Versioned scans
Every scan-driven page keeps a version history. Activate, view, or delete past versions. Counter for non-deterministic AI output.
Executive summary
Three short paragraphs explaining the feature and value.
Versioned scans is the version history layer running underneath every scan-driven page in the workspace. Every time a scan runs (visibility scan, audit scan, blog generation, brand profile regeneration), a new version is created. The active version drives the current view. The full history of past versions remains accessible through the Versions tab on each relevant surface throughout the entire workspace.
AI output is probabilistic. The same scan run twice can produce slightly different language, slightly different scores, slightly different recommendations. Versioning lets a user keep the language they liked from yesterday's scan and reject the one they got today. Without versioning, every fresh scan overwrites the previous output and good language gets lost to model drift quickly.
The Versions tab on each surface lists past scans with activate, view, and delete affordances. Click activate on any prior version to make it the current view. Click view to inspect without changing the active version. Click delete to remove versions no longer needed. The versioning model preserves user agency over probabilistic output rather than forcing acceptance of whatever the latest model run produced.
Key highlights
Five capability points teams should know about quickly.
- Version history layer for every scan driven page
- Activate, view, or delete past versions per surface
- Counter for probabilistic AI model output drift
- Keep language you liked, reject what you do not
- Versions tab on every relevant workspace surface
Top FAQs
Five common questions answered for fast practical clarity.
Which surfaces support versioned scans?
Every surface where AI generates output: Brand profile, AEO visibility, GEO visibility, SEO visibility, AEO tech audit, GEO tech audit, SEO tech audit, every individual blog draft, the Dashboard score panels, and the Visibility report. Anywhere an LLM or scan produced the visible content, versioning runs underneath and the Versions tab is available for review.
How many versions are kept?
The last fifty versions per surface, with the active version always preserved separately. Older versions beyond fifty are pruned automatically to keep storage reasonable. Users can manually pin specific versions to prevent pruning (useful for capturing a particularly good draft or score moment before later scans drift away from it across model updates running).
Can I export specific versions?
Yes. From the Versions tab, click any past version and use the export affordance to download a snapshot. PDFs export with the version timestamp embedded in the metadata so the audit trail is preserved. Useful for compliance scenarios where the brand needs to demonstrate what specific output the AI system produced at a specific date for review later.
What happens when I activate a past version?
The selected version becomes the current active view across the workspace. All downstream surfaces that reference the changed surface refresh to reflect the activated version. The Dashboard score recomputes if a visibility scan version changed. The audit summary updates if an audit version changed. Activation is reversible: switch back to a different version anytime without data loss.
How does versioning interact with the XEO calendar?
Blog versions in particular tie into the calendar. Each blog has multiple draft versions before publishing. The active version is what publishes. After publishing, future edits create new versions while preserving the published one. The calendar shows the active draft per blog with an indicator when newer drafts exist that have not yet been promoted to active by the user.