The current operating path
How Cloneshield works when a persona needs a paper trail.
Five bounded steps connect a persona reference record to a reviewed candidate, a deliberate request, and a clear deletion state. The details below describe what the product does today, including the limits that matter.
Intake → review → action
Scheduled checks surface candidates. People confirm the finding before a platform request is sent.
The five-step workflow
What happens after you start
Each handoff has a defined owner, state, or record in the current product.
- 01
Intake
Start with the persona you want to protect.
Add the persona’s references and context: URLs, content hashes, identifiers, and prose describing the voice, visual style, backstory, and monitoring scope. The intake stores references, not uploaded audio or image files.
- 02
Matching
Scheduled checks use the sources you configured.
On its scheduled run, CloneShield checks enabled Source URLs for personas whose monitoring status is ACTIVE or QUEUED. A candidate is queued when the current matching threshold is met; the source list is configured, not the entire web.
- 03
Review queue
A person reviews what deserves action.
Pending candidates appear in the persona’s review queue. The owner can review them, and an active franchise member with the Reviewer role can review candidates for personas shared into that franchise.
- 04
Takedown
A confirmed finding can become a documented request.
A confirmed finding can be previewed before the owner or authorized reviewer sends it. CloneShield routes the notice to the matched platform’s configured abuse contact and records a takedown_requested activity entry. The platform decides whether to respond or remove anything.
- 05
Deletion
Remove a saved persona from normal product views.
The owner can delete a saved persona. Today that operation is a soft delete: the row stays in the database, while normal reads and listings hide it and direct persona reads treat it as not found.
Scope, stated plainly
What CloneShield does not do
These boundaries are part of the product description, not footnotes. They keep the workflow useful without claiming more than the current implementation can show.
- The intake does not upload reference files; it accepts URLs, content hashes, identifiers, and prose.
- No biometric or voice-model training, and no facial identity matching.
- There is no promise of entire-web coverage or instant, continuous scanning; monitoring runs against enabled configured sources on a schedule.
- A match is not a legal conclusion that infringement occurred, and a takedown request is not a guarantee that a platform will remove content.
- Persona deletion is currently a soft delete, not a hard-erasure guarantee for every related record.
Ready for the first step?