|
|
Dialogue AI Assist Events
Processes dialogue events from Dialogue Cloud and sends agent join/leave notifications to Dialogue AI Assist. Also triggers transcriptor start messages when agents join a dialogue.
Configuration
- Dialogue AI Assist Config
- A configuration node containing your Dialogue AI Assist API credentials. This is required and must be set up before using this node.
- Enable Debugging
- When enabled, logs detailed debug information to the Dialogue Studio debug sidebar for troubleshooting purposes.
Inputs
This node accepts two types of input messages:
1. InDialogue “first” message
First means the first message you send for a given dialogue. This node uses it as the template for later output (including the required state.cancellationId).
- session object
- Session object containing:
id - The dialogue ID.sipUri - The URI of the dialogue.type - Session type (e.g., audiovideo).state.cancellationId - Required. Used to start transcriptors.originalRecipientUri - Original recipient URI.initialSessionId - Initial session identifier.
- endpoint object (optional)
- Endpoint details (uri, modality, etc.) passed through to transcriptor start.
2. Eventbus Events
- payload.event object
- Event object with
dialogueId and event-specific properties.
- payload.eventType string
- One of the supported event types:
CurrentlyHuntingEvent - Agent hunt initiated.ParticipantHuntResultEvent - Hunt answered/failed.DialogueParticipantAddedEvent - Participant joined (triggers transcriptor start for agents).DialogueParticipantRemovedEvent - Participant left.DialogueEndedEvent - Dialogue ended, cleans up context.
Outputs
Output 1: Dialogue AI Assist Status
- payload object
- Response from Dialogue AI Assist API containing:
response - The response data from Dialogue AI Assist.statusCode - The HTTP status code.headers - The HTTP response headers.
Output 2: Transcriptor Start
- session object
- Full session template from the InDialogue first message with
sipUri overwritten to the agent's SIP/tel URI:
id - Dialogue ID.sipUri - Agent's SIP or tel URI.type - Session type.state - Includes cancellation ID.originalRecipientUri, initialSessionId - Preserved from the first message.
- endpoint object (if present)
- Endpoint details from the original InDialogue first message.
Details
This node tracks agent participation in voice dialogues:
- InDialogue first message: Caches the full session template (including
cancellationId) and endpoint. Flushes any pending transcriptor starts.
- Hunt events: Tracks which agent is being hunted and whether they answered.
- Participant added: When an agent joins, sends
DialogueAgentJoinedEvent to Dialogue AI Assist and emits a transcriptor start message on output 2.
- Participant removed: Sends
DialogueAgentLeftEvent to Dialogue AI Assist.
- Dialogue ended: Cleans up all stored context for the dialogue.
The node uses flow context keyed by dialogueId to track state across multiple events.
Error Handling
The node provides status indicators:
- Green dot: Successfully processed event or sent to Dialogue AI Assist.
- Red ring: An error occurred (details in debug sidebar).
- Yellow ring: Warning (e.g., tel: URI without SIP mapping).
- Blue dot: Processing or informational status.
Example Flow
- Connect an InDialogue node to provide the first message with session/endpoint.
- Connect an Eventbus node to provide dialogue participant events.
- Output 1 goes to a debug node or further Dialogue AI Assist processing.
- Output 2 connects to your transcriptor start logic.
|
|
|
Dialogue AI Assist Transcript
This node allows you to push transcript data to Dialogue AI Assist, enabling seamless integration with Dialogue Cloud communication systems. It processes transcript data from conversations and sends it to Dialogue AI Assist for further analysis or storage.
Configuration
- Dialogue AI Assist Config
- A configuration node containing your Dialogue AI Assist API credentials. This is required and must be set up before using this node.
- Stream
- Defines the source of the transcript. Options are
visitor, agent, system, bot, or unknown. This determines the source type of the transcript data.
- Tags
- Custom key-value pairs to include in the tags sent to Dialogue AI Assist. The default
channel: voice tag is always included. You can add one or more additional tags:
- Key: Free text tag name (e.g., "department", "priority", "language"). Entries with empty keys are ignored.
- Value: Can be a string literal, or dynamically resolved at runtime from
msg properties, flow context, or global context. If a referenced property doesn't exist, the tag will be set to undefined.
Tags are resolved when each message is processed and merged with the default tag before sending to Dialogue AI Assist. Multiple tags can be added, removed, and reordered. When debugging is enabled, resolved tag values are logged before sending the request.
- Enable Debugging
- When enabled, logs detailed debug information to the Dialogue Studio debug sidebar, including the resolved tags object and transcript details for troubleshooting purposes.
Inputs
- payload
object
- The input message containing transcript data. Two payload formats are supported:
Format 1: Transcriptor Format (for visitor/agent transcripts)
transcriptor.sessionId - The unique ID of the dialogue session.transcriptor.transcript - The text of the transcript.transcriptor.transcriptId - The unique ID of the transcript message.transcriptor.participantUri.uri - The URI of the participant (e.g., SIP URI).
Format 2: Simple Format (for bot/system messages)
payload.text - The text content of the message.session.id - The unique ID of the dialogue session.
Note: For simple format, messageId is auto-generated and participantUri uses the configured stream type.
Outputs
- payload
object
-
A JSON object containing:
response - The response data from Dialogue AI Assist.statusCode - The HTTP status code of the request.headers - The HTTP headers from the response.
Details
This node processes transcript data from Dialogue Cloud conversations and sends it to Dialogue AI Assist. It supports multiple stream types and payload formats:
Stream Types:
- visitor/agent: Use transcriptor format with full participant details
- system/bot/unknown: Use simple format with just text and session.id
Payload Formats:
- Transcriptor Format: Full structured format for real-time transcriptions
- Simple Format: Streamlined format for bot/system messages requiring only text and session.id
The node automatically detects the payload format and processes accordingly, making it ideal for analyzing various types of interactions and communication patterns.
Error Handling
The node provides status indicators to help identify issues:
- Green dot: Successfully sent the transcript to Dialogue AI Assist.
- Red ring: An error occurred (details available in the debug sidebar).
- Yellow ring: Invalid or missing configuration or payload data.
Example Flow
A basic flow might include:
- An input node (e.g., HTTP or WebSocket) that receives transcript data from Dialogue Cloud.
- The Post Transcript node configured with the appropriate Dialogue AI Assist credentials and stream type.
- An output node to log or display the response from Dialogue AI Assist.
Example: Using Custom Tags
To add metadata to your transcripts, configure custom tags in the node editor:
- department =
msg.session.department (resolved from message)
- priority =
high (static string)
- language =
flow.detectedLanguage (resolved from flow context)
These tags will be included in every Dialogue AI Assist request from this node, along with the default channel: voice tag.
|