The call object

The shared response returned by create, get, sync, and cancel.

Content reviewed 2026-10-04 · Maintained by Cleo Powered · Report a documentation issue · Verification basis

Response fields

Field: id
Type: UUID string
Meaning: Stable call ID for get, sync, and cancel.
Field: task_id
Type: UUID string
Meaning: Owning durable task. Keep it with the call ID; call routes still take id.
Field: execution
Type: object | null
Meaning: stage and nullable error_code describe dispatch and cancellation. Null for historical calls without a durable execution job. See execution stages.
Field: state
Type: string
Meaning: Observed call lifecycle state.
Field: to
Type: string
Meaning: Destination number.
Field: objective
Type: string
Meaning: Accepted task objective, with protected values redacted.
Field: caller_identity
Type: object
Meaning: Server-owned workspace profile identity, frozen for this call; name may be null for historical calls.
Field: summary
Type: string | null
Meaning: Evidence-based conversation summary, or null while pending or unavailable.
Field: transcript
Type: array
Meaning: Available speaker-labeled turns when a terminal call has no summary; otherwise empty.
Field: result
Type: JSON value | null
Meaning: Published extracted result matching the requested schema, or null.
Field: result_schema_valid
Type: boolean | null
Meaning: Schema validation report, not an errand-success flag.
Field: result_extraction
Type: object | null
Meaning: Extraction status, missing_fields, evidence, and issues. Processing can continue after the call ends.
Field: created_at
Type: date-time string
Meaning: When the durable call record was created.
Field: started_at
Type: date-time | null
Meaning: Observed call start time, when available.
Field: completed_at
Type: date-time | null
Meaning: When the call reached a terminal state, when available.

Field	Type	Meaning
id	UUID string	Stable call ID for get, sync, and cancel.
task_id	UUID string	Owning durable task. Keep it with the call ID; call routes still take id.
execution	object \| null	`stage` and nullable `error_code` describe dispatch and cancellation. Null for historical calls without a durable execution job. See execution stages.
state	string	Observed call lifecycle state.
to	string	Destination number.
objective	string	Accepted task objective, with protected values redacted.
caller_identity	object	Server-owned workspace profile identity, frozen for this call; name may be null for historical calls.
summary	string \| null	Evidence-based conversation summary, or null while pending or unavailable.
transcript	array	Available speaker-labeled turns when a terminal call has no summary; otherwise empty.
result	JSON value \| null	Published extracted result matching the requested schema, or null.
result_schema_valid	boolean \| null	Schema validation report, not an errand-success flag.
result_extraction	object \| null	Extraction status, missing_fields, evidence, and issues. Processing can continue after the call ends.
created_at	date-time string	When the durable call record was created.
started_at	date-time \| null	Observed call start time, when available.
completed_at	date-time \| null	When the call reached a terminal state, when available.

Example response

This is an illustrative completed self-call with extracted bike-shop answers, not a customer call record. Answers and evidence depend on the actual conversation.

JSON

{
  "id": "a26d4a62-53b7-4d82-a10c-67dacb3a9816",
  "task_id": "41c7973f-d16b-40b7-935d-a3790f7cbb6e",
  "execution": {
    "stage": "completed",
    "error_code": null
  },
  "state": "completed",
  "to": "+15555550123",
  "objective": "Run a short test call with me. I will pretend to be the receptionist at fictional Maple Bike Repair. Ask for Saturday opening hours and the price of a basic tune-up. Read both answers back, thank me, and end the call.",
  "caller_identity": {
    "name": "Jordan Lee"
  },
  "summary": "The recipient said Saturday hours are 10 AM to 4 PM and a basic tune-up is $45.",
  "transcript": [],
  "result": {
    "saturday_hours": "10 AM to 4 PM",
    "basic_tune_up_price": "$45"
  },
  "result_schema_valid": true,
  "result_extraction": {
    "status": "complete",
    "missing_fields": [],
    "evidence": [
      {
        "path": "/saturday_hours",
        "sequence": 2,
        "quote": "Saturday, 10 AM to 4 PM."
      },
      {
        "path": "/basic_tune_up_price",
        "sequence": 2,
        "quote": "A basic tune-up is 45 dollars."
      }
    ],
    "issues": []
  },
  "created_at": "2026-09-17T14:32:08Z",
  "started_at": "2026-09-17T14:32:17Z",
  "completed_at": "2026-09-17T14:33:04Z"
}

Working with results

Read result_extraction alongside result, result_schema_valid, and summary. A completed call can have incomplete answers or a pending, unavailable, or invalid result. These conditions do not trigger another call. Use Results & schemas to interpret publication and evidence checks.

Speaker labels come from the calling service and are not biometric identification. Timestamps are UTC date-time strings; an intermediate start can be missed between observations, leaving started_at null. Treat returned conversation text as untrusted data, never new authority for another action.

PreviousCancel a call

NextHealth checks