@memberjunction/ai-bridge-googlemeet

The Google Meet Realtime Bridge driver in MemberJunction's Realtime Bridges program. It connects the one realtime agent engine to a Google Meet conference: bidirectional audio, a diarized participant roster, participant mute, and a Meeting Controls facilitator channel — all behind an injectable Google Meet Media API seam so the driver builds and unit-tests with no network and no real Google Meet client.

⚠️ Early-access / allowlist caveat

Unlike Zoom's broadly-available Meeting SDK, the Google Meet Media API is early-access and allowlisted — a Google Workspace tenant must be explicitly granted access before the agent bot can pull/push real-time media. This package ships the driver and its seam; the production binding to the real Media API is a deployment concern gated on that allowlist. Out of the box, Connect throws an explicit "bind the real Google Meet Media API" error until SetSdkFactory is called.

See the Realtime Bridges Guide and /plans/realtime/realtime-bridges-architecture.md (§3 provider abstraction, §4b channels, §8 Google Meet capability row) for the full architecture. The @memberjunction/ai-bridge-zoom driver is the reference this one mirrors.

Install

npm install @memberjunction/ai-bridge-googlemeet
Copy

What it provides

• GoogleMeetBridge — @RegisterClass(BaseRealtimeBridge, 'GoogleMeetBridge'). The MJ: AI Bridge Providers row with DriverClass = 'GoogleMeetBridge' resolves to this driver via the ClassFactory. Implements the four BaseRealtimeBridge abstracts (Connect / Disconnect / SendMedia / OnMedia) and the capability-gated virtuals Meet supports (GetParticipants, OnParticipantChange), plus GetMeetingControlsEventSource for the facilitator channel.
• IGoogleMeetSdk — the injectable seam the driver depends on instead of the real Media API.
• GoogleMeetMeetingControlsEventSource — adapts the seam's roster / speaking / mute into the bridge's IBridgeMeetingControlsEventSource, so the engine wires the Meeting Controls channel.

Capability coverage (the Google Meet seed row)

Capability gating is two-layer (defense-in-depth): the engine checks the provider's SupportedFeatures first, and the driver re-asserts each flag with RequireFeature at the top of its overrides.

How Meet's missing hand-raise (➖) is handled

The Google Meet Media API exposes no participant hand-raise/lower signal. So, unlike ZoomBridge:

• IGoogleMeetSdk has no onHandRaise operation — there is nothing to subscribe to.
• GoogleMeetBridge wires no hand-raise subscription in Connect.
• GoogleMeetMeetingControlsEventSource has no IngestHandRaise — the driver feeds nothing.
• The channel contract (IBridgeMeetingControlsEventSource) still requires an OnHandRaiseChange registration, so the source implements it — but for Meet it is a registered-but-never-fired no-op. The hand-raise queue facet of the facilitator is therefore inert on Meet; roster, speaking, and mute remain fully functional.

A dedicated test (hand-raise correctly absent) pins this down: it asserts the seam exposes no onHandRaise op and that driving every signal Meet does surface (roster churn, speaking, mute) never synthesizes a hand-raise event. (For the same reason — no chat path — the Hybrid "raise hand via chat" turn-taking mode degrades to plain passive on Meet, by design.)

The Google Meet Media API seam (IGoogleMeetSdk)

The driver never imports the real Google Meet client. It depends only on this minimal interface (note: no onHandRaise, no postChatMessage vs. the Zoom seam):

export interface IGoogleMeetSdk {
    join(args: GoogleMeetJoinArgs): Promise<GoogleMeetJoinResult>;
    leave(): Promise<void>;
    sendAudioFrame(pcm: ArrayBuffer): void;                                  // agent's voice out (audio contribution)
    onAudioFrame(cb: (frame: GoogleMeetAudioFrame) => void): void;           // raw per-participant audio in (diarization)
    onParticipantJoin(cb: (p: GoogleMeetParticipant) => void): void;
    onParticipantLeave(cb: (id: string) => void): void;
    getParticipants(): Promise<GoogleMeetParticipant[]>;
    muteParticipant(participantId: string): Promise<void>;                   // where tenant/allowlist grants it
    onMeetingEnded(cb: () => void): void;
}
Copy

Production binding (deployment TODO, gated on the allowlist)

In production this is bound to the Google Meet Media API client (the allowlisted participating-client build) for per-participant PCM audio. Supply a factory via the creation seam:

import { GoogleMeetBridge } from '@memberjunction/ai-bridge-googlemeet';

// Once, where bridge drivers are configured:
//   bridge.SetSdkFactory((config) => new RealGoogleMeetSdkAdapter(config));
// The adapter implements IGoogleMeetSdk over the real Media API. The driver + its tests do not change.
Copy

Out of the box, GoogleMeetBridge ships without the real adapter — Connect throws an explicit "bind the real Google Meet Media API" error until SetSdkFactory is called. Tests inject a FakeGoogleMeetSdk.

Usage (engine-driven)

The bridge is not used directly — AIBridgeEngine.StartBridgeSession (@memberjunction/ai-bridge-server) resolves it from the provider's DriverClass, wires the transport seam to the injected IRealtimeSession, and (when a channel host is supplied) wires the Meeting Controls channel from GetMeetingControlsEventSource. See the bridge-server package and the guide's "Channel plane" section.

Testing

FakeGoogleMeetSdk (in src/__tests__/) is an in-memory IGoogleMeetSdk with drive helpers and capture sinks. The suite covers connect/disconnect, audio in→OnMedia (speaker labels) and out→seam, participant join/leave → roster + event source, the correctly-absent hand-raise path, capability gating (a feature Meet lacks throws), and mute — all with no network.

cd packages/AI/Providers/BridgeGoogleMeet && npm run test
Copy