Crux takes names, contact details and health details out of your recordings before it keeps anything, and its AI providers keep nothing at all.
Tick what a participant says. Crux keeps the point, not the person.
Tick what they say. Change anything.
7 details swapped out
Speaker B · 04:12
Hi, I’m [PERSON_NAME]. You can email me at [EMAIL_ADDRESS]. I came in because of my [MEDICAL_CONDITION], so the frames had to fit new lenses. My insurance number is [HEALTHCARE_NUMBER].
[PERSON_NAME] said the old frames pinched. Write to [EMAIL_ADDRESS] about her [MEDICAL_CONDITION] check.
A made-up example that runs in your browser; nothing you type is sent. In Crux, the transcription service finds these details in the recording itself.
These happen on their own, on every plan, for every project.
Before a transcript reaches Crux, names, contact details, addresses, dates of birth, health conditions, medicines, and money and ID numbers are swapped for what kind of detail they were. Summaries, findings and quotes are made from that text. A transcript that comes back with them still in is refused.
Once Crux has the transcript, it asks the transcription service to delete its copy of the transcript and of the audio it was sent.
Every request that carries your content goes only to AI providers that do not store or train on what they receive. If none is available, the request stops instead. Crux does not train models on your content.
Comments, approvals and research are shown only to signed-in people invited to the project, checked on the server for every request. Viewers look, reviewers comment, approvers approve.
Uploads live in private storage, never on public links, and open only through Crux for people with access.
Deleting an item removes it and its stored file. Ask us and we delete your account and everything stored with it. Crux never stores your password.
Crux runs on a few well-known services, each for one job.
Evidence behind every decision, in every tool you build with.