What is UOR?
UOR (Universal Object Reference) is a framework that creates a shared coordinate system for digital artifacts, enabling semantic clarity and traceability across systems.
It extends the object-oriented programming paradigm to the entire universe, treating every entity—whether digital or physical—as an object with properties, relationships, and behaviors.
UOR was first introduced in a pioneering article from Red Hat's Emerging Technologies blog, which laid the foundation for what would become the UOR Foundation.
Universal Object Reference
The semantic foundation for all PrimeOS applications
Core Components
UOR treats every application as a structured logic circuit with three essential components that have direct parallels in modern AI/ML systems
Canon
The source — similar to RAG inputs or trusted documents in AI systems, but with a crucial difference: Canon is not just a database but an ontological core of the system.
Unlike RAG systems that retrieve facts from external databases that the LLM doesn't deeply understand, UOR's Canon is a native part of the system: an encoded, evolving library of concepts, rules, and relationships that shape the reasoning process itself.
Cortex
The logic engine — analogous to LLMs like GPT-4 or Claude, but operates on structured reality circuits rather than statistical correlations.
While AI systems mostly detect patterns in huge datasets, UOR's Cortex runs structured, recursive circuits through layered meaning stacks (the Primal Domain Stack), embedding logic and geometry at each level for deterministic, reproducible results.
Codex
The output — similar to a structured output log in AI systems, but using Containers rather than vectors or tokens.
Unlike AI outputs that are often represented as floating-point vectors (embeddings), UOR's Codex uses structured Containers that are symbolic, interpretable, and verifiable by design, with complete lineage information for tracing exactly how each conclusion was reached.
The 12-Layer Model
All UOR-powered applications follow a 12-layer model that ensures every step of computation is well-defined, auditable, and deterministic. This layered approach enables semantic clarity and traceability across systems.
UOR vs. Traditional AI/ML Frameworks
UOR represents a fundamental paradigm shift from current AI approaches, focusing on meaning and structure rather than statistical patterns
Meaning-First, Not Model-First
Traditional AI starts with statistical models (like neural networks) and adds knowledge as a plugin (e.g., RAG). UOR inverts this approach.
UOR starts with structured meaning (the Canon), where knowledge is defined up front — the logic engine operates on known, stable units rather than probabilistic approximations of meaning.
Handles and Containers vs. Tokens and Vectors
AI/ML frameworks represent meaning with floating-point vectors (embeddings), often opaque and hard to audit.
UOR uses Handles (stable references) and Containers (structured outputs), which are symbolic, interpretable, and verifiable by design, making every step of computation transparent.
Structured Reality Circuit vs. Statistical Correlation
AI systems mostly detect patterns and correlations in huge datasets, leading to impressive but sometimes unpredictable results.
UOR systems run structured, recursive circuits through layered meaning stacks (the Primal Domain Stack), embedding logic and geometry at each level for consistent, explainable outcomes.
Coherence and Stability over Probability
ML models rely on probabilities and often produce inconsistent outputs across runs, making them unsuitable for applications requiring absolute reliability.
UOR circuits emphasize coherence norms and deterministic mappings, aiming for stable, reproducible answers that can be verified and trusted in critical applications.
End-to-End Semantic Traceability
In LLMs, tracing why the model said something is fuzzy and approximate, often requiring post-hoc explanations that may not reflect the actual computation.
In UOR, each step in the circuit — from input Handle to final Codex Container — is traceable and grounded in a logic stack, enabling complete auditability of the reasoning process.
Symbolic-Structural Fusion
Most ML models are black boxes: they learn weights but don't expose logic, making them difficult to debug or improve systematically.
UOR integrates symbolic reasoning (like a logic engine) with structured transformation layers, enabling both computation and explanation in a unified framework that combines the best of rule-based and learning-based approaches.
UOR Framework Foundation
Our framework is built on three key technical pillars that form the foundation for all our products
Math Package
A comprehensive mathematical foundation that enables deterministic computation, cryptographic verification, and semantic reasoning across distributed systems.
- Post-quantum cryptographic primitives
- Verifiable computation models
- Semantic graph algorithms
Data Protocols
Standardized data exchange formats and protocols that ensure semantic integrity, interoperability, and sovereignty across all UOR-native applications.
- Universal Object Reference format
- Semantic graph exchange protocols
- Data sovereignty mechanisms
Client/Server
A distributed architecture that enables secure, verifiable communication between clients and servers while maintaining data sovereignty and semantic integrity.
- Trust-minimized communication
- Deterministic execution environment
- Peer-to-peer verification mechanisms
From Foundation to Products
Foundation Layer
Product Layer
BUILD
SHIP
RUN
Key Benefits
How UOR transforms the way we work with data and computation
Semantic Context
UOR preserves the meaning and context of data as it moves between systems, addressing the semantic drift problem common in AI pipelines where information loses fidelity across processing stages.
Unlike AI systems where meaning is approximated through statistical patterns, UOR's approach ensures that semantic intent is preserved throughout the entire computation process, making it particularly valuable for applications requiring high precision.
Distributed Coherence
By providing a universal reference system, UOR enables disparate AI systems to maintain a coherent understanding of shared objects, even in decentralized environments where traditional knowledge graphs struggle.
This makes it possible to build truly distributed AI applications that maintain semantic integrity across nodes without requiring centralized coordination.
Data Sovereignty
UOR supports true data ownership by allowing individuals and organizations to maintain control over their data while still enabling secure, verifiable AI computation and inference.
This shifts the paradigm from centralized AI training on pooled data to a model where computation moves to the data, preserving privacy while enabling collaborative intelligence.
Deterministic Computation
Unlike the stochastic nature of neural networks, UOR ensures that computations are deterministic and reproducible, with every step traceable back to its inputs and transformation logic.
This coherence-first approach addresses a fundamental limitation of current AI systems, enabling the kind of reliability and verification that is essential for high-stakes applications in healthcare, finance, and critical infrastructure.
Applications
The UOR Framework powers a range of products developed by the UOR Foundation
Prime SDK
A developer toolkit for building UOR-native applications with deterministic behavior and full source traceability.
Learn morePrime AppStore
A distribution platform for UOR-native applications, ensuring semantic integrity and source transparency.
Learn morePrimeOS
A trust-minimized runtime environment for executing UOR-native applications with guaranteed deterministic behavior.
Learn moreJoin us in creating a world where data has meaning, people have ownership, and systems work together seamlessly.