hi there!"

✅ Expected result: Friendly greeting with offer to help

3. Verify Each Node

4. Monitor Success Metrics

• ✅ Response time: Under 5 seconds
• ✅ Message delivery: Confirmed in WhatsApp
• ✅ Intent accuracy: Correct routing for test messages
• ✅ Response quality: Relevant and helpful answers
• ✅ No errors: Check TaskAGI logs for failures

5. Troubleshooting Common Issues

Congratulations! Your WhatsApp AI Customer Support Bot is now live and ready to delight your customers with instant, intelligent support. Monitor performance regularly and refine your prompts based on real customer interactions.