How it works

One engine. Swap the detector.

Suvard sits in-line between your people and the AI, inside your own network. Every request passes through the same four moves — in every sector. Only the detector changes.

Detectsector-tuned recognition of what must never leave
Maskreplaced with safe placeholders before the AI ever sees it
Blockwhat should never leave at all is stopped at the door
Loghash-chained, court-ready evidence of every decision

The same engine — through your sector's lens

Watch the same four moves on your data.

detector: CPR · clinical ontology Live
What your employee sends

A clinician asks the AI:

Draft a discharge summary for Jens Hansen, CPR 010190-1234, admitted with atrial fibrillation, treated with Marevan.

What leaves your network

Same question — nothing sensitive in it:

Draft a discharge summary for [PATIENT NAME], CPR [CPR NUMBER], admitted with [DIAGNOSIS], treated with [MEDICATION].

Live today — validated against a real clinical evaluation corpus. (Example data is synthetic.)

detector: SCADA · OT tags In development · illustrative
What your employee sends

A grid engineer asks the AI:

Why is turbine WT-A117 at Anholt East underperforming? SCADA tag A117-GEN-RPM reads 1,180 rpm on feeder F-42.

What leaves your network

Same question — nothing sensitive in it:

Why is turbine [ASSET ID] at [SITE] underperforming? SCADA tag [SCADA TAG] reads 1,180 rpm on feeder [GRID TOPOLOGY].

Illustrative — the energy detector pack is in development. The engine underneath is the same.

detector: IBAN · MNPI In development · illustrative
What your employee sends

An analyst asks the AI:

Draft a client note for Peter Madsen, IBAN DK50 0040 0440 1162 43, on the pending acquisition of NordFin A/S.

What leaves your network

Same question — nothing sensitive in it:

Draft a client note for [CLIENT NAME], IBAN [IBAN], on the pending [MNPI].

Illustrative — the finance detector pack is in development. IBAN detection is deterministic (mod-97) and quick to ship.



Questions we hear

Straight answers.

Is Suvard a cloud service?

No. Suvard is deployed inside your own network, on your own infrastructure. Nothing sensitive leaves the building — that is the point.

Do we need to run our own AI models or GPUs?

No. The shield works in front of the cloud AI your teams already use. Local models are an option, never a requirement.

Which AI services does it work with?

Suvard sits in-line as a proxy, so it is provider-agnostic: ChatGPT, Claude, Gemini, Copilot — same engine, same evidence.

What is live today — honestly?

Healthcare, through CareProxy: live pilots; in dialogue with a Danish region. The energy and finance detector packs are in development; the engine underneath is the same.

Will Suvard make us compliant with GDPR and the EU AI Act?

No tool can promise that. Suvard supports compliance: it enforces your policies technically and produces the evidence. Your legal assessment stays yours.

What evidence do we get when the regulator calls?

A hash-chained, Ed25519-signed audit log of every AI request and every decision — exportable in one click and verifiable without trusting us.

How is this different from a firewall or classic DLP?

Firewalls see packets; classic DLP sees patterns. Suvard understands what is being sent to AI — masks it with placeholders your people can keep working with, blocks what must never leave, and proves what happened.

How do we start?

A conversation, a demo, then a pilot in your own environment. Write hello@suvard.com.

See it run on your own data.

Book a demo