Skip to main content

Privacy

Privacy architecture

This page describes what the system actually does. Everything below is a property of the deployment, verifiable in your own tenant from the Privacy Centre.

Principles

Process where the video already is

Analysis happens on an appliance inside the branch. Frames are read from the local camera network and are not forwarded anywhere.

Anonymise before transmission

Face regions are blurred on the device as part of the pipeline, before any payload is produced. Anonymisation is not a post-processing step in the cloud.

Keep the counts, discard the frames

Frames are held in memory for the length of the inference and then dropped. The PrivacyVision cloud has no raw video store.

Separate metadata from imagery

What leaves the branch is a numeric record: a time bucket, a zone, a count, a duration. It has no image payload.

Minimise what is collected

The system collects what a count requires and nothing else. There is no demographic estimation, no dwell-by-person and no revisit matching.

Keep data in the Kingdom

Analytics metadata is stored on Saudi-hosted infrastructure, scoped to your organisation.

What PrivacyVision does not do

  • No facial recognition and no biometric identification
  • No identity tracking between visits, zones or branches
  • No customer profiling and no personal records
  • No raw video retention in the PrivacyVision cloud
  • No sale or sharing of analytics with third parties

PDPL-oriented design

PrivacyVision is designed around the privacy-by-design expectations of the Saudi Personal Data Protection Law: minimise collection, avoid biometric identification, keep data in the Kingdom and make processing visible to the operator. Responsibility for the lawful basis, the in-store notice and the data subject process stays with you as the controller — the Privacy Centre and the bilingual signage template are there to support it.

PrivacyVision does not claim a certification or an official PDPL registration on this page. Compliance is assessed per deployment, with your own legal and privacy teams.

In-store notice

Every entrance should carry the bilingual notice. The Brand Kit includes the A3 entrance and A5 checkout templates with a data-request contact.

هذا المكان يستخدم تحليلًا ذكيًا مع إخفاء الوجوه

This location uses smart analytics with faces blurred.