No. Twofold does not use patient audio, transcripts, generated notes, or clinician edits to train AI models. That commitment is contractual rather than a stated policy, and it covers every category of content you put through the product. Separately, patient audio is never written to storage on Twofold's servers at all — it is processed in memory and discarded once the transcript exists.
This guide sets out exactly what Twofold does with each kind of data: what is never stored, what is retained and for how long, what you can delete and when, and why the difference between a policy and a contract matters when you are comparing vendors. Verified 28 July 2026.
Disclosure: this guide is published by Twofold Health and describes Twofold's own practices. Where a claim rests on how our systems are built, we say so specifically enough that you can hold us to it.
The Short Answer
Twofold does not train AI models on any of the following:
- Patient audio from recorded or uploaded sessions.
- Transcripts generated from that audio.
- The clinical notes Twofold produces.
- The edits you make to those notes — including the corrections that teach Twofold your writing style within your own account.
Your data is not sold, and it is not shared for anyone else's product development. Style learning happens inside your account and improves your notes; it does not feed a shared model.
What Happens to Patient Audio
Audio gets the strictest treatment of anything in the system, because it is the richest and least necessary thing to keep. Twofold does not write patient audio to storage at any point on its servers: recordings are held in memory, passed to Twofold's own transcription servers, and discarded once transcribed. Those transcription servers do not write audio to disk either. There is no bucket, no archive, and no database column anywhere in Twofold that holds audio.
One detail worth stating plainly, because it is the part most vendors leave out: your recording is held on your own device while the session is in progress and until it has been transcribed, then deleted from it. That local buffer is deliberate — it is what stops a dropped connection or a closed laptop costing you a session. It lives in your browser's storage on your machine, not on ours.
What Happens to Transcripts and Notes
This is where Twofold's answer differs from the audio answer, and we would rather tell you than have you discover it. Transcripts and generated notes are retained in your account until you delete them. There is no automatic expiry.
That is a deliberate choice rather than an oversight. Your notes are your clinical record, and a documentation tool that silently deleted them on a timer would be a liability, not a feature. What Twofold gives you instead is control over when they go:
Data | Stored on Twofold's servers? | How long |
|---|---|---|
Patient audio | No — never written to storage | Discarded once transcribed |
Transcript | Yes | Until you delete it |
Generated note | Yes | Until you delete it |
Your edits | Yes, within your account | Until you delete the visit |
Deleting a visit does not simply hide the row — the transcript content itself is scrubbed. You can also delete a transcript while keeping the finished note, which is the option most clinicians want once a note is signed: the record stays, the verbatim conversation does not. That action is irreversible by design.
Why This Distinction Matters When You Compare Vendors
"We don't train on your data" is close to universal as a marketing line, and almost meaningless without the specifics behind it. When you evaluate any AI scribe, the questions that separate real commitments from comfortable ones are:
We wrote the vendor‑neutral version of this as a checklist you can take to any supplier: what to ask your AI notes vendor about subprocessors, storage and training. Use it on us too.
Bottom Line
Twofold does not train AI models on your audio, transcripts, notes, or edits, and that commitment is contractual rather than a webpage promise. Patient audio is never written to storage on Twofold's servers — it is processed in memory and discarded. Transcripts and notes stay in your account until you choose to delete them, and you can remove a transcript while keeping the note. Full details are in our privacy policy, and our team will walk a compliance reviewer through the specifics on request.