Recording & Transcribing Meetings

How Nutan Captures Audio

Nutan uses system audio capture to record your meetings. This is fundamentally different from tools that put a bot in your meeting:

• No bot joins your call — Participants never see a recording indicator from Nutan.
• System audio capture — Nutan captures what your computer hears and outputs, similar to screen recording audio.
• On-device processing — Audio is transcribed locally by an on-device AI model. No audio is ever sent to any server.

Setting Up Audio Capture

Step 1 — Grant Accessibility Permission

Nutan requires Accessibility permission to capture system audio. On first launch, you'll be prompted:

1. Open System Settings → Privacy & Security → Accessibility.
2. Toggle Nutan on.
3. Restart Nutan after granting permission.

Step 2 — Select Your Audio Device

Go to Settings in Nutan to select your preferred audio input and output devices. Nutan supports:

• Built-in MacBook speakers and microphone
• External headphones and headsets
• Bluetooth audio devices
• USB audio interfaces

For best transcription accuracy, use a headset with a microphone — this separates your voice from the prospect's voice for better speaker attribution.

Step 3 — Start a Meeting

Nutan can detect meetings automatically from your calendar. When a calendar event starts:

1. Nutan prepares a pre-call briefing with deal context.
2. Audio capture begins when the meeting starts.
3. Real-time transcription appears in the meeting panel.

You can also manually start and stop capture from the meeting controls.

Push-to-Talk

Push-to-talk is a separate feature for recording quick voice notes or dictation — it's not the same as meeting capture.

• Hold the designated key to record a voice note.
• Release to stop recording and trigger transcription.
• The transcribed text can be used for deal notes, chat input, or action items.

Push-to-talk requires Microphone permission (separate from Accessibility).

During a Meeting

While recording:

• The transcript updates in real time as words are spoken.
• Speaker detection attributes dialogue to different participants.
• The coaching sidebar shows real-time suggestions based on the conversation.
• Audio is processed in memory — it's transcribed as it arrives, not saved as a file.

After a Meeting

When capture stops:

1. Audio is discarded — Raw audio is never saved to disk. Only the transcript remains.
2. Summary is generated — Key discussion points and overall sentiment.
3. Action items are extracted — Commitments from both sides.
4. Deal signals are updated — New evidence added to the associated deal.
5. Follow-up email is drafted — Based on the conversation.

Audio Quality Tips

Troubleshooting

• No audio captured — Check Accessibility permission and restart Nutan.
• Transcription not appearing — The transcription model may still be downloading. Check Settings for model status.
• Poor accuracy — Try using a headset instead of built-in speakers. Reduce background noise.

See the Troubleshooting article for more detailed solutions.