Security and deployment
Your data stays on your site.
Synsa is built for companies whose operational data cannot leave the building: regulated, export-controlled or simply too valuable to send to a cloud.
-
Runs on your hardware
Installs as one signed package. Works without internet.
-
Read-only
Synsa reads your systems and writes nothing back.
-
Sensitive data masked
Names and personal data are masked on arrival, before any model sees them.
-
Your experts decide
Every pattern is confirmed by your people before it is used.
-
Every answer has a source
Each match links back to the original records and the versions that produced it.
-
Your memory is yours
Patterns built from your data stay with you.
Every answer can be traced and replayed.
Pick any step to see what Synsa records about it. This is one real answer's history, anonymised.
What Synsa showed the engineer, and the records it cited.
Match strength 0.87. Cites 3 source records.
Audit trail
A tamper-evident log of every read, match and decision. Exports to your security monitoring.
Replay
Any answer can be re-run with the exact versions that produced it.
Standards support
Built to support EU AI Act record-keeping and the traceability expected under IATF 16949 and AS9100.
Deployment that fits your site.
-
Pilot
A CPU server or a laptop. Enough for a first memory and a small history.
-
Standard site
One GPU server. One site, a few hundred users.
-
Large or multi-site
Several GPUs. More users, several sites, one memory per site or shared.
Questions security teams ask
- Does any data leave our network?
- No. Synsa runs on your hardware with no outbound connection. Updates arrive as signed packages that your team brings in.
- Do you train your model on our data?
- Only on your site, only with your written agreement, and the result stays with you.
- How do you keep the model's behaviour stable?
- Versions are pinned. Every update is tested against a fixed set of your confirmed cases before it goes live, and can be rolled back in one step.
- Can we see how it works before we decide?
- Yes. We hold a technical briefing for your architects and security team under NDA.