AI Agents: Automate Customer Conversations with AI

AI Agents are the core automation engine in ChatbotX. Each Agent uses a large language model (LLM) connected to your business data to respond to customers instantly, accurately, and around the clock — without requiring a live team member for every interaction. You can configure an Agent with a written prompt, business documents, external API functions, and MCP Server connections so it answers questions and takes action in ways that reflect exactly how your business operates.

​

How to Create a New AI Agent

​

Set a Bot Reply Delay

The Bot reply delay setting controls how long an Agent waits before sending a reply. This is especially useful when customers tend to send several short messages in quick succession — the Agent waits until the customer finishes typing, then composes a single reply that addresses everything instead of responding to each message separately. The timer resets every time the customer sends a new message during the waiting window. When the timer expires without any new messages, the Agent combines all received messages and sends one unified response.

​

Provide Information to Your AI Agent

To help your Agent answer customers accurately, you need to give it context about your business. There are two main ways to do this.

​

1. System Message (Prompt)

The System Message is the core instruction set for your Agent. In the Prompt field you can write your business introduction, define the tone and response style, and set any rules the Agent must follow during every conversation. Think of it as the standing brief you would give a new team member on their first day.

​

2. Tools

Tools extend what an Agent can do beyond answering from memory. There are three types of Tools you can attach to an Agent: When you combine a well-written System Message with the right Tools, your Agent can respond more accurately and handle more complex tasks than either approach alone.

​

Rich Responses

Beyond plain text, AI Agents in ChatbotX can send visually rich, interactive messages and execute automated system actions. Supported response types include:

• Card — a single product or service card with an image, title, description, and action button
• Carousel — a horizontally scrollable collection of cards
• Quick Replies — tappable buttons that let customers select an option without typing
• Actions — background operations such as assigning Tags, updating Custom Fields, triggering Flows, or transferring the chat to a human agent

To enable Rich Responses, toggle the Rich Response switch in the Agent configuration modal and add JSON response rules to the System Message (Prompt) field.