Privacy

NRVLR interprets conversations. That means it processes recordings and transcripts of people speaking — which is personal data, and can contain sensitive personal data, because human conversation is not restricted to safe topics.

This page describes what actually happens to that data in the system as built. Where something is unresolved, it says so rather than reassuring you.

Who is responsible

Martin Schubert — m.schubert@1tm.solutions. Contact this address for any question about your data, including access, correction, export or deletion.

What we process

Account data — your email address. Sign-in is passwordless: we send a one-time link and store no password.

Conversation sources — audio or video you upload, a URL you submit, microphone input you start, or audio captured from a browser tab you select. Stored in access-controlled storage tied to your organisation.

Derived content — transcripts, speaker segmentation, and the outputs of the interpretation layers you run, including generated imagery. These are stored so a conversation can be reopened.

Billing data — handled by Stripe. Card details never reach our servers.

The people who are not you

This is the part most policies skip. A conversation usually has more than one participant, and only one of them uploads it. The others are data subjects too, and they never used this product.

You are responsible for having the right to record and process a conversation before you bring it here. In many jurisdictions that means informing every participant, and sometimes obtaining consent. NRVLR cannot verify that and does not attempt to.

If you are a participant in a conversation someone else processed here and you want it removed, write to the address above.

Automated processing and AI

Conversation content is sent to third-party AI models for transcription, interpretation and — where you run those layers — image or video generation. Content leaves our infrastructure to reach them.

Model-derived layers are marked in the interface, and generated imagery always carries an AI-generated label.

The system does not make automated decisions producing legal or similarly significant effects about individuals. It describes and summarises a conversation; it does not score, rank or gate people. Layers that inferred speakers' emotional state were removed in August 2026.

Processors we use

Supabase — database, authentication and file storage.

Vercel — application hosting, and the AI Gateway that routes model requests.

Google — Gemini models for transcription, interpretation, live sessions and generated media.

Stripe — payments.

Cloudflare — Turnstile, to stop automated abuse of the sign-in form.

Google Fonts — web fonts are currently loaded from Google's servers, which discloses your IP address to Google when you load a page. We consider this a defect and intend to self-host them.

How long we keep things

Conversations and their derived content are kept until you delete them or your account. There is currently no automatic expiry — deletion is under your control, not on a timer.

Accounts created by automated abuse of the sign-in form are deleted when found; twenty such accounts were removed in August 2026.

Your rights

Under the GDPR you can request access to your data, correction, deletion, restriction, portability, and you can object to processing. Write to the address above. You may also complain to a supervisory authority.

Status of this document

Last updated 2026-08-03. It describes the system as implemented. If you find a discrepancy between this page and what the product does, the discrepancy is a bug — please report it.