The Authorization Layer for Every Automated Decision | Demonstrator V1
NON_CONFORMANT Rate: 0% (0 / 0 checks)
Demo
Audit Log
Product
About
Select preset or use Custom:
Audit Log — CIDRE
ID
Timestamp
Use Case
Result
Question
Answer
Hash
What ORYNX becomes
The Authorization Layer for Every Automated Decision.
Every output — from any system — checked against a client-defined reference framework before it produces legal effect. Court-admissible. Hashsecured. Retrievable on demand. Continuously.
This is a Demonstrator. Full production implementation follows with the first strategic partner.
Client-Side Reference Framework — The client defines the criteria. ORYNX checks against them. Provider and deployer are separated by architecture, not contract.
Court-Admissible Finding
Every Finding-Report is hashsecured and timestamped. References the exact framework version at the moment of decision. Retrievable on demand. Designed for evidentiary requirements.
RLCF — Improvement Signal
Every NON_CONFORMANT finding generates a structured corrective signal back to the model. The only real-world, auditable KPI for LLM performance in regulated deployments. This data exists nowhere today.
Non-Disclosable Criteria
Internal rules — age-based risk thresholds, denial criteria, pricing tiers — marked as non-disclosable. ORYNX ensures they never surface in outputs. Drift is detected before the regulator does.
Roadmap
V0 — Complete
Documentation
Architecture. Whitepaper. API schema. Regulatory Mapping. MindForge Mapping.
V1 — Live
Demonstrator
Four regulated use cases. CIDRE audit log. Hashsecured Finding Reports. RLCF export.
Multi-step workflow authorization. Full RLCF pipeline. Agentic chain governance.
Academic Analysis — Nannini et al. (arXiv 2604.04604, April 2026)
"A fourth tier is absent: infrastructure governing human-agent interaction at runtime, capable of classifying individual actions against a structured ontology and maintaining an immutable oversight record. The absence of this tier is not only a market gap: it is a compliance gap."
"High-risk agentic systems with untraceable behavioral drift cannot currently be placed on the EU market." — Conclusion (8)
The first partner gains what no later partner can achieve.
Permanent reputation and market position as the one who set the standard.
Every output is checked against a client-defined reference framework. Binary result: CONFORMANT or NON_CONFORMANT. Hashsecured. Timestamped. Retrievable on demand.
ORYNX diagnoses. It does not decide.
Nannini et al. (arXiv 2604.04604, April 2026) identify ORYNX as the missing fourth tier in AI governance tooling — the action-level authority record that meets EU AI Act Articles 12–14.
Oliver Richter — Founder ORYNX
Background in emergency medicine and aviation logistics. Work where every decision is documented because liability is real. ORYNX was recognised, not invented — the same structural problem that exists in regulated medical decision-making exists in every automated financial, insurance, and compliance decision. In medicine, the authorization layer exists. In AI, it did not.