Platform
Observe live calls — invisibly, server-enforced
Supervisors join any live WebRTC session without ever appearing in the agent's participant list. Enforced at the mediasoup SFU layer — not the UI — so it cannot be bypassed by inspecting the browser or sniffing packets.
<200ms
Time to drop into a live session
0 packets
Sent to the agent on monitor join
100%
Server-side enforcement (not UI flag)
Unlimited
Concurrent supervisors per room
When a supervisor joins in monitor mode, the GlobeMeet API sends a recv-only flag with the join request. The mediasoup router skips creating producer transports for that peer, never advertises them in the room's participant list, and never publishes their RTP stream to any consumer. The agent and customer have no signaling, no ICE candidates, no DTLS handshake — and therefore no evidence — that a third peer exists.
From the live monitor view, see every active session in real time across every queue you supervise. Drop into any of them with a single click and hear both sides of the conversation, watch the agent's webcam, and view their shared screen — all in HD.
Most contact center products implement monitoring as a UI flag — the agent's browser is told 'don't show this peer'. A motivated agent can open dev tools, edit the React state, and reveal the supervisor. GlobeMeet eliminates this class of bypass entirely. The supervisor's media simply does not exist in the agent's WebRTC peer connection.
Recording and monitoring policies are configurable per tenant. You can enable an audible chime when a supervisor joins, a visible 'monitored' indicator on the agent dashboard, and a per-session consent banner for customers. Different jurisdictions have different rules — GlobeMeet ships with India, EU, and US presets out of the box.
Every supervisor session generates an audit entry: who joined, when, which session, duration, and what action they took (passive monitor / whisper / barge-in / tagged for QA). The log is tamper-evident — each entry is hash-chained — and exportable as CSV or via the REST API for SIEM ingest.
No. The supervisor's peer is never added to the agent's room from the SFU's perspective. Inspecting peerConnection.getReceivers(), getStats(), or the participant list reveals nothing.
No. The supervisor consumes a separate forwarded copy of the agent's and customer's media. The customer's downlink is unchanged. Only the SFU's CPU sees a few extra percent of load per supervisor.
Yes. A supervisor can monitor any session whose queue they have permission for, regardless of which mediasoup worker or geographic region the session is routed through.
The session continues uninterrupted for agent and customer. The supervisor's client reconnects automatically and can resume monitoring within a few seconds.
See it for yourself in 15 minutes
We'll spin up a live demo room, join two test agents, and let you drop in as a supervisor. You'll see the agent dashboard from their side and confirm the supervisor never appears.