Put a document checkpoint before reusable AI context.
Start with RAG, knowledge bases and document-using agents, then apply the same boundary to research, client work and other AI workflows. Check outside sources as they change, follow the decision and retrieve the current Approved copy.
Coming soon
Your document workflow
Check. Decide. Retrieve.
Check an incoming document, organise it for the workflow and retrieve the current Approved version when it is eligible for reuse.
1Upload the outside document for checking.
2Choose a Collection for the client or project.
3Run Standard Scan and follow Approved, Needs review or Quarantine.
4Keep every status visible in its Collections without treating membership as approval.
5Retrieve the current Approved copy for your workflow while its checks remain valid.
Knowledge bases and RAG
Source documents for reusable AI context
Check incoming sources before preparation, chunking or indexing. Retrieve current Approved copies for the knowledge base and check replacements as sources change.
Check reports, papers and source packs before they enter research agents or analysis workflows. Keep current Approved copies ready as the source set grows and changes.
Check briefs and outside source materials for the AI systems you build or operate for clients. Follow each version’s decision and retrieve the current Approved copy.
The status leads. Findings, completed checks and coverage explain why and what to do next.
Supported checks clear
Approved
Required supported checks completed without a stop or review finding for this exact version under the current policy.
Available for Approved retrieval only while this is the current eligible version.
A next decision is needed
Needs review
A finding needs attention, or required coverage is missing. Read the reason and next step before using this version.
Keep it organised while you review the concern, replace the file or retry as appropriate.
In Quarantine
Quarantine
A registered hard-stop or blocking finding keeps this exact version out of normal retrieval and reuse.
Keep the record organised while you handle the finding, and keep this version out of your AI workflow.
Technical failure: The scan could not complete, so no security decision was made. Retry is available and no new scan credit is consumed.
When a corrected version is accepted, the previous version stops being downloadable immediately. Both remain in version history, but the correction needs its own scan and approval; there is no fallback while it is pending or held.