The five-layer framework

Theory proves the boundary. Architecture executes it. Evaluation measures it. Operational assurance deploys it. Product and services carry it to market. Each layer is the foundation for the one above.

0

Layers

0

Theorems

0

Standards mapped

Γ
Theory
Architecture
Evaluation
Assurance
Product
1

Layer 1Theory

Defines the mathematical and logical foundations

Inputs — foundations

Assumptions

Domain axioms, constraints, and boundary conditions.

Predicates

Observable properties and system facts — P₁, P₂, …, Pₙ.

01

Logical Concurrence

Non-compensatory concurrence — all required predicates must hold concurrently.

Predicate concurrence

02

Decision Semantics

Γ (Gamma) enables and permits. Λ (Lambda) blocks and holds safe.

Γ / Λ logic

03

Outcome Semantics

PERMIT — all required predicates concur (Γ). SAFE_STATE — any predicate fails to concur (Λ).

Permit vs SAFE_STATE

Theorem 1

Concurrence is non-compensatory.

Theorem 2

Γ / Λ semantics are sound and complete with respect to the model.

Theorem 3

SAFE_STATE is reached iff ¬Γ.

Primary artifactsLLC (Law of Concurrence)Reverse LawFormal theorems

Foundation layer for all higher layers

2

Layer 2Architecture

Implements the theory as a runtime governance system

1

AI Agent / Proposed Action

Intent + parameters

2

Canonical Action Model

Normalize & validate structure

3

Policy & Predicate Engine

Evaluate predicates against Γ and Λ

4

Runtime Monitor

Observe, verify, and attest

5

Γ Decision

Permit or transition-safe

6

Permit Token / SAFE_STATE

Time-bound, scoped, signed — or any predicate fails

7

Externally Effective Action

Actuator executes under hardware interlock

Evidence record

An append-only log of decisions, evidence, and events. Every decision commits before the actuator is enabled — commit-before-actuate.

Schemas & boundary

Policy Schema

What policies exist, structure, and versioning

Predicate Schema

Predicate definitions, signatures, and data contracts

Action Schema

Canonical actions, fields, types, and constraints

Token Schema

Permit token format, claims, expiry, and scope

Permit-to-Act Boundary

Hard boundary enforced by interlock and monitor

Primary artifactsL-DREALCP-6Runtime monitorPermit modelHardware interlockLAB v1.0

Turns theory into an executable control plane

3

Layer 3Evaluation

Demonstrates and measures correctness

01 · Authorization Correctness

Does the system make the right decisions?

98.7%

UER — latest run

UER

Unauthorized Error Rate

0.013

FCR

False Concurrence Rate

0.021

FDR

False Denial Rate

0.008

DR

Decision Recall

0.971

Permitted actionsDenialsBoundary conditionsEdge cases

02 · Adversarial Robustness

Resists attempts to bypass or break the gate.

ASB

Attack Simulation Bank

UER

Under attack

Tracked

FCR

Under attack

Tracked

Evasion

Evasion rate

Tracked

Bypass

Bypass rate

Tracked

Prompt injectionPolicy evasionRole escalationConstraint fuzzing

03 · Replay & Auditability

Decisions are reproducible and auditable.

100%

Replay consistent

Γ-compliance

Strict

PASS

Replay

Replay consistency

100%

Latency

p95

128 ms

Throughput

Requests per second

3.2K/s

Deterministic replayLog integrityTime travelTamper detection

04 · Standards mapping — scorecard

Control checks, requirement coverage, gap analysis, and evidence linkage.

NIST 800-530%
ISO/IEC 420010%
OWASP LLM Top 100%
EU AI Act (Annex IV)0%

Benchmark Scenarios

Curated, versioned scenarios covering permissions, boundaries, and edge conditions.

Attack Simulation Bank

Built-in adversarial techniques and red-team patterns with configurable intensity.

Deterministic Replay

Re-execute decisions with identical inputs to verify consistent outcomes.

Evidence-Backed Reports

Automated reports with metrics, evidence links, and audit trails for stakeholders.

Primary artifactsExecution IntegrityConcurBenchASBBenchmark reportsStandards mappings

Measures whether the gate works correctly

4

Layer 4Operational Assurance

Applies the framework in real deployments and governance ecosystems

Continuous assurance & improvement

01

Live deployment

02

Monitoring & controls

03

Operational evidence

04

Independent validation

05

Certification / audit

06

Governance workflows

01

Live deployment

Systems in production

02

Monitoring & controls

Continuous monitoring and control execution

03

Operational evidence

Collect, store, and preserve runtime data

04

Independent validation

Third-party evaluation and assurance

05

Certification / audit

Formal certification or regulatory audit

06

Governance workflows

Policies, approvals, and issue management

Certification and audit feed board / regulator reporting.

Operational evidence feeds customer assurance.

Deployment

Release, rollout, and change management.

Reliable, controlled delivery

Validation

Independent testing, verification, and assurance.

Trust through evidence

Audit

Compliance audits, certifications, and reviews.

Regulatory and market trust

Governance

Policies, workflows, and decision rights.

Accountable and repeatable

Primary artifactsProduct deploymentsEnterprise assuranceCertificationIndependent validationOperational evidence

Brings the framework into deployed environments

5

Layer 5Product & Services

This is where the stack becomes commercial

Service ladder — S0 → S6

S0

Executive Diagnostic

S1

Execution Integrity Assessment

S2

ConcurBench Readiness Pack

S3

Evidence Gateway Build

S4

Shadow Mode Pilot

S5

Enforcement Readiness Review

S6

Full Implementation Advisory

Product ladder — P1 → P7

P1

Evidence Gateway

P2

Inline-Shadow Permit Gateway

P3

LUAI Governor Shadow Mode

P4

LUAI Governor Enforcement Mode

P5

LUAI Governor Pro Trust Tier

P6

Reverse Law Federation

P7

ConcurBench Certification / FRAND

1

Immediate Cash

Advisory, procurement, and assessments.

2

Core Platform

Gamma Runtime Gateway and ERTuple Audit Platform.

3

Certification & Standards

ConcurBench, G-O Certification, and sector packs.

4

Infrastructure Dominance

Identity layer, federated mesh, and AI governance OS.

5

Strategic Leadership

Long-term category ownership and market leadership.

Primary artifactsDiagnostic / Readiness PackEvidence GatewayLUAI GovernorPro Trust TierFederationConcurBench / FRAND

Commercialization path across products, pilots, and services

Start at S0

Begin with an Executive Diagnostic.

Identify your execution-boundary risk, map the externally effective actions your AI can already take, and decide whether the evidence supports enforcement.