These docs are an early preview — some sections are still being expanded.
Vidext Docs

Data handling

What Vidext stores, what it deliberately does not, and what happens when people or workspaces leave.

Looking for the AI providers?

This page covers what Vidext stores first-party. For what a request sends to model providers, who those providers are, and how long they may keep it, see AI processing. Admins and members can open Settings → Data & AI for today's exact model catalog.

Where your data lives

  • Your sources (uploads, imported pages) are stored in your organization's workspace, immutable as provided. Everything derived from them — extracted text, images, summaries, concept indexes — is separate and rebuildable.
  • Your content (plans, drafts, published modules, branding) is versioned workspace state, with published versions kept as stable snapshots.
  • Learning signals (sessions, progress, quiz results) are first-party product data scoped to your organization.

All of it sits behind the organization boundary — see Access control.

Deliberate restraint in what is recorded

  • Quiz answers are structured facts — which option, correct or not, which attempt. Learner free text is not stored from quiz interactions.
  • Activity metadata is sanitized. The workspace activity history strips emails, names, tokens, and content bodies from event metadata; events reference entities by ID with minimal display labels.
  • Credentials are encrypted. Connector tokens (Google Drive, Microsoft Teams, Notion) are encrypted at rest and revocable instantly.

When a person leaves

Deleting a user does not silently rewrite your history: their past actions in the activity history are anonymized — the record survives, the identity is removed. Their sessions end; their connector credentials die with their access.

When an organization leaves

Deleting an organization cascades: its activity history and workspace data are removed with it.

The boundary with model training

Your workspace knowledge is used to generate your learning content. It is not repurposed as generic training material for models serving other customers. That restriction is contractual and binds Vidext's AI sub-processors too — see AI processing.

Formal commitments

For DPAs, retention schedules, residency, and other contractual data commitments, contact the Vidext team — this page describes product behavior, not legal terms.

Last updated on

On this page