Skip to content
Technical focusPolicy-bound AI-to-device action across legacy IoT and OT paths

Technical order engine

Ritqo

Decide deterministically whether an AI or human command may reach a protected device.

Request a private briefing
Product 05Authorize

The operational problem Ritqo is designed to solve.

An AI model can propose a useful action, but it should never become the authorization authority for a physical or operational system.

Ritqo separates intelligence from permission. It binds identity, workload, device capability, current state, policy, risk, approval, expiry, nonce, replay controls, and provenance before issuing a signed command envelope or an explainable denial.

QUANTUMGUARD / RITQOIllustrative product view
Action request / Chiller 07

Authorization decision

Decision signed
Normalized intentAdjust cooling setpointdevice/chiller-07 → 18.5°C
Workload identitymaintenance-02
Device capabilitysetpoint.adjust
Current statemaintenance window
Human approvalAPR-884
Expiry + replayfresh / nonce unused
AllowSigned envelope QG-1042Bound to destination, capability, policy, and expiry
Example view: intelligence proposes the action; deterministic policy grants permission.

Operational contract

What enters, what Ritqo does, and what leaves.

Inputs

Evidence entering the boundary

  • Normalized action intent
  • Identity, capability, and device state
  • Policy, approval, risk, and provenance
Ritqo

Authorize

Decide deterministically whether an AI or human command may reach a protected device.

Outputs

Evidence moving forward

  • Authorization decision
  • Signed command envelope
  • Explainable denial evidence

Core capabilities

A focused part of one controlled system.

  1. 01

    Normalize AI and human intent into a versioned action request.

  2. 02

    Evaluate identity, capability, device state, policy, risk, approval, freshness, and provenance.

  3. 03

    Issue a signed, bounded envelope when every required condition passes.

  4. 04

    Return deterministic denial reasons when an action is stale, replayed, unsupported, or unapproved.

Operational value

What teams gain.

  • Keep probabilistic intelligence outside the permission boundary.

  • Make every allowed or refused action attributable and reviewable.

  • Send only bounded commands to Action Shield for gateway enforcement.

Continue through the control loop

Next: Action Shield

Explore Action Shield Full platform