NoteWright

FAQ

Does NoteWright replace my scribe?

No — it assumes you have one. NoteWright is downstream: your scribe turns the encounter into text, NoteWright turns that text into your note. If you don't use a scribe, you can paste any draft text, but the product is designed for the scribe-output problem.

Which scribes does it work with?

Any of them. NoteWright doesn't integrate with your scribe — it reads pasted text. If your scribe produces a draft you can copy, NoteWright can restyle it. No integration also means nothing to install and no vendor sitting between you and your note.

Do I paste my actual note?

Paste the text from wherever it comes — scribe output, dictation, your own draft — but remove patient identifiers first. The style engine doesn't read them: identifiers don't have style, so stripping them costs nothing in output quality. What makes a note yours is how it's written, and that's the only thing NoteWright works with.

Will it ever add something I didn't say?

No. This is the founding rule: if a value isn't in your input, it isn't in your note. Gaps become *** placeholders. Ambiguities become visible choices. Suggestions come with accept and reject buttons and are never applied silently. The system handles formatting and restyling; clinical judgment stays yours.

What happens to my notes?

Your note text is sent to the model provider's API for processing and returned — that's the entire data path. NoteWright doesn't store your notes on any server, doesn't record audio, and doesn't connect to your EHR. Your style profile and examples live on your own machine. The Premium tier goes further: the model itself runs on your hardware, so the note never leaves it at all.

Every AI tool says it's 'transparent.' Why believe this one?

Because the claim here is narrow and checkable. NoteWright doesn't claim to be right about everything — it claims to never decide for you. Which diagnosis owns which symptom, whether a relationship between findings is real — those don't have a single right answer, so NoteWright surfaces them as choices instead of answering them. You can verify that in the product in about two minutes.

Why should I believe the 'runs on your own hardware' claim for Premium?

Because it's structurally verifiable, not a policy promise: the model file is on your machine, and there's no network traffic during inference. You don't have to trust a document — you can watch the wire.

When does it launch?

Early users in 2027. The waitlist is the way in — and it's also where the build log lands, if you want to watch it come together honestly, dead ends included.