Privacy

Privacy posture.

This page summarises how VVON™ handles policy documents, claim evidence, and personal information uploaded for forensic analysis. A formal privacy policy is being finalised with counsel; the substantive practices below already apply to every upload.

What we collect

Policy PDFs, denial letters, and damage photos you upload — kept in a private object store and accessible only to your organisation.
Structured analysis the AI produces (parsed coverages, classifications, estimates, supplements) — stored against your organisation.
Account identity from your Clerk session: email, optional name. We do not store passwords.
Operational metadata: request timestamps, rate-limit counters, billing tier from Stripe.

What we do not do

We do not sell user data to anyone, for any purpose.
We do not use your claim materials to train public foundation models. Uploads are processed via vendor APIs (Anthropic) under no-training data-handling terms.
We do not share your uploads with the carrier, opposing counsel, or any third party without your explicit instruction.
We do not retain bytes longer than necessary — Resolve-tier uploads are not retained beyond the resulting analysis record.

Vendors with access to processing

Anthropic — language and vision model inference.
Vercel — application hosting; Vercel Blob (private store) for uploaded files.
Neon — managed Postgres for structured records.
Clerk — identity and session management.
Stripe — billing and subscription management (payment details handled directly by Stripe; we never see card numbers).

Your rights

Export — request a copy of your organisation’s records at any time.
Delete — request hard deletion of your organisation’s records; we honour within 30 days.
Correct — flag any inaccurate AI output via support; corrections feed our eval set.
Object — opt out of any optional analytics (we ship with none enabled by default).