Confidant runs entirely on your computer — there's no cloud component to lean on, so the hardware needs to be capable of running the AI locally. Here's what you need.
Minimum
macOS 13 (Ventura) or newer, on Apple Silicon (M1, M2, M3, M4) or recent Intel
Windows 10/11, 64-bit
8 GB of RAM for the Lite tier
~10 GB of free disk space for the app plus transcription and note-generation models
A microphone (built-in laptop mic works for testing; an external mic is much better for real sessions)
Two tiers, picked automatically
Confidant ships two note-generation models:
Lite — about 3 GB on disk. Runs well on 8 GB machines. Faster and lighter; produces solid clinical notes with a slightly tighter writing style.
Pro — about 5 GB on disk. Larger model, more nuanced clinical voice, slightly longer notes.
Confidant picks the right tier automatically based on your machine's total RAM. Machines with exactly 8 GB get Lite; machines with more than 8 GB get Pro. You can override the choice in Settings → Notes → Note Generation AI Model — for example, forcing Lite on a Pro-capable machine to keep more RAM free for other apps.
Recommended
16 GB of RAM or more — gives the Pro tier comfortable headroom; notes generate faster and have less impact on the other apps you have open
Apple Silicon (M2 or newer) — significantly faster transcription and note generation than Intel Macs, especially for long sessions
Wired microphone or headset — clean audio dramatically improves transcription accuracy
What about iPhone, iPad, or Android?
Mobile versions are on the roadmap but not shipping yet. The underlying constraint is the same as the desktop version — running a clinical-grade language model locally takes a lot of memory, which is harder to come by on phones. We'll release mobile when we can ship a clinical experience that doesn't compromise on note quality.