Get Voicemail Settings
What the AI voice agent does when an outbound call reaches an answering machine: ``hang_up`` (the default) or ``leave_message`` (a short, pre-approved message naming the company, saying it is an AI assistant, with ``callback_number`` and how to opt out). Either way the call ends ``voicemail``.
API key access
workspace:read scope with an API key (ro_pat_...). A signed-in session uses the caller's role./api/v1/workspaces/current/voicemailAccess token from /api/v1/auth/login, or an API key (ro_pat_..., RO-176) from /api/v1/workspaces/current/api-keys: the key implies its workspace and needs the route's scope. MCP connectors (/api/v1/mcp only) may use an OAuth access token (ro_oat_..., RO-284).
In: header
Header Parameters
Id of the workspace to act on (must be a member).
Response Body
curl -X GET "http://localhost:8000/api/v1/workspaces/current/voicemail" \ -H "X-Workspace-Id: string"{
"mode": "hang_up",
"callback_number": "string",
"configured": true,
"updated_at": "2019-08-24T14:15:22Z",
"updated_by": "deea00dc-b6b6-4412-a483-26ac61e1f6fe"
}{
"detail": [
{
"loc": [
"string"
],
"msg": "string",
"type": "string",
"input": null,
"ctx": {}
}
]
}Get Voice Capacity
Concurrent AI calls (RO-181): the workspace's limit in effect (the platform's per-workspace limit, lowered by the workspace's own setting), its calls ringing or in progress, its dials waiting for a free slot, and whether the platform-wide limits (all workspaces, voice workers, SIP channels, speech-to-text / LLM / text-to-speech capacity) leave room now. Every dial is admitted against these limits right before it is placed; a dial without a slot waits and is tried again (it never fails for lack of capacity until it has waited ``VOICE_ADMISSION_MAX_WAIT_SECONDS``).
Get Writing Style
What the email writer learned from this workspace's own decisions on AI drafts (RO-282): approvals, edits (with their diffs), rejections with reasons and Ask Zero's thumbs. The rules, the short versioned block the writer prompt gets, and the decisions counted. No model training; nothing is shared across workspaces.