AI Audit

AIAudit

{
    "id": "<string>",
    "customer_id": "<string>",
    "aicall_id": "<string>",
    "ai_id": "<string>",
    "prompt_history_id": "<string>",
    "status": "<string>",
    "overall_score": <integer or null>,
    "evaluation": <object or null>,
    "message_ids": <array of strings or null>,
    "language": "<string>",
    "error": "<string>",
    "tm_create": "<string>",
    "tm_update": "<string>",
    "tm_delete": "<string>"
}

• id (UUID): The audit record’s unique identifier.

• customer_id (UUID): The customer who owns this audit. Obtained from the id field of GET /customers.

• aicall_id (UUID): The AI call session this audit evaluates. Obtained from the id field of GET /aicalls.

• ai_id (UUID): The AI configuration used during the call. Obtained from the id field of GET /ais.

• prompt_history_id (UUID): The prompt snapshot version evaluated by this audit. Obtained from the id field of GET /ai-prompt-histories. Set to 00000000-0000-0000-0000-000000000000 if no prompt history is associated.

• status (enum string): The audit’s current processing status. See Status.

• overall_score (integer, nullable): A numeric score summarising the evaluation, typically 1–5. Null while status is progressing.

• evaluation (object, nullable): Structured evaluation detail produced by the evaluator. The exact schema depends on the configured evaluator. Null while status is progressing.

• message_ids (array of strings, nullable): Ordered list of message IDs (newest-first) that were evaluated by Gemini. Null while progressing, on failure, or for audits completed before this feature was introduced. Present and non-empty on successful completion for calls that have messages.

• language (string): The BCP47 language code used for the evaluation (e.g., en-US, ko-KR).

• error (enum string): Machine-readable error code set when status is failed. Empty otherwise. See Error.

• tm_create (string, ISO 8601): Timestamp when this audit record was created.

• tm_update (string, ISO 8601): Timestamp of the last update to this audit.

• tm_delete (string, ISO 8601): Timestamp when this audit was deleted. Set to 9999-01-01 00:00:00.000000 if not deleted.

Note

Audit Processing

Audit evaluation is asynchronous. After triggering an audit, poll the record until status changes from progressing to completed or failed. The overall_score, evaluation, and message_ids fields will be populated only when status is completed.

Status

All possible values for the status field:

Status	Description
progressing	The audit evaluation is currently running
completed	The audit has finished successfully; scores are available
failed	The audit could not be completed; see error for the reason

Error

All possible values for the error field (non-empty only when status is failed):

Error	Description
invalid_call_metadata	The call metadata required for evaluation was missing or malformed
prompt_snapshot_not_found	The prompt snapshot referenced by prompt_history_id could not be found
prompt_snapshot_has_no_history_id	The associated AI configuration has no recorded prompt history
invalid_evaluator_response	The evaluator returned an unrecognisable response
evaluator_unavailable	The external evaluator service was unreachable
cancelled	The audit was cancelled before completion

Example

{
    "id": "d4e5f6a7-b8c9-0123-defa-b12345678901",
    "customer_id": "5e4a0680-804e-11ec-8477-2fea5968d85b",
    "aicall_id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
    "ai_id": "b2c3d4e5-f6a7-8901-bcde-f12345678901",
    "prompt_history_id": "c3d4e5f6-a7b8-9012-cdef-123456789012",
    "status": "completed",
    "overall_score": 4,
    "evaluation": {
        "tone": 5,
        "accuracy": 4,
        "resolution": 3
    },
    "message_ids": [
        "550e8400-e29b-41d4-a716-446655440001",
        "550e8400-e29b-41d4-a716-446655440002"
    ],
    "language": "en-US",
    "error": "",
    "tm_create": "2024-03-01T10:05:00.000000Z",
    "tm_update": "2024-03-01T10:05:45.000000Z",
    "tm_delete": "9999-01-01T00:00:00.000000Z"
}