Skip to main content
A chat agent is a conversational AI assistant that visitors can interact with on your website through a chat widget. This guide walks you through the complete process of creating and setting up your chat agent from start to finish.

​
Prerequisites

Before you begin, ensure you have:
  • An account and are logged in
  • An organization set up (created automatically when you sign up)
  • A project created (see the “How to Create a Project” tutorial if needed)
  • Permission to edit the project (you have this if you’re the project creator or admin)

​
Access Project Settings

1

Navigate to your project

After logging in, go to the Projects page from the main navigation. Find and click on the project you want to configure as a text chat agent. You’ll be taken to the project dashboard.
2

Open Chat Agent

Look for the Chat Agent option in the sidebar of your project dashboard. Click on Chat Settings to open the configuration page. This is where you’ll configure your chat agent’s logic (by writing prompts) and update the UI to match your brand. You’ll see several tabs at the top: General, Chat Experience, AI Configuration, Appearance, Integration, and more.

​
Configure Basic Information

The General tab is usually selected by default. Here you’ll set up the basic information about your chat agent:

​
Project Name

The name of your chat agent project. The name should already be filled in from when you created the project. You can change it if needed. Use a clear, descriptive name.

​
Description

A brief explanation of what your chat agent does. Add a description that explains the chat agent’s purpose. Example: “Customer support chat agent for answering product questions”. This is optional but helpful.

​
Language

The primary language your chat agent will communicate in. Click the Language dropdown and select the language that matches how your chat agent will communicate with users. Examples: English, Spanish, French, German, Chinese, and many more.
Choose the language that most of your website visitors speak.

​
Timezone

The timezone your chat agent will use for date and time information. Click the timezone dropdown and select your timezone from the list. This helps the chat agent provide accurate time-based information.
After filling these fields, click Save at the bottom of the page to save your changes.

​
Set Up Chat Experience

Click on the Chat & Lead Capture tab to configure how your chat agent interacts with users:

​
Welcome Message

The first message users see when they start a conversation. Type a friendly greeting message. Example: “Hi! I’m here to help. What can I assist you with today?” Keep it short, friendly, and welcoming.
This message appears when users first open the chat.

​
Hello Message (Widget Bubble)

The message shown on the chat agent button before users click it. Type a short message that appears on the chat bubble. Example: “Hello! How can I help you?” This is what users see before they click to start chatting. Keep it brief (usually 1-2 sentences).

​
Lead Capture Settings

A feature that collects user information during conversations. Toggle the Lead Capture switch to ON if you want to collect user information. Toggle it to OFF if you don’t need to collect leads. When enabled: You can set up questions to ask users (like name, email, phone number).
When disabled: The chat agent won’t ask for personal information.
1

Add lead questions

If Lead Capture is enabled, click Add Question to create questions that collect user information. For each question, you’ll need to:
  • Enter the question text (e.g., “What’s your name?”)
  • Select what information it captures (Name, Email, Phone, etc.)
You can add multiple questions.
Common questions include name, email, phone number, company name.
2

Set thank you message

Type a thank you message that appears after users complete the lead capture questions. Example: “Thank you for providing your information! How can I help you today?” This appears after users answer all the lead capture questions.

​
Quick Questions

Pre-written questions users can click to start conversations quickly. Click Add Question to create quick question buttons. Type a question (e.g., “What are your business hours?”). You can add up to 5 quick questions. These appear as clickable buttons when users open the chat.
Use questions that are commonly asked by your customers.
After configuring these settings, click Save to save your changes.

​
Configure AI Behavior

Click on the AI Configuration tab to set up how your chat agent thinks and responds:

​
Conversation Model (LLM)

The AI model that powers your chat agent’s conversations. Click the Model dropdown and select an AI model (e.g., “gpt-4o-mini”, “gpt-4”, etc.).
Start with the default model (usually “gpt-4o-mini”) - it’s cost-effective and works well for most chat agents.

​
Temperature

Controls how creative or focused your chat agent’s responses are. Use a slider or number input to set the temperature.
  • Lower values (0-0.3): More focused, consistent responses
  • Higher values (0.7-1.0): More creative, varied responses
Start with 0.3 for customer support, or 0.7 for more conversational chat agents.

​
AI Actions (Tools)

Special capabilities that extend your chat agent beyond basic conversation. You’ll see a list of available AI tools and actions. Each tool has a checkbox next to it with a description. Check the boxes for the tools you want your chat agent to use. Common tools include:
  • Book Appointment: Allows the chat agent to schedule appointments (requires Calendly setup)
  • Track Shipment: Enables tracking of shipments (requires Shiprocket setup)
  • Human Handover: Allows the chat agent to transfer conversations to human agents
  • And more based on what’s available in your organization
Only enable tools that you actually need - each tool adds functionality but may require additional setup.
Some tools (like Book Appointment) may require you to configure credentials first. If you check a tool that needs setup, a popup will appear to guide you through the configuration.

​
Prompt Template (Agent Template)

Pre-written instructions (templates) that tell your chat agent how to behave, or you can write your own custom instructions.
1

Select a template

Click the Select Template dropdown. You’ll see a list of available prompt templates. Each template has a name and description. Select a template that matches your chat agent’s purpose.
Templates are pre-configured prompts designed for specific use cases (e.g., customer support, sales, FAQ assistant).
2

Customize the prompt (optional)

After selecting a template, you’ll see the prompt content in a text editor below. You can edit this prompt to customize it for your specific needs. The prompt tells your chat agent:
  • What it should do
  • How it should respond
  • What tone to use (professional, friendly, casual)
  • What topics it should focus on
  • What it should avoid
Example prompt: “You are a helpful customer support assistant for an online store. Answer questions about products, shipping, and returns. Be friendly and professional. If you don’t know something, say so and offer to connect the user with a human agent.”
Be specific and clear in your instructions - this determines how well your chat agent performs.
3

Preview (optional)

Click the Preview button to see how your prompt will look when formatted. This helps you check if your prompt is well-structured.
After setting up the AI configuration, click Save to save your changes.

​
Configure Notifications

Click on the Notifications tab to set up email notifications for your chat agent:

​
Notification Email

The email address that will receive notifications about your chat agent’s activity. Click in the Email Address field and type the email address where you want to receive notifications. This email will receive project notifications and updates. Examples: You might want to use a support email, your business email, or a dedicated notifications email.
Use an email address that you check regularly so you don’t miss important notifications.
After setting up notifications, click Save to save your changes.

​
Configure Security Settings

Click on the Security tab to set up rate limits and security controls:

​
Rate Limits

Controls that limit how many requests your chat agent can handle to prevent abuse and control costs. Project Level Rate Limits:
  • Requests Per Minute: Enter the maximum number of requests allowed per minute for your entire project
  • Requests Per Hour: Enter the maximum number of requests allowed per hour for your entire project
  • These limits apply to all users combined across your project
User Level Rate Limits (Lead Capture):
  • Leads Per Minute: Enter the maximum number of lead capture requests allowed per minute per user
  • Leads Per Hour: Enter the maximum number of lead capture requests allowed per hour per user
  • These limits apply to individual users to prevent spam
Recommended values:
  • Project per minute: 100 (adjust based on your traffic)
  • Project per hour: 500 (adjust based on your traffic)
  • Lead per minute: 20 (prevents spam)
  • Lead per hour: 100 (prevents abuse)
Start with the default values and adjust based on your actual usage. Higher limits allow more traffic but may increase costs.
After configuring security settings, click Save to save your changes.

​
Set Up Human Handover

Click on the Human Handover tab to enable transferring conversations to human team members:

​
Enable Human Handover

A feature that allows your chat agent to transfer conversations to human team members when needed. Toggle the Enable Human Handover switch to ON. When enabled, your chat agent can transfer conversations to human agents. This is useful when the chat agent can’t answer a question or when a user requests to speak with a human.

​
Select Team Members

Choose which team members can receive handover notifications and take over conversations. After enabling handover, you’ll see a list of your project’s team members. Check the boxes next to the team members who should receive handover notifications. These team members will be notified when a conversation is transferred to them. You can select multiple team members.
Select team members who are available to handle customer inquiries and have the right expertise.
After setting up human handover, click Save to save your changes.

​
Customize Appearance

Click on the Appearance tab to make your chat agent look the way you want:

​
Bot Name

The name your chat agent will use in conversations. Type a name for your chat agent. Example: “Support Bot”, “Sarah”, “AI Assistant”. This name appears in the chat interface.

​
Bot Icon

An icon or image that represents your chat agent. You can select from available icons or upload a custom image. Choose something that matches your brand. This appears in the chat widget.

​
Primary Theme Color

The main color used in your chat agent widget. Click the color picker and choose a color that matches your website’s branding. You can enter a hex color code (like #204677) or pick from the color picker. This color is used for buttons and highlights in the chat widget.

​
Widget Position

Where the chat agent button appears on your website. Select Right to show the chat agent button in the bottom-right corner. Select Left to show it in the bottom-left corner. Most websites use the right side.
After customizing the appearance, click Save to save your changes.

​
Integration

Once you have configured your chat agent, you need to add it to your website. Check out the Integration Guide for detailed instructions on how to:
  • Get your integration code
  • Add it to HTML, WordPress, or Website Builders
  • Test your chat agent
  • Troubleshoot common integration issues

​
Tips for Success

​
Using AI Actions

  • Start simple: Only enable AI Actions that you actually need.
  • Read descriptions: Each tool has a description - read it to understand what it does.
  • Configure credentials: Some tools (like Book Appointment) require you to set up credentials first.
  • Test after enabling: After enabling a tool, test it to make sure it works as expected.

​
Choosing and Customizing Prompt Templates

  • Start with templates: Use a pre-made template that matches your use case.
  • Customize carefully: When editing a template, make small changes first and test them.
  • Be specific: Clearly describe what your chat agent should do.
  • Set the tone: Specify if it should be formal, friendly, casual, etc.

​
Lead Capture Best Practices

  • Only ask for information you actually need.
  • Don’t ask too many questions (3-5 is usually enough).
  • Make questions clear and easy to understand.

​
What’s Next?

After your chat agent is set up and working:
  1. Monitor Conversations: Check how users are interacting with your chat agent.
  2. Improve Over Time: Update your prompt based on real conversations.
  3. Add Knowledge Sources: Consider adding documents or FAQs to help the chat agent answer better.
  4. Set Up Notifications: Configure email notifications to stay informed.

​
Troubleshooting Configuration

What to check:
  • Verify that the AI configuration (model, prompt) is saved correctly.
  • Make sure you’ve clicked Save after making changes.
  • Check if there are any error messages in the browser console.
How to improve:
  • Improve the prompt: Write more detailed and specific instructions.
  • Adjust temperature: Try different temperature values.
  • Test and iterate: Try different prompts and see what works best.
What to check:
  • Make sure “Lead Capture” is enabled in the Chat Experience tab.
  • Verify that lead questions are added and configured correctly.
  • Check that the thank you message is set.

​
Summary

Creating a text chat agent involves these main steps:
  1. Configure Basic Info: Set the name, language, and timezone.
  2. Design Experience: Set up welcome messages, lead capture, and quick questions.
  3. Setup AI: Choose a model and provide instructions (prompts).
  4. Customize Appearance: Match the chat agent’s look to your brand.
  5. Integrate: Add the code to your website and test it.
For a deep dive into the integration process, see our Integrate Chat Agent guide.