Governed conversational AI

QiXChat is where teams use AI with controls switched on.

QiXChat gives teams a familiar chat workflow while IOA governance runs underneath. You can start in shadow mode, move to enforce mode, and attach domain governance layers like QiXHealth or QiXCite when workflows become regulated.

Current access is by invitation through limited early-adopter pilots while validation is in progress.

What changes with governance

  • Policy checks before model calls
  • Session-level evidence and audit signals
  • Tenant controls for rollout posture
  • Domain adapters for sector workflows
  • One chat surface across multiple model providers

How teams start

Roll out in phases instead of forcing one big change.

Phase 1 — Shadow mode

Observe usage and policy outcomes without blocking users. Build a baseline before hard controls.

Phase 2 — Enforce mode

Block or flag governed violations in runtime with clear user-facing reasons and evidence.

Phase 3 — Domain adapters

Attach QiXHealth, QiXCite, and other frameworks when use cases need sector-specific controls.

Where it fits

QiXChat is the surface. QIX frameworks provide domain governance.

Layer Role Example
QiXChat Day-to-day governed chat interface for clinicians, legal teams, and operators. Clinician asks a query in a familiar chat UI.
QiXHealth Clinical consent, patient-information handling, and evidence controls. Session is blocked without valid scoped consent.
QiXCite Citation verification and quote-lock controls for legal workflows. Draft filing is checked before submission.

Current status

What is live now for managed rollouts.

Live capabilities

  • Shadow, enforce, and consensus governance modes
  • Runtime policy and evidence emission per session
  • Domain adapter path for QiXHealth and QiXCite
  • Tenant-scoped controls in IOA Cloud Console

Rollout boundaries

  • Some sector adapters are pilot-gated rather than self-serve
  • Integration depth differs by workflow and tenant posture
  • High-assurance healthcare paths remain assisted-onboarding
  • Regulated production claims should stay aligned with the specific deployment posture

Next step

Start with one governed workflow, then expand.