About the Themis Project
Semantic validity architecture for AI systems
The Themis Project is a research and development initiative dedicated to solving the ontological crisis in artificial intelligence. We believe that current AI architectures are fundamentally incomplete because they lack the structural capacity for knowing. They process information through statistical correlation without validity through semantic grounding.
We have developed the Tetrahedral Ontological Closure Architecture — a framework that enforces six necessary constraints on AI outputs: Referential, Contextual, Premissive, Inferential, Constraining, and Teleological. This architecture is protected by ten patent families (A through I-Extension) and is available for licensing and integration.
The Origin
The project began with a simple question: Why do LLMs hallucinate?
The answer was not “insufficient data” or “temperature settings.” The answer was that they have no mechanism for checking if a claim is valid. They only check if it is probable.
The structure of validity was recovered from fifteen years of studying ancient Greek philosophy in original texts — Homer through Aristotle, read in Greek, never in translation. The six constraints emerged not as invention but as discovery: the geometric minimum for semantic completeness, encoded in Aristotle’s four causes and mapped onto the simplest Platonic solid.
The result is a system that can say “I don’t know” — and mean it.
The Founder
Steven Easley brings an unusual combination of rigorous financial discipline and deep philosophical scholarship to the problem of AI validity.
His professional career began at Ernst & Young as a CPA, followed by partnership at his own firm, CFO roles, and work as a turnaround specialist where he raised over $25 million in funding. This background in financial rigor — where claims must be auditable and structures must close — shaped the architectural thinking that would later produce the Tetrahedral Architecture.
After this career, he spent fifteen years studying ancient Greek philosophy in original texts, developing fluency in the philological methods required to read Homer, Heraclitus, Socrates (through Plato), Plato, and Aristotle as they wrote — not as translators interpreted them. This work yielded the discovery that Aristotle’s framework for knowing (εἰδέναι) encodes a geometric structure: exactly four components connected by exactly six constraints, forming the minimal closed solid in three-dimensional space.
The insight that this ancient structure maps directly onto the failures of modern AI systems — and can be engineered into patentable architectures — is the foundation of Echosphere.
The Patent Portfolio
Ten patent families protect this architecture:
| Family | Coverage |
|---|---|
| A | Constraint-governed AI output validation |
| B | Form-preserving memory and semantic persistence |
| C | Logos-based judgment and inference validation |
| D | Rhetorical state signaling and discourse marking |
| E | Cognitive pedagogy and structural learning |
| F | Multi-domain constraint instantiation |
| G | Tetrahedral closure verification |
| H | Semantic condensation and reconstitution |
| I | Hexis-governed cognitive pedagogy |
| I-Ext | Extended pedagogical and diagnostic applications |
Contact
For licensing inquiries, research collaboration, or technical questions:
Echosphere.io LLC
Orlando, Florida
GitHub ↗