Skip to main content

Installing Confidant

J
Written by Jason

Confidant is a single download for macOS and Windows.

macOS

  1. Download the .dmg from your account page

  2. Open the .dmg and drag Confidant into your Applications folder

  3. Launch Confidant from Applications or Spotlight (Cmd-Space, type "Confidant")

  4. The first time you launch, macOS may briefly show a "verifying" dialog — this is Gatekeeper checking the signature. After the first launch, this doesn't happen again.

If macOS says the app is "from an unidentified developer," you have an outdated copy — please re-download from your account. Our Mac builds are signed and notarized; that warning shouldn't appear.

Windows

  1. Download the Confidant-setup.exe installer from your account page

  2. Double-click the installer and follow the prompts

  3. Launch Confidant from the Start menu

If Windows SmartScreen flags the installer, click More info → Run anyway. SmartScreen warnings are common on new releases until the installer builds enough reputation.

After install

The onboarding wizard opens the first time you launch. It walks you through:

  1. Accepting the license agreement

  2. Your name, practice, and therapy approach

  3. On-device setup — a few quick choices, then one download: the transcription model (about 1.6 GB recommended, or 466 MB), the note-generation model (Standard, about 3.5 GB, or Pro, about 5 GB), and optional speaker separation (about 370 MB on Mac, 30 MB on Windows)

  4. Microphone permission

  5. Starting the 14-day trial, or activating your license if you've already purchased

Total onboarding time: about 10–15 minutes on a normal connection, mostly waiting for the note-generation model to download. After this, the app works fully offline. See Completing onboarding for every screen.

Where Confidant lives on disk

If you ever need to find Confidant's data:

  • macOS: ~/Library/Application Support/com.confidant.notes/

  • Windows: %APPDATA%\com.confidant.notes\

That folder contains your encrypted database, audio recordings, and the downloaded model files.

Did this answer your question?