start a web call
start a voice-agent call and return the credentials a browser joins it with over webrtc. media flows browser ↔ agent directly. two ways in: agentId with an api key that has the agent scope starts the call in one hop — the call id is minted inside this request, so the agent’s before-call tools receive it as {call_id} before you see it in the response. accessToken from /v1/register-call starts the call under the call_id that response already gave you, needs no api key, and is what to use when a before-call tool has to recognise the id, or when the page starts the call itself. either way the agent must be deployed and the returned callId identifies the call in conversations and usage.
Authorizations
your rumik api key, e.g. rk_live_.... create one in the rumik dashboard.
Body
- Option 1
- Option 2
exactly one of agentId or accessToken.
the agent to run — its uuid or its ua_… handle. agent_id is accepted too. needs your api key in Authorization.
"ua_fe3277d8"
the single-use wct_ token from /v1/register-call. it is the credential on its own — no api key — and the call starts under that response's call_id. access_token is accepted too.
"wct_9f8c2b1e4d5a4e6fa7b8c9d0e1f2a3b4"
Response
the call started; join with these credentials.
short-lived livekit access token for the browser to join with.
livekit websocket url to connect to.
"wss://livekit.rumik.ai"
the room created for this call.
"call-891e23fe503944c28dbf5e5a2a105190"
the call id, as shown in conversations and usage, and what your tools receive as {call_id}. for a registered call it is the call_id register-call already gave you.