AI Voice Assistant

This activity enables workflow designers to integrate conversational AI capabilities directly within Mitel Workflow Studio. It connects to supported AI platforms such as Google Gemini, OpenAI, ChatGPT and Anthropic, allowing administrators to configure intelligent voice or chat-based agents that interact naturally with callers.

This activity contains three tabs: AI Settings, Voice, and Chat.

Table 1. AI Voice Assistant action - AI Settings tab
Parameters Description

ConnectionType

Select the AI Integrations from the list. The options are Google Gemini, OpenAI, ChatGPT, and Anthropic.

Connection

Select the list of created connections, filtered based on the selected ConnectionType.

Model

Select the list of supported models, filtered based on the selected ConnectionType.

Instructions

Enter the clear and practical guidance for creating effective instruction sets for AI systems. These instructions help define how an AI behaves, communicates, and makes decisions within a defined context — ensuring clarity, reliability, and alignment with human expectations. This field supports up to 20,000 characters.

CompletionFormatResult

Enter the object type that will be returned by the action once it terminates.

Table 2. AI Voice Assistant action - Voice tab
Parameters Description

Language

Select the voice language from the supported language list.

Voice

Select the list of AWS Text to Speech voices.

Greeting

Enter the initial message to be played to the customer.

ErrorMessage

Enter the text to be played when the system is unable to determine what the customer said.

MOH_AudioSource

Select a WAV file that will be played when the AI is thinking. It is only recommended to populate this field when AI results take longer than 3-4 seconds.

RetryCount

Enter the number of attempts before exiting the action. The count is increased when the customer's speech is not detected. When the retry count has been exceeded, the action will terminate.

RetryExceedMessage

Enter the message to the customer when the RetryCount is exceeded.

Table 3. AI Voice Assistant action - Chat tab
Parameters Description

EnableRealTimeTranscriptionChatAPI

Enable this checkbox to allow the voice-to-text to be captured into a Cloudlink chat stream.

AssistantName

Enter the name of the AI bot that will appear in the Chat Conversation.

UserName

Enter the name of the customer. If the field is left blank, it will take the caller's name from the phone.

DeleteChatPostTranscription

If this field is selected as True, the chat conversation will be deleted once the actions terminates.

InsertSummaryOnCompletion

If this field is selected as True, a summary will be generated by the AI and added to the chat transcript.

SupervisorParticipantId

Enter the Cloudlink UserId of a user that you want to add to the Chat Conversation.

SummaryResult

Enter the text that will be prefixed to the Summary generated by the AI

Table 4. AI Voice Assistant action - Outputs
Parameters Description

Success

Indicates whether the activity was completed successfully (value = True) or not (value = False).

Result

Full result sent from the AI bot.

TranscriptionConversationId

If a conversation is created, this is set to the ConversationId of the created conversation. Otherwise, it is set to the value of the input variable ConversationId.

TranscriptionContent

JSON-formatted content of the chat conversation transcript.