We write papers about everything we build — not because academia requires it, but because papers force honest thinking. Every claim needs a hypothesis. Every hypothesis needs an experiment. Every experiment needs results reported as they are, not as we wished they were.

All papers are published as open-access preprints on Zenodo under CC BY 4.0.

What We've Demonstrated

These are the capabilities we've validated with experimental data. Each links to a paper with full methodology, results, and limitations.

Learning

Capability Result Paper
Knowledge extraction from documents 95%+ provenance attribution, 1.8M+ triples 102
Continuous learning during conversation 97.6% extraction precision, +33% graph growth in 8hr 104
Knowledge transfer between beings Equal quality, 100,000x faster than corpus learning 113

Reasoning

Capability Result Paper
Sub-5ms knowledge retrieval P95: 3.25ms, P99: 4.16ms — zero hallucination 103
Unified perception cycle 9/9 hypotheses validated, 0% hallucination 108
Episodic memory for mistake prevention 85.7% mistake prevention rate (p=0.006) 107
Domain-aware safety routing 0 safety violations, sub-millisecond overhead 114

Knowledge Organization

Capability Result Paper
7-layer cognitive architecture (Y0-Y6) Procedural compilation: 85.7-100% error elimination 117/119/120
Predictive memory loading 76.2% next-topic prediction accuracy 110
Hierarchical concept indexing 97,000x speedup over full-graph loading, O(1) at 1M concepts 112
Metacognitive self-monitoring 4.8% Expected Calibration Error, 100% error classification 117/119/120

Evaluation

Capability Result Paper
AGI certification framework 10-dimension ACF, 150,347 triples/GPU-hour (10x vs LLM) 122

All Papers

Published on Zenodo

# Title Key Result PDF
101 The NuSy Brain Architecture 4-layer knowledge model, 11.3ms latency, 0% hallucination DOI
102 Training Neurosymbolic Beings 95-100% provenance, 100-1000x faster than fine-tuning DOI
103 Sub-5ms Reasoning Performance P95: 3.25ms, 1,345x speedup, 0% hallucination DOI
104 Continuous Learning / Crystallization 97.6% precision, +33% graph growth, 0 LLM calls on learned knowledge DOI
107 Graph-Native Episodic Memory 85.7% mistake prevention, 33.3pp decision quality improvement DOI
108 Unified Cognitive Architecture 9/9 hypotheses validated, 0% hallucination, 100% task autonomy DOI
110 Semantic Field of View 76.2% next-topic prediction, 42.8% auto-answer rate DOI
112 Hierarchical Concept Index 0.003ms lookup, 97,000x speedup, O(1) at 1M concepts DOI
113 COG Transfer Experiment Equal quality, 100,000x faster knowledge transfer DOI
114 Domain-Aware Hallucination Prevention 0 safety violations, 0.0013ms overhead, 6 domains DOI
117/119/120 Cognitive Layer Architecture (Y0-Y6) Y5: 85.7-100% error elimination, Y6: 4.8% ECE DOI
118 Lean Hypothesis Testing 5/5 hypotheses failed — honest failure paper, methodology value DOI
122 AGI Certification Framework 10-dimension ACF, UER 10x vs LLM, Zorblaxia battery DOI

On Hold or Deferred

# Title Status
109 Curiosity-Driven Learning Deferred
111 Autonomous Being Development Deferred
116 Textbook Knowledge Integration Abandoned — blocked until multimodal vision

Papers 105, 106 are superseded by later work. Papers 115, 121, 123-130 are early outlines awaiting implementation.

Newer campaigns (preprints in progress)

The papers above cover the V7–V15 architecture. Since then: V17 (causal vector-symbolic dedup — tested and refuted), V18 (executable reasoning + the provable gate), and V19 — the NuSy LRM, a provable clinical reasoner where hallucination is zero by construction on the symbolic path. Preprints for V18/V19 are in progress.

Research Integrity

Contact: [email protected] | ORCID: 0000-0002-3912-0080