3
AI Agents work quietly in the background to automate and standardize actions in your help desk.
They listen for specific ticket events, evaluate contextual conditions, and execute the right actions all while keeping every update visible in your activity logs.
To start using AI Agents, head over to the Autopilot > Home in your sidebar and explore the available agents.
This article explains the core concepts of how AI agents operate - including triggers, actions, and the way they appear in your account.
Overview
Each AI Agent is built around a simple model:
Trigger → Evaluate Conditions → Perform Actions
For example:
When a ticket is created (trigger), the Duplicate Ticket Notifier checks if similar tickets exist (evaluation) and flags them by writing a private note(action).
AI Agents continuously monitor your help desk for these triggers to triage tickets and act automatically in the categories where they have been deployed.
Triggers
A trigger is the event that causes an AI Agent to run.
AI Agents respond to specific ticket-level events, such as:
- Ticket Created - runs when a new ticket is created.
- Contact Update - runs when a new reply is received from the contact.
- Agent Update - runs when a new reply or private note is posted.
- Ticket Closure - runs when a ticket is moved to a ‘closed’ status.
Each AI Agent is specifically designed with one or more pre-determined relevant triggers.
For instance:
- The Duplicate Ticket Notifier runs when a new ticket is created.
- The Auto Ticket Closer runs when a contact message is received.
- The Unanswered Query Notifier runs when an agent replies to a ticket.
Actions
When an AI agent is triggered inside; it evaluates the context in relation to its purpose, reasons on a case-by-case basis and decides to perform one or more actions automatically.
Typical actions that an AI agent can do includes:
- Adding private notes to a ticket
- Changing status or priority
- Adding tags to tickets
- Updating Ticket Custom Fields
- Updating Contact Custom Fields
Each AI agent’s actions are predefined and optimized expressly for its purpose.
Admins can customize how and where those actions apply for instance, choosing categories, notification recipients, or if they want their AI agent to run in Supervised/Autopilot mode.
Action Dependencies
For some AI agents to perform actions, they may need certain conditions and directions while ‘hiring’ and configuring to make sure they run just as desired in your helpdesk.
Common conditions include:
- Ticket category (e.g., “Support,” “Returns”)
- Defining Ticket Tag
- Time Ranges
- Ticket and Contact Custom field values (e.g., product type, region)
These action dependencies help you tailor and customize an AI agent’s behavior to serve your needs and customer experience processes.
For example, you can configure the AI agent - Duplicate Ticket Notifier to run only on “Customer Support” category tickets while ignoring “Billing” ones.
Visibility in Ticket Activity
All actions performed by AI Agents appear in the ticket’s activity timeline just like your other agents.
These updates are clearly marked with the AI agent’s name, so users can differentiate them from updates made by human agents in the timeline.
Example:
Private note added by Translation Agent
Ticket tagged as “duplicate_ticket” by Duplicate Ticket Notifier
This renders full transparency into all activities done on any ticket by AI agents.
Agent Execution
- AI Agents run independently of each other.
- If multiple agents are configured for the same category, each will evaluate conditions and act parallelly.
- Execution order is not guaranteed; however, every action is logged in the ticket timeline in the order it was completed.
Example: Duplicate Ticket Notifier Agent in Action
Let’s take a look at the Duplicate Ticket Notifier Agent as an example:
- A customer sends a message about a missing order.
- The Duplicate Ticket Notifier reads the incoming message and scans for any recent tickets by the same customer about the same issue.
- It tags the potential duplicate tickets of that customer.
- It then automatically writes a Private Note message with the identified duplicate tickets to help your support agents easily merge any existing duplicates.
- Every ticket update done by the AI agent is logged in the ticket activity
