Say it once. Keep the record.
Describe how you are feeling, or hand over a discharge summary, a lab report or a letter. Scribe turns it into clinical records with real codes, and hands them back for you to check.
Not for diagnostic use. Scribe drafts; a clinician decides.
My blood pressure this morning was 128 over 821. I’ve had a bit of a cough2 since after Christmas, and a funny feeling in my chest3. I take metformin 500 mg tablets4.
Every record keeps what was said, the way it was said.
A code is only there if SNOMED CT, LOINC or RxNorm confirms it.
Nothing is saved until you, or a clinician, keeps it.
A model alone writes a convincing record. Grounded, it writes only what was said.
Any assistant can turn a sentence into tidy clinical data. What changes here is where each code comes from, and what happens to the parts no code fits.
Fluent, specific, and wrong in ways nobody downstream can see: codes recalled rather than looked up, a vague onset rounded into a date that was never given, and diagnoses nobody made.
Each code is looked up in SNOMED CT, LOINC or RxNorm and verified before it is kept, and the person’s own words stay beside it. Where no code fits, the words stay alone: a feeling in the chest is not angina until a clinician says so.
An honest description beats a confident wrong code, which goes on reading as authoritative for as long as it stays in the chart.
Three code systems, each asked about the part of the record it owns.
The US Edition, kept current with every release.
Each value named for exactly what was measured.
Each medication in its RxNorm form, and only if still current.
Vaccines keep the name they were given, without a code, for now.
Connect in seconds
https://scribe.maxhealth.tech/mcp
Also installable as a plugin, which brings a skill teaching your assistant how to take this down without overstating it.
Two ways in
What you describe
Conditions, allergies, medications and measurements, picked out of ordinary language, in whichever language you said it. Nothing is added that you did not say: no laterality, no dose, no severity invented to fill a gap.
A clinical document
A discharge summary, a lab report, a letter. The original stays attached to whatever is read out of it, so every value keeps its source and its date. A scan with no text in it is reported as unreadable rather than guessed at.
Questions people ask before they trust it.
Does Scribe save anything?
No. It hands records back for review, and the app you use saves only what you keep.
Can it invent a code?
No. Every code it returns is confirmed by the code system itself. Anything it cannot confirm is left out.
What happens when no code fits?
The record keeps your words, without a code. An honest description is better than a confident wrong code.
Does it change what I said?
Your words stay on every record. Vague details stay vague: “after Christmas” is not turned into a date.
Is it a medical device?
No. Scribe drafts records for a person to check. It does not diagnose, and nothing it returns is meant for diagnostic use.