AI Agent Reporting (Datasource)

conversation_id

String

Unique AI Agent conversation identifier — the primary key for the interaction at the provider level.

interaction_id

String

Internal unique identifier for the AI Agent record within Insights. Distinct from conversation_id where applicable.

agent_id

String

AI Agent configuration identifier used for the interaction.

agent_name

String

Human-readable name of the AI Agent as configured.

organization_id

String

Internal identifier for the customer organization or account associated with the interaction.

user_id

String

Customer-provided user identifier associated with the interaction, if supplied to the AI Agent.

call_direction

String

Whether the interaction was inbound or outbound, derived from dialed vs. caller number routing logic.

call_duration

Integer

Total call duration in seconds from start to end.

start_time

Timestamp

Call initiation timestamp in ISO 8601 format.

end_time

Timestamp

Call completion timestamp in ISO 8601 format.

caller_number

String

The originating phone number of the caller.

called_number

String

The destination phone number that was dialed.

path_taken

String

Identifier of the conversation flow or route taken through the AI Agent — for example, which dialog path or scenario was followed.

call_successful

Boolean

Whether the call completed successfully per the AI Agent. This reflects the conversation's own success flag, not a telephony-level CDR.

call_rating

String

Call quality or experience rating, if configured (e.g., positive, neutral, negative).

self_serve_status

Boolean

Whether the customer was fully self-served by the AI Agent without requiring a transfer to a human agent.

transfer_required

Boolean

Whether the interaction required escalation or transfer to a human agent.

caller_sentiment

String

Overall sentiment for the caller's side of the conversation (e.g., positive, neutral, negative).

caller_frustration_level

String

Qualitative assessment of caller frustration (e.g., low, medium, high).

abandon_reason

String

Reason the call was abandoned (e.g., user hung up, no answer). Primarily relevant for incomplete or abandoned interactions.

summary

Text

AI-generated summary describing the full interaction.

call_topic

String

Primary topic identified for the conversation (e.g., billing_inquiry, cancellation).

next_actions

Text

Recommended follow-up tasks or actions derived from the conversation.

custom_llm_response

Text

Additional AI-generated analysis or output configured in prompts, beyond the core summary.

has_audio

Boolean

Indicates that a conversation-level audio recording exists for the call.

has_user_audio

Boolean

Indicates that the caller's audio is available.

has_response_audio

Boolean

Indicates that the AI Agent's response audio is available.

recording_url

String

URL where the final audio recording is accessible after being moved to its final storage location.

speaking_breakdown

String

Text representation of speaking-time distribution between caller and agent (e.g., caller: 60%, agent: 40%).

tool_calls

JSON Array

Detailed list of tool invocations during the conversation, including tool name, parameters, results, timestamps, and success flags.

tool_results

JSON / Derived

Aggregated or extracted results from tool executions — for example, combined outcomes or key values returned by tools.

custom_data_points

JSON Object

All custom dynamic variables set during the interaction, typically customer-specific metrics or identifiers passed into the AI Agent.

crm_companyId

String

Identifier for the customer's account or company record in a CRM. Expected to be populated via integration or dynamic variables.

error_code

String

Technical error code for failed call initiation (e.g., NO_ANSWER, or provider-specific codes).

failure_reason

String

Human-readable explanation for call initiation failure.

retry_attempts

Integer

Number of attempts to initiate the call before failure was finalized.