Starting today, a supervisor can join a live call between a customer and one of your agents in one of three modes: listen to both sides without being heard, whisper to the agent while the customer hears nothing, or barge into the conversation. They switch between the three without hanging up, from a browser, a SIP phone or their mobile. It is one API call to start, and a supervisor from a browser or SIP phone costs KES 0.50 a minute.
Recordings are good for QA, but they cannot change how a call ends. Coaching that changes the outcome has to happen during the call. That means hearing the call as it happens, a way to speak to the agent without the customer hearing, and the option to take over when the call needs it, all on the same line and without redialling anyone.
Three modes on one line. Listen, whisper and barge, switched through the API at any point. The supervisor's keypad cannot change modes, so nobody barges by accident.
Supervisors join from anywhere. A browser tab running @sautikit/webrtc, which answers the supervisor's leg by itself; a desk phone or softphone on a SIP account; or any phone number.
Your choice of who knows. A join tone for the agent, for the agent and the customer, or for nobody. Agents in the browser can also get an on-screen notice, such as "Mary Wanjiku is listening".
Events for your dashboards. Five call.supervision.* webhooks, from ringing to ended with the billable seconds.
Recording that matches what the customer heard. Barge audio is in the call recording; listen and whisper are not. The join tone never is.
import { SautikitClient } from "@sautikit/node";const sautikit = new SautikitClient({ apiKey: process.env.SAUTIKIT_API_KEY });await sautikit.calls.supervise(callId, { mode: "listen", supervisor: { type: "phone", number: "+254700000002" }, label: "Mary Wanjiku",});// Later, once your call.supervision.started webhook arrives:await sautikit.calls.setSupervisionMode(callId, "whisper");
Switching modes waits until the supervisor has answered; before that the API answers 409 calls.supervisor_not_connected. The developer guide walks through the browser version, the agent's on-screen notice and every event.
A supervisor is billed like any call leg, from answer to leave: KES 0.50 a minute, metered per second, from a browser or a SIP phone, or your normal outbound rate when Sautikit rings their mobile. A supervisor who does not answer costs nothing. Every rate is on /pricing.
Monitoring (listen) lets a supervisor hear both sides of a live call without being heard. Whisper lets the supervisor speak to the agent only, so they can coach without the customer hearing. Barge brings the supervisor into the conversation so the agent and the customer both hear them.
A live, answered call between a customer and one of your agents on Sautikit, whether the agent answers in a browser (including Sautikit Desk), on a SIP phone or in a mobile app. Calls answered by an AI agent, conference rooms and callers still waiting in a queue cannot be supervised.