Update realtime mode
Configure realtime mode: running calls speech to speech on a provider's realtime API instead of the cascaded speech recognition, AI model and voice stack. provider picks one of three: the default, a second realtime provider, and openai_live, a full-duplex live model. Send only what changes. Each provider reads its own fields: model, voice, webSearch, xSearch with xSearchAllowedHandles, and fileSearchVectorStoreIds belong to the default provider; openaiModel and openaiVoice to the second; liveModel, liveVoice and liveBackendModel to openai_live. customVoiceId overrides voice on the default provider and liveVoice on openai_live. turnMode picks who decides that the caller has finished: server_vad (the provider, on silence), semantic_vad (the model judges from the words; second provider only) or local_vad (our own tuning); it does not apply on openai_live, where the model owns every turn boundary. speakGreetingFirst opens the call with the agent's greeting. All three providers process calls outside Australia, and a dropped provider connection ends the call. The input schema lists every accepted value. Use this for 'enable realtime mode' (enabled true), 'switch the realtime provider' (provider), 'use the cedar voice' (openaiVoice), 'use the ripple voice' (liveVoice), 'use a different backend model for the live provider' (liveBackendModel), 'switch the realtime voice' (voice, or customVoiceId for a cloned voice), 'use our designed voice' (customVoiceId), 'let the model judge when the caller is done' (turnMode semantic_vad), 'let the agent search the web' (webSearch true), 'let it search X / Twitter' (xSearch true, optionally with allowed handles), 'use our turn detection instead of the provider's' (turnMode local_vad). While realtime is enabled the cascade-only Voice, Hearing and Model settings do not apply to calls; use this op's fields instead of set_voice / set_model for a realtime agent.
Authorization
bearerAuth A personal access token from Settings > API tokens.
In: header
Header Parameters
A key you choose, such as a UUID, that makes this POST safe to retry. Within 24 hours the same key and request answers with the first successful answer and runs nothing, the same key with a different request answers 422 idempotency_key_reused, and a key whose request is still running answers 409 idempotency_key_in_use. A failed request keeps nothing, so the same key runs it again.
1 <= length <= 255Request Body
application/json
TypeScript Definitions
Use the request body type in TypeScript.
Response Body
application/json
application/problem+json
application/problem+json
application/problem+json
application/problem+json
application/problem+json
application/problem+json
application/problem+json
curl -X POST "https://example.com/operations/set_realtime" \ -H "Content-Type: application/json" \ -d '{ "agent": "string" }'{}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}{
"type": "http://example.com",
"title": "string",
"status": 0,
"detail": "string",
"code": "string",
"errors": [
{
"path": "string",
"message": "string"
}
],
"estimate": {},
"retryAfterSeconds": 0
}Update public share limits POST
Configure the limits on the public share surface (/a/<slug>) - the caps behind the Share tab's 'Limits on the public surface' panel. Fields (all optional, send only what changes): maxCallSeconds (0-3600, default 180 = 3 min; 0 or null = no cap - longest a public VOICE call runs; text ignores it); maxTurns (0-500, default 30; 0 or null = no cap - for voice the pipeline turn cap, for text the number of user messages allowed in one shared chat); rateLimitStarts (1-1000, default 10 - public starts one IP may make per window); rateLimitWindowSeconds (60-86400, default 600 = 10 min - the rate-limit window). Bot detection (a challenge the share page passes before a call starts) is not configurable. Use when the operator says 'let the public demo run longer', 'cap the shared call at N minutes', 'no time limit on the share link', 'allow more turns', or 'tighten the rate limit on the share link'.
Set returning greeting POST
Pin the greeting spoken word-for-word to a returning caller the agent recognises by phone number (needs caller memory on). Distinct from set_greeting_message, which is the opener EVERY caller hears; this one fires only for a recognised repeat caller and can carry the {{caller_name}} token for the remembered name. Use when the operator says things like 'welcome returning callers back by name', 'greet repeat callers differently', or 'say Welcome back {{caller_name}} to people we know'. Pass null to clear it, so recognised callers hear the normal greeting. Applies to the current draft version.