FAQs

No. Core is the substrate your systems plug into. We don’t need access to your model internals and you don’t need to retrain. (XAI-Lite is the lightweight wrapper you can land first; it plugs into Core later.)

XAI-Lite = fast on-ramp: decision receipts, plain-English reasons, and pre-release policy/ethics checks around one flow.

Core = full governed decision substrate across the estate: common grammar, simulation and selection, ethics before execution, federation/sovereignty, and the classical-to-quantum portability bridge

Weeks, not months. XAI-Lite pilots land first; Core pilots extend coverage and capabilities.

Yes. PARCIS produces per-decision evidence you can replay, shows which oversight checks were applied, and exports a regulator-friendly technical pack (e.g., Annex IV alignment). It supports your compliance work but is not legal advice.

Each audience gets the view they need—end-users, operators, auditors, and regulators—drawn from the same governed record.

Yes. Policies are configurable per jurisdiction and product. Changes follow maker–checker approval and are logged for audit.

Default is Hybrid: inline guardrails plus off-path analytics. You can also choose a posture per flow:

Black — observe-only: off-path bus-tap (Kafka/AMQP) and/or API/SDK adapters. No gating; no latency risk.

Grey — advisory inline: gateway proxy with soft-fail pass-through; flags and routes (e.g., human-in-the-loop).

White — enforcing inline: gateway proxy with hard pre-release policy & ethics checks (permit/modify/block).

No. It runs inside your boundary (e.g., your cloud VPC). By default, it stores only minimal metadata and evidence hashes; raw sovereign data and PII remain in your systems.

Yes. Core runs fully local (air-gapped). When needed, it synchronises during scheduled windows using signed trust bundles, not raw data—so sovereignty is preserved.

Inline guardrails add a small overhead; the off-path listener has no user-visible latency. You choose posture per flow.

APIs/SDKs, gateways and message buses, plus hooks into GRC/SIEM/case-management tools. Evidence exports to CSV/PDF/JSON.

A standard pack per decision stream with:

(1) a signed replay

(2) the safety/oversight checks that were active

(3) a technical dossier aligned to regulator expectations.

It’s machine-checkable and exportable, which shortens investigations.

You can replay the decision, see the checks in force at the time, and export the pack—dramatically reducing reconstruction effort.

No. Governance artefacts and evidence export in open formats, and the approach is portable across vendors—and, when appropriate, to quantum compute.

Retention is configurable; evidence stores minimal fields. Deployments respect customer data-residency requirements.

Single sign-on and role-based access are supported. Evidence is encrypted in transit/at rest, and customer-managed keys (BYOK) are available.

You benefit today on classical infrastructure. Our IP includes a synthetic-quantum path and a bridge to quantum processors, so governance and evidence carry forward when quantum is appropriate—no re-audit just because compute evolves.

Pilot first (often with design-partner credits on success), then platform subscription and services as you move into production.