Once you've finished onboarding and added a patient, recording is two clicks.
Before you start
Pick a quiet space and a good mic. Confidant transcribes whatever it hears, so background noise gets transcribed too.
macOS will prompt you for microphone permission the first time you record. Approve it. If you accidentally denied it, fix this in System Settings → Privacy & Security → Microphone and toggle Confidant on.
If you're on a laptop, plug in. Local transcription is CPU/GPU-heavy and will drain a battery quickly during a long session.
Start the session
Click the patient's name in the sidebar to open their detail page
Click the red Start New Session button near the top of the page
The recording panel slides in from the right
Pick In-Person or Telehealth
Click the red Record button
You're live. The status bar at the top of the panel turns red and shows Recording, and a red indicator appears at the top of every screen so you don't lose track.
During the session
You'll see a live audio level meter so you know the mic is picking you up. If the level looks dead, check that the right input device is selected — head to Settings → Recording → Audio Device to switch.
A few things to know:
Pause stops capture without ending the session. Useful if you need to step out, take a private call, or talk about something off the record. Resume picks up exactly where you left off.
Stop ends the session and kicks off transcription. There's no "discard" — once stopped, the audio is saved.
The session keeps recording even if you close the panel, switch patients, or use other parts of Confidant. The red indicator at the top reminds you a session is live.
After you stop
Confidant runs through three phases automatically:
Transcribing — the transcription engine turns the audio into text. This usually takes about a quarter of the session length (a 50-minute session takes ~12 minutes to transcribe). The status bar shows you which phase you're on.
Transcribed — the raw transcript is saved. You can read it inside the session panel.
Generate notes — click the Generate notes button when you're ready. The AI reads the transcript and writes a draft note for you to review.
What if something goes wrong?
No audio in the transcript → the mic likely wasn't picking you up. Check the recording panel — if you saw the level meter moving, the audio is there; otherwise the recording captured silence and there's nothing to transcribe.
Transcription is taking forever → it's CPU-bound. Close other heavy apps (especially other transcription/video apps). If it still feels stuck, see Troubleshooting.
Notes look wrong → you can regenerate them as many times as you want, and editing the draft directly is encouraged. The AI gets you 80–90% of the way there; you finish.