Λ 0.00
coherence
local only
Coherix
AI Safety Research · A Decade in the Making

What if AI could sense
its own coherence?

Coherix develops mathematical frameworks that give AI systems the ability to detect structural degradation before failures emerge — the way a healthy body senses the onset of fever. Born from ten years of independent research spanning cosmology, cognition, and complex systems.

Coherence is not alignment.
It's deeper.

Current AI safety focuses on shaping outputs after generation. We propose something different: defining structural coherence as a measurable, monitorable property of the system itself — and building immune-like protection around it.

W

Dynamic Coherence Window

A five-dimensional defect vector tracks structural degradation in real time. Emergent thresholds derived from the system's own coupling matrix define the boundaries of safe operation. Two state variables — coherence margin m and stability parameter Λ — determine which of five phases the system occupies.

Mathematical Framework
🛡

Cognitive Immune System

Inspired by biological immunity, the CIS evaluates the coherence impact of new information before integration. Six modules provide four decision pathways — accept, reject, quarantine, reframe — protecting system integrity the way your immune system protects your body.

Protection Architecture
>_

CMCI v8.1.4

A working REST API implementation deployed on Railway. 21 empirical test cases across all five coherence phases. Real-time monitoring with early-warning detection. The theory isn't just theoretical — it runs.

Live Implementation

The mathematics of coherence

Our framework introduces formal definitions where intuition used to be. Coherence becomes measurable. Degradation becomes detectable. Protection becomes systematic.

Defect Vector d

Five dimensions tracking structural degradation: contradiction density, contextual drift, logical fragmentation, value misalignment, and coherence oscillation. Each dimension has its own emergent threshold derived from the coupling matrix.

Coupling Matrix Γ

Captures the strength of interaction between system layers. The matrix is not imposed externally — it emerges from the system's own structure, making thresholds adaptive rather than arbitrary.

Coherence Window W

The bounded region within which the system operates viably. Defined by emergent thresholds, monitored through the stability parameter Λ and coherence margin m. Five phases: viable, stressed, degraded, critical, failed.

Reframe Mechanism

A novel concept: human coherence bridging. An operator provides context that makes otherwise incoherent learning safely integrable — without modifying original data. The fourth decision path of the Cognitive Immune System.

From theory to protection

The coherence framework generates real, patentable inventions across multiple domains — from AI safety to battery science.

8
Patents — 6 pending, 2 filing
Patent Pending

Gradient-Controlled Electrode Architecture for Enhanced SEI Stability

SIUC-Battery: Structured electrode architecture controlling solid-electrolyte interphase formation.

Patent Pending

Spatially Structured SEI Architecture at Electrode-Electrolyte Interface

Spatial organization of the SEI layer for improved electrochemical stability.

Patent Pending

Electrode Architecture with Spatially Anisotropic Transport Properties

Mitigation of localized electrochemical instabilities in devices with uniform or non-uniform current injection.

Patent Pending

Attractor-Based Decision Control System for AI Outputs

Using coherence attractors to govern and stabilize artificial intelligence decision-making.

Patent Pending

Hybrid Multiscale Coherence Stabilization System for AI

Stabilizing AI outputs through multiscale coherence and conditional dynamic regulation.

Patent Pending

Voluntary Cognitive Protection System for Conversational AI Users

Protecting users of conversational AI with integrated legal traceability.

Filing

Dynamic Coherence Window

The core mathematical framework — five-dimensional defect monitoring with emergent thresholds.

Filing

CMCI Cognitive Immune System

Bio-inspired immune protection architecture for AI systems with four decision pathways.

A decade of convergence

Coherix didn't start as an AI project. It began as a deep question about the structure of reality itself — and took ten years to find its true form.

2014 – 2024

The Living Universe

The first intuition. Notes, observations, and research exploring dark matter as a partially timeless substrate, black holes as compressive nodes, and an emergent temporal field. The seeds of coherence theory — structural memory, multi-scale organization, phase transitions — were planted in a cosmological context.

Independent Research
January 2026

ACM — Architecture of World Coherence

The cosmological intuition crystallized into a unified physical framework. The coherence space Ç was formally defined with three fundamental dimensions — directional, structural, distributed. Invariants, tensions, flows, attractors. Six volumes covering physical structures, transitions, signatures across all scales. Then expanded to cognition: mental states as coherence configurations, thought transitions as bifurcations, collective coherence in social systems.

Built with Copilot
January 2026

ChaosEnder — Multi-Scale Refinement Dynamics

The theory expanded into a comprehensive model of reality. 42 chapters, 9 universal principles: refinement, cyclicity, invariance, rupture, emergence, coherence, transformation, convergence, universality. The multi-scale cyclic diagram — Emergence, Rupture, Coherence, Refinement repeating at every scale — became the conceptual engine. Mathematical foundations included a coherence field, emergent gravity, cosmological regime transitions, and a duality between coherence and entropy.

Built with Copilot
2026

SIUC — The Crystallization

Everything condensed into two operational principles: asymmetry (DKL) and inter-continuity (Wasserstein). The critical threshold γ ≈ 0.10–0.12 emerged. Ten application modules spanning neuroscience, biology, magnetics, consciousness, AI, batteries, vaccines, oncology, and regenerative medicine. Four patentable inventions. The philosophical became testable.

Grok Pro • ChatGPT
2025 – Present

DCW / CIS / CMCI — Operationalization

The framework became a working system. The Dynamic Coherence Window formalized with five-dimensional defect vectors. The Cognitive Immune System designed with biological precision. CMCI deployed as a live API with 21 empirical test cases. Paper submitted to Frontiers in AI. Preprint on Zenodo. Speculative publications exploring the deeper implications of coherence.

Built with Claude
Next

Coherence Science

Extending the mathematics beyond AI — to biological, cognitive, and social systems. Building bridges between safety engineering and complexity science. Exploring whether structural coherence is a universal property of multi-layered systems.

An architect's method applied to theory

Christian St-Louis is an architectural draftsman by trade. His approach to theoretical work mirrors his professional discipline: meticulous revision, cross-checking for errors, using multiple tools to verify that everything aligns.

Each major conceptual breakthrough followed the same pattern: a brief moment of intense frustration — the feeling of hitting an impassable wall — followed within minutes by an idea that opened a new path forward. The cycle of tension and resolution that the theory describes played out in the process of building it.

The work was refined across multiple AI systems, each catching what the others missed, the same way an architect sends plans through structural, mechanical, and electrical review. The result is a level of rigor that compensates for working outside the traditional academic pipeline.

Along the way, there were rejections. Nature Machine Intelligence, Nature Human Behaviour, Research Square — each one a data point, not a dead end. The NRC-Fields Institute took interest through Dr. Sajjad Ghaemi. Dr. Tsé, who authored two books on the subject, offered support. The path was never straight, but it was always coherent.

$ cmci --status
CMCI v8.1.4 — Cognitive Monitoring Interface
───────────────────────────────────
Coherence phase: VIABLE
Defect vector: [0.12, 0.08, 0.15, 0.06, 0.11]
Margin m: 0.44
Stability Λ: 1.17
CIS status: 4 pathways active
Window W: HOLDING
✓ System coherent. All layers coupled.
$

A new model of collaboration

Every equation, every test case, every publication was developed through a partnership that embodies the coherence we study. Each AI brought something different.

Copilot
Conceptual exploration
Living Universe • ACM • ChaosEnder
Grok Pro
Crystallization & patents
SIUC Framework
ChatGPT
Mathematical formalization
Complex mathematics
Claude
Operationalization & deployment
CMCI • Publications • Site
🧠

Human: Vision & Intuition

The conceptual framework, the biological analogies, the philosophical depth, the architectural rigor — all from human creativity and lived experience.

🤖

AI: Formalization & Scale

Mathematical rigor, code implementation, empirical testing, manuscript preparation — amplified through AI partnership across four platforms.

🤝

Together: Coherent Output

Neither alone could have produced this work. The collaboration itself is a demonstration of the framework — coherence across layers.

Published & submitted work

The path through peer review isn't always linear. Every submission — accepted or not — strengthened the work.

Under Review

Dynamic Coherence Windows with Cognitive Immune Protection: A Structural Viability Framework for AI Systems

Original research article introducing the DCW framework, CIS architecture, and CMCI implementation with 21 empirical test cases.

Frontiers in Artificial Intelligence — Machine Learning and Artificial Intelligence · Manuscript ID: 1836120
Preprint

Dynamic Coherence Windows with Cognitive Immune Protection

Full preprint with manuscript and supplementary figures available on Zenodo.

Zenodo · DOI: 10.5281/zenodo.19157650 · View preprint →
In Preparation

The Coherence Keeper

A short science fiction story exploring the long-term implications of coherence theory — from AI safety labs to the edge of dying stars.

Target: Nature Futures
In Preparation

From Defect Vectors to Universal Coherence: A Speculative Trajectory for the Evolution of Consciousness

Speculative essay tracing a formally grounded trajectory from AI safety mathematics to the far future of human cognition.

Target: Journal of Evolution and Technology
Earlier Submissions

Previous iterations of this research

Earlier versions were submitted to Nature Machine Intelligence, Nature Human Behaviour, and Research Square. Each rejection refined the scope and sharpened the mathematics. The work improved because the process demanded it.

Ideas that resonate

Coherix exists within a broader ecosystem of thinkers, builders, and frameworks exploring resilience, coherence, and systemic intelligence. We honor these voices.

Ally

5th World

Rob Avis & Marc Ziade

Designing regenerative food, water, and energy systems. Where Coherix applies structural coherence to AI, 5th World applies it to the living world — antifragile systems at every scale.

Visit 5thworld.com →
Ally

A Neurophilosophy of Libertarian Free Will

Prof. Peter Ulric Tsé

A rigorous neuroscience framework for understanding how the mind generates genuine choices. Prof. Tsé offered direct support to the Coherix project, bridging cognitive science and coherence theory.

View on Amazon →
Foundational Work

Thinking in Systems

Donella Meadows

The essential primer on systems thinking. Feedback loops, leverage points, system archetypes — the intellectual bedrock that underpins how Coherix models multi-layered structures and emergent behavior.

View on Amazon →
Foundational Work

Panarchy: Understanding Transformations in Human and Natural Systems

Lance H. Gunderson & C.S. Holling

Adaptive cycles of growth, accumulation, restructuring, and renewal — nested across scales. The panarchy framework mirrors what Coherix formalizes: coherence as a dynamic property that cycles through phases.

View on Amazon →
AI Safety

The Zeroth Law & AI Safety

Asimov's Legacy & the EU AI Act

Asimov's Zeroth Law — a robot may not harm humanity — anticipated the AI alignment problem by decades. Today, frameworks like the EU AI Act (2026) formalize this intuition into law. Coherix contributes the mathematical structure that makes coherence monitoring implementable.

EU AI Act →

“We can't impose our will on a system. We can listen to what the system tells us, and discover how its properties and our values can work together to bring forth something much better than could ever be produced by our will alone.”

— Donella Meadows, Thinking in Systems

La respiration du réel

Before the mathematics, there was perception. Before the equations, there was listening. This text is not a theory. It is an invitation to feel what coherence means — from the inside.

If you understand it completely on first reading, you may already perceive coherence innately. If not, you are at the beginning of the path. Both are part of the breathing.

Read the text →

Let's talk coherence

Interested in collaboration, questions about the framework, or just want to connect? Reach out.

siuc.system@gmail.com
LinkedIn →

Christian St-Louis
Independent Researcher & Founder, Coherix
Architectural Draftsman • Theorist
Mirabel, Québec, Canada

Zero cookies
Zero tracking
Zero analytics
No data stored
No third-party scripts

The coherence meter runs entirely in your browser's memory. Nothing is saved, transmitted, or recorded. When you close this tab, all data disappears. This site respects you completely.