RT Assist

Ready for your conversation?

Go live in one click for a quick test, or customize the meeting when you need more control.

1
PrepareSet the goal and confirm what RT Assist may use.
2
Listen & coachSee clean notes and one useful prompt at a time.
3
Review recapConfirm the outcome, commitments, and follow-up.

IdleNot listeningNot startedRecovery status: connected

Preparation is ready.

Selecting the button confirms consent to transcription. No saved background exists for this meeting; your preparation fields still apply. It uses one microphone and keeps live web on.

Before the meeting

Quick start

One optional goal, then three required safety checks.

Fastest path

Start in under a minute

Goal is optional

Configured start

Confirm two choices below

Confirm everyone’s consent and choose the saved background this meeting may use—or explicitly choose none. RT Assist then verifies your workspace before opening the selected audio source. Live web stays on.

The microphone stays off until consent, background, and workspace checks pass.

What do you want to accomplish?Optional. Leave it blank to coach from the conversation alone.

Live web readyReady by default for current public facts. Turn it off for this meeting whenever you prefer.

Search starts only during a live conversation and every result shows its sources. Private-looking requests are blocked when detected; screening is best effort.

Try a guided demo insteadFabricated conversation · no microphone · no API cost

Safe guided demo

See the whole flow without a microphone

No audio · no API cost

Choose a fabricated conversation

A short client conversation that ends with a clear owner and time.

Optional: personalize your preparationAdd attendees, tone, background notes, guardrails, or a reusable playbook.

Your nameWhat is your role?What kind of conversation is it?Who is taking part?Who are you meeting?How should RT Assist help you sound?What should RT Assist know for this meeting?What should RT Assist avoid?Topics to handle with care

Reusable playbookSave this preparation once and reuse it for the next conversation of the same kind.

Playbook name

Optional: change how audio is capturedThe default mixed microphone works for the fastest start.

How will the conversation reach RT Assist?

Choose explicitly. No permission prompt opens until consent is confirmed and you use the button below.

Mixed microphoneSupported pilot. Transcribes only sound reaching your microphone; no speaker identity is inferred.Microphone + browser-tab audioSupported desktop Chrome pilot. Choose the microphone and a browser tab with Share tab audio; each source gets its own neutral transcript label and stop control.

One mixed microphone channel

All voices reaching this microphone stay in one unseparated channel. Participant labels are unavailable because RT Assist cannot verify who spoke.

Language spoken in this conversationChoose the main language for the strongest noise and wrong-language filtering. Choose multilingual only when people will intentionally switch languages.Everyone knows this conversation is being transcribed and has agreed.

Background for this meeting

Your preparation fields above still apply. Of your saved sources and approved memories, RT Assist may only use what you confirm here.

Checking your saved sources…

Exactly 3 checks are required to start listening

The goal and every personalization field are optional. Your microphone stays off until these three safety checks pass; the guided demo does not need them.

Confirm everyone’s consentUse the consent checkbox above before enabling any microphone or tab audio.
Confirm meeting backgroundUse the "Background for this meeting" box above. You may confirm an empty saved list; your preparation fields still apply.
Live workspace will be checkedNothing to fill in: RT Assist checks your signed-in workspace after you press start.

Optional: transcription and voice

RT Assist chooses sensible defaults. Change these only if needed.

How to transcribeBrowser-managed captions may process audio on your device or through your browser or operating-system provider; that provider's terms and retention apply. Audio transcription sends bounded chunks to the configured OpenAI account. RT Assist does not save raw audio in either mode.VoiceUsed only when you pause listening and choose “Read aloud.”

Browser captions API: not detectedMicrophone recording API: not detectedTab-sharing API: not detected

API detected does not mean permission granted or audio active. Permissions are checked fresh after consent.

Supported-device matrix

Chrome 126+ · macOS 14+ / Windows 11Mixed microphone · supportedOne consent-bound microphone stream feeds audio transcription. Manual hardware verification remains pending. Evidence: deterministic browser-API mocks.
Edge 126+ · Windows 11Mixed microphone · supportedUses the same MediaRecorder contract as desktop Chrome. Manual hardware verification remains pending. Evidence: deterministic browser-API mocks.
Chrome 126+ · macOS 14+ / Windows 11Microphone + tab audio · supportedThe exact microphone and shared-tab streams feed separate transcription recorders with independent stop controls. Manual hardware verification remains pending. Evidence: deterministic browser-API mocks.
Edge 126+ · Windows 11Microphone + tab audio · capability onlyThe same independent recorder path is covered with browser-API mocks, but Edge remains outside the manually qualified primary pilot. Evidence: deterministic browser-API mocks.
Firefox 127+ · macOS / Windows / LinuxMixed microphone · capability onlyMediaRecorder support is expected, but this pilot is not manually qualified on Firefox. Evidence: manual device check pending.
Firefox 127+ · macOS / Windows / LinuxMicrophone + tab audio · unsupportedDual-source meeting-tab audio is outside the supported pilot topology. Evidence: manual device check pending.
Safari 17+ · macOS 14+Mixed microphone · capability onlyMicrophone recording may work, but the personal pilot is not manually qualified on Safari. Evidence: manual device check pending.
Safari 17+ · macOS 14+Microphone + tab audio · unsupportedShared-tab audio is not a supported RT Assist capture path. Evidence: manual device check pending.
Modern mobile browser · Android / iOSMixed microphone · capability onlyVisible speakerphone or second-device use may capture room audio; labels remain neutral because there is no verified diarization. Evidence: manual device check pending.
Chrome / Edge · macOS / WindowsNative meeting-app audio · unsupportedRT Assist does not claim capture of native Zoom, Teams, or other desktop-app audio. Evidence: manual device check pending.
Any · Android / iOSPhone-call audio · unsupportedDirect phone-call audio, locked-phone capture, and background listening are not supported. Evidence: manual device check pending.

Usage and spending

Totals come from the private spend ledger. No account keys or conversation text appear here.

Open full diagnosticsProvider, model, latency, retries, and cue feedback for this workspace.

During the meeting

Conversation notes

What RT Assist hears will appear here

Nothing captured yet.

Start listening and your conversation notes will appear here.

Live coach

Helpful suggestions

Read the large prompt first. Evidence and reasoning stay secondary.

Say next

Your opening will appear here as soon as live listening starts.

Confidence: pending

Reasoning will show here after the first pass.

Debrief

Session summary

Generated when you stop the session

No summary yet.

Stop the session after the meeting to generate the recap.