How to Set Up Customer Service AI Chatbot

Engage customers instantly with an AI chatbot trained on your business data, with seamless human handover when needed. Follow the steps below to properly install and configure your Customer Service AI Chatbot.

Step-by-Step Setup Guide

Go to the Automation Menu

• Open your app dashboard

• Click Automation

• Select Add Rule

This is where you’ll install and manage your chatbot automation.

Install the Customer Service AI Chatbot

• Locate Customer Service GPT for Your Business – AI Chatbot

• Click Install

You can install it for:

• WhatsApp

• Instagram

• Messenger

• Telegram

• Live Chat

The setup process is the same across supported channels.

Confirm Installed Rules

After installation, you’ll see two automation rules:

• custom-training-ai-bot

• ai-to-human-transfer

These work together:

• The AI bot handles conversations

• The transfer rule ensures smooth handoff to your team when needed

Configure the Custom Training AI Bot

Open the custom-training-ai-bot rule and scroll to the Action section to configure the chatbot properly.

1. Customize the Prompt

A default customer service prompt is provided.

You can:

• Update it to match your brand tone

• Define how the bot should respond

• Add business rules

• Set response style (formal, friendly, concise, etc.)

Example:

• “Answer politely and professionally.”

• “If unsure, ask for clarification.”

• “Transfer to a human if customer requests an agent.”

2. Model Configuration

Set your AI parameters:

• OpenAI Model

• Temperature (creativity level)

• Max tokens (response length)

• Conversation memory settings

For customer support, lower temperature is usually recommended for more accurate responses.

3. Enable Built-In Tools

You can allow the AI to:

• Update contact information

• Search the web

• Add internal notes for your team

These tools help the AI act beyond simple replies.

4. Integrate Your APIs (Function Calling)

You can connect external systems using function calling.

Provide:

• API endpoint URL

• HTTP method (GET, POST, etc.)

• Headers (authorization if required)

• Request/response schema

This allows your AI to:

• Check order status

• Book appointments

• Pull customer data

• Create tickets

5. Use the Playground to Test

Before going live:

• Use the Playground to test chatbot responses

• Simulate real customer conversations

• Verify API functions work properly

• Adjust prompt if necessary

Testing ensures smooth performance before customers interact with the bot.

6. Load Your Business Data (Embedding Config)

To train the chatbot with your actual business knowledge:

• Go to Embedding Config

• Upload documents, FAQs, policies, or website content

• Ensure the knowledge base reflects accurate information

This improves response accuracy and relevance.

7. Save the Rule

After configuration:

• Click Save

• Confirm all settings are correct

8. Enable AI-to-Human Transfer

Open the ai-to-human-transfer rule.

This rule ensures:

• When AI cannot resolve a query

• Or when customer requests a human

• The conversation is transferred to your support team

This guarantees a seamless customer experience.

Best Practices

• Clearly define your AI boundaries

• Always include human fallback

• Train with accurate business data

• Keep responses concise and helpful

• Regularly test after updating prompts