Sovereignty Reality Check
DETERMINISM
LLMs are probabilistic by design. VertexPace builds expert systems for when you need a traceable cause-and-effect chain, not a coherent-sounding guess.
SOVEREIGNTY
Enterprise AI interactions are logged and used to retrain models without explicit consent. Lumina runs on your hardware, under your control.
LATENCY
Cloud-dependent AI inference adds 100–800ms of network latency. Lumina logic resolves in under 1ms on a Raspberry Pi.
AUDITABILITY
When a model hallucination causes an operational failure, there is no "why" in the logs. Lumina decisions are traceable back to grounded facts.
VENDORS
Organisations depending on cloud APIs have zero control when models are deprecated or prices are hiked. Sovereignty is your hedge against vendor lock-in.
HARDWARE SOVEREIGNTY

LUMINA OS

A deterministic expert system that processes intelligence as a fixed 75-concept grid. Runs on macOS and Raspberry Pi 5 today.

Hardware Sovereignty

Deterministic. Auditable. Local.

LUMINA OS is a fundamental rewrite of the relationship between hardware and intelligence. Built as a deterministic expert system, Lumina provides absolute independent intelligence. Where other AI systems phone home, Lumina thinks locally—on your hardware, with your data never leaving the device. It's not a chatbot; it's a high-speed logic engine.
  • The Genesis GridGenesis GridA 75-concept semantic encoding where every operation maps to a single byte.: A 75-concept semantic encoding where every operation maps to a single byte.
  • Inference EngineInference EngineA Euclidean resolver that finds logical relationships through spatial computation.: Resolving logical relationships through Euclidean spatial computation—no neural network required.
  • Fact GroundingFact ValidationA hardware-level grounding mechanism that filters unverified data.: Thoughts must be traceable to loaded facts to stay on the grid—unverified data is filtered.
Meaning-Based Recall

The Centroid Lookup.

Lumina identifies the core of a thought using spatial relationships, moving beyond keyword matching. It converts text into a geometric map. When no exact match exists, the engine finds the nearest concept by Euclidean distance on the grid. This means Lumina finds meaning even when exact words differ—because it matches the shape of ideas, not the letters.
  • Boil-DownDe-noisingA process that reduces thoughts to their Light Language core.: Reducing input to its Light Language core.
  • MappingSpatial MappingA mapping that positions concepts at deterministic coordinates on the grid.: Representing thoughts as physical constellations.
  • Centroid InferenceCentroid LookupA method for recalling conceptually related memories via Euclidean distance.: Recognising conceptual relations through centre-of-gravity proximity on the grid.
Semantic Compression

Light Language Protocol.

All communication within Lumina OS is handled by Light Language—a compact semantic ontology. The resulting protocol has demonstrated an order-of-magnitude reduction in payload size versus general-purpose compression for AI corpora. Less data, same meaning, zero translation overhead.
Containing hallucination

The Grounded Filter.

LUMINA OS uses a filter called Fact Grounding to keep unverified data off the grid. Every entry is checked against grounded sources and logical consistency to remain in active memory. An answer is always traceable to a grounded fact, not invented. This provides deterministic lookups instead of probabilistic generation.
Distributed Intelligence

The Autonomous Swarm.

Our goal is a peer-to-peer network where specialised agents share hyper-focused knowledge to complete large tasks collaboratively. Each node on each device focuses on one topic, contributing its expertise to the wider mesh. Lumina maintains self-verified integrity through continuous grounding against its local knowledge set.
  • Collaborative Nodes: Agents communicate across devices to share specialised data.
  • Energy Efficiency: Distributed intelligence modules that operate without centralised power draw.
  • Auditable LogicSelf-VerificationAutonomous reflection where Lumina synthesises its own logs to maintain integrity.: Reflective AI that maintains honesty through traceable, deterministic lookups.
Current Status

What's working today.

AreaStatusWhat that means in v0.4.0
Expert SystemImplemented75-concept grid encoding and deterministic Euclidean inference engine.
Core stabilityIn testingNo crashes seen in internal long-runs on Raspberry Pi 5.
LatencyVerifiedSub-millisecond local in-memory lookups on target hardware.
Fact GroundingImplementedDeterministic recall of facts traceable to the user-loaded knowledge set.
Performance claims derived from internal tests. We value audibility over probabilistic sophistry. Methodology shared by email.

GET IN TOUCH.

Interested in Lumina OS, NovaGlyph, or any other VertexPace project? Let's talk about deterministic expert systems and sovereign hardware.
EMAIL ZAC
robertpage@vertexpaceorganization.com
© 2026 VERTEXPACE. NOVAGLYPH KIT AND BABY AI LITE ARE PROPRIETARY PRODUCTS.
LIGHT LANGUAGE SPECIFICATION AND CORE COMPILER SOURCE AVAILABLE UNDER OPEN-SOURCE TERMS.
STATUS: ACTIVE R&D · SOVEREIGN BY DESIGN · 🐝