Skip to main content
An Action represents a task your Agent can perform during an interaction.
Actions handle information gathering, decision-making, or triggering external processes.

​
Available Actions

Actions are triggered when certain conditions are met.
For End Conversation and Transfer Call, there are two primary trigger types:
  1. Phrase-based
  2. Function-based

​
Phrase-Based Triggers

Phrase-based triggers are reliable but less flexible.
The Action is executed when the Agent says a specific phrase. Example — transferring a call to a human: Phrase-based Example You can prepare your prompt so that if the customer asks for support, the Agent uses a predefined phrase. For example:
“If the customer wants to talk to or be transferred to the support team, say the phrase ‘Let me transfer you to the support team.’”
Once the phrase is detected, the Action is triggered.

​
Function-Based Triggers

Function-based triggers give the Agent flexibility.
Here, the Agent decides when conditions are met and when to perform the Action.
  • Pros: More dynamic, less restrictive.
  • Cons: Less predictable.
Experiment with both methods to find the best fit for your workflow.

​
Action Types

  • End Conversation — Ends the call when triggered.
  • DTMF — Sends touch-tone digits (no specific trigger required).
  • Transfer Call — Transfers the call to another number, triggered by phrase or function.
  • External — Connects to external systems (see below).

​
External Actions

External Actions allow your Agent to call an external endpoint during a conversation to create or retrieve data.

​
Configuration Fields

  • Action Name — Unique name for the Action.
  • Description — Brief explanation of its purpose.
  • URL — Endpoint where data is sent.
  • Speak on Send — Message the Agent says before sending data.
    • Example: “Let me retrieve your contact information.”
  • Speak on Receive — Message the Agent says after receiving data.
    • Example: “Okay, I got your contact information.”
  • Processing Model — Currently only Muted is supported.
  • Signature Secret — Not yet supported.
External Action Example

​
Input Schema

The Input Schema defines the structure of data the Agent collects before executing an External Action.

​
Options in the Interface

  • Add new properties (e.g., user_id, issue, date).
  • Mark properties as required or optional.
  • Define types (string, number, boolean, etc.).
  • Add descriptions to guide Agent behavior.
Once all required fields are collected, the External Action is triggered automatically.

​
Next Steps

⌘I