In this document, we will talk about the Start Trigger and how any developer can configure it to initiate an AI agent based on specific events and leverage intent-based filtering for greater control. The Start Trigger is automatically generated when the Agent Builder platform loads a new agent. It acts as the entry point of any agent flow and defines the conditions under which the AI agent will be initiated.

​
Trigger Event

The Trigger Event setting specifies what kind of event will activate the AI agent.

SigmaMindAI supports a variety of channels and conversation types, and selecting "Conversation Started" as the trigger event makes it easy to build conventional AI agents for these use cases.

Common Use Cases for “Conversation Started” Trigger:

  • A customer initiating a query via the Helpdesk Chat
  • A support request received through Helpdesk Email
  • A support conversation started via Slack or WhatsApp
  • An inbound phone call from a customer
  • An outbound call initiated by the system (e.g., scheduled callback)

By simply selecting "Conversation Started" as the start trigger, developers can build AI agents across multiple customer interaction channels using a unified approach.

Other Supported Trigger Events:

  • "Conversation Updated": Triggered when a message is added or modified in an ongoing conversation
  • "Conversation Ended": Useful for logging, follow-up actions, or handoff summaries

​
Filter by Intent (Inbound AI Agents)

The Filter by Intent option enables developers to scope down agent execution to specific types of user intents, creating highly focused inbound AI agents.

​
🔧 Functionality

When enabled, the AI agent will only activate if the customer’s initial message (or the relevant message based on the trigger event) matches one of the pre-configured intents in AI Studio.

The SigmaMind AI system comes with pre-configured categories and over 200 intents, which have been trained on millions of real-life customer queries to detect these intents with high accuracy.

Only when SigmaMind AI is more than 95% confident about a particular intent (meaning the confidence interval is above 95%), the system will trigger an AI agent based on that intent. If the confidence level is below 95%, SigmaMind AI will not trigger the agent, thereby providing strong guardrails against low-confidence classifications and reducing hallucinations. This ensures that customers receive highly accurate responses and a reliable conversational experience.

​
💡 Benefits

  • Triage by Intent: Route conversations to different agents based on what the customer is asking.
  • Workflow Automation: Build specialized agent behaviors such as:
    • Offering discounts before cancellation
    • Confirming shipping status before processing returns
    • Answering FAQs vs. handling transactional support differently

📝 Note: Intents are useful primarily when dealing with inbound text or email queries, where it is necessary to detect the customer’s intent before triggering the AI agent.

In the case of voice calls, typically the AI agent speaks first before the customer responds. Thus, configuring the intent in the start trigger is generally not useful for inbound or outbound voice flows. Instead, you can use the Customer Intent node later in the flow to triage into different paths once the customer’s response is captured.