Serova context
Why Ficta sits in Labs and how it fits the wider Serova portfolio.
Serova Labs
self-hosted GatewayModel traffic, redacted before it leaves.
Ficta is a Serova Labs project for teams evaluating AI chat with a local redaction boundary. This page keeps the portfolio context short; the full product site, docs, install path, and FAQs live at ficta.sh.
Serova Labs tracks the experiment. Ficta's product surface stays canonical at ficta.sh.
POST /v1/messageson your machine
Authorization: Bearer sk-live-4a9f...c2
{
"content": "email me at ada@acme.co"
}ficta redact
-> leaves for the model
Authorization: Bearer sk-live-4a9f...c2
{
"content": "email me at FICTA_1b8e4d..."
}ficta restore
<- restored in the reply
{
"content": "Here is your draft to ada@acme.co"
}The provider receives required auth headers; protected payload values leave as tokens, and the mapping never leaves your machine.
Why Ficta is in Serova Labs
Ficta extends Serova beyond communication automation into trust infrastructure for AI workflows. Labs is the right place to show that direction without turning serova.ai into a second Ficta marketing site.
Full product detail
The current Gateway pitch, install path, FAQs, and product updates live on ficta.sh.
Proof links
Source, docs, and threat model stay one click away instead of being re-maintained here.
The boundary in one pass
This is the only mechanism summary we keep here. For exact coverage, current limitations, and setup details, use Ficta's product site and threat model.
Register what must stay local
Known sensitive values are loaded into the local boundary before the model request is prepared.
Send surrogates, not values
The provider sees useful request shape with FICTA_ tokens where protected values would have appeared.
Restore inside your environment
Replies are mapped back locally so users can read coherent output without sending the mapping away.