Skip to main content
The FusionDesk AI Agent automatically drafts responses or directly resolves incoming customer tickets using RAG (Retrieval-Augmented Generation) context from your Knowledge Base.

Accessing AI Agent Settings

To configure the AI Agent:
  1. Navigate to AutomationAI Agent from the sidebar.
  2. If your organization is on the Free plan, the AI Agent interface will be locked. Upgrade to a paid plan (such as Standard or PAYG) to activate this module.

Configuration Settings

1. Enable / Disable Toggle

A master switch at the top enables or disables the AI Agent. When turned off, all configuration fields below it are disabled and ticket auto-responses are suspended.

2. Custom System Instructions

Provide instructions that define the role, scope of support, and directives for the AI model (e.g., “You are a friendly support assistant for Acme Corp. Do not answer questions outside of our product catalog.”).

3. Tone of Voice

Select the desired persona for responses:
  • Professional (default)
  • Empathetic
  • Friendly
  • Concise

4. Confidence Handoff Threshold

Set a percentage threshold (0-100%).
  • When the AI Agent’s response confidence is at or above this threshold, the reply is automatically approved and sent to the customer.
  • When response confidence falls below this threshold, the reply is saved as a private draft, allowing your human agents to review, edit, and approve it before sending.

Cost Controls & Limits

Establish strict limits to control generative model API usage:
  • Max Messages per Chat: The maximum number of auto-responses the agent will generate within a single ticket or conversation session.
  • Max Messages per Month: The total monthly API message budget for your organization. Once reached, the AI Agent suspends generation until the monthly cycle resets.