Actions

Below you'll find a list of actions that are specific to AI Agents, as well as where they can be used: either in the Flow Editor, the Function Editor, or both.

Send a Message

Available in: Flow Editor, Function Editor

Send a message to the customer.

You can send messages to the customer just about anywhere in an AI Agent flow. The Send a Message behavior is really just a special wrapper around the Send a Message action with a provision to wait for a customer response before proceeding in the flow.

The Send a Message action, in both its visual and programmatic forms, has full access to Quiq's deep rich messaging feature set including quiq replies, cards, carousels, auth and pay message support. The studio provides deep support for visually configuring rich message content. For programmatically configuring rich message content, see the Rich Messaging Gude

Visually configured Send a Message actions furthermore support Message Variables, including session fields. Use this if you have LLM-generated message content that you want to reference in an otherwise visually configured message.

Programmatic Payload Example

Sending a basic message that only contains text content:

{
  "action": "sendMessage",
  "message": {
    "text": "Hello Customer!"
  }
}

Send Catalog Message

Available in: Flow Editor

Enables you to send a message from the Message Catalog. Catalog messages are configured outside of AI Agents and can be maintained by less technical users. They are useful for static messaging content (w/ Variable Support) such as legalese or seasonal promotion content.

Start Typing Indicator

Available in: Flow Editor

Starts the typing indicator on messaging platforms that support it, such as Apple Messages for Business, Facebook etc.. Showing a typing indicator is always a good idea to provide a more natural feeling conversation, especially if your AI Agent has any latency due to chained LLM calls.

AI Agents have Special Events that can interrupt the normal flow. These are essentially events/triggers that can occur anywhere in an AI Agent's flow and have a special global handler to deal with them.

Voice Options

When the AI Agent Voice Enabled toggle button in the Voice Configuration tab is enabled, builders can define a waiting message to be sent on Voice, as well as define the delay before the message gets sent:

Voice Specific Actions

Below you'll find a list of actions that are specific to building AI Agents on Voice. In order to see these actions, you'll need to enable them via the AI Agent Voice Enabled toggle button in the Voice Configuration tab.

Set Text to Speech Voice Model

Available in: Flow Editor

The Set Text to Speech Voice Model action enables builders to define a specific voice model, as well as a fallback voice model, at any point in the flow. This can be coupled with decision steps to contextually switch voices depending on the customer's language, or any other field:

Set Speech to Text Voice Model

Available in: Flow Editor

The Set Speech to Text Voice Model action functions much in the same way as the Set Text to Speech Voice Model action does, enabling builders to define a Primary Model and a Fallback Model for Speech to Text transcription:

Builders can also define an optional language the customer will be speaking, as well as define how long the model should wait before submitting the customer utterances.

Continue the Conversation when the Call Ends

Available in: Flow Editor

This action enables builders to define whether or not they'd like to continue the conversation on a secondary messaging platform (so long as the user has one established) after the call ends:

This can be useful if the AI Agent is trying to shift from voice to messaging at a particular point in the conversation, or if you want to ensure that certain types of conversations don't end if the user accidentally hangs up.