We build AI systems grounded in physics and mathematical first principles.
Hamiltonian dynamics. Statistical mechanics. Spectral theory. Quantum chaos. These aren't textbook abstractions -- they are the engine.
Pick the path that matches what you're looking for:
| Start with | What you'll find |
|---|---|
| DSC-1-Spectral-Unity | 28,160 testable Lehmer pair predictions + Monster Group moonshine peaks. Full dataset (CC-BY-4.0), reproduction notebooks, validation scripts. 6 confirmed, 1 falsified (reported). |
| On-chain papers (BSV) | Two paper series permanently anchored to the blockchain -- immutable, timestamped, publicly verifiable. |
| arXiv:2601.19500 | Cross-chain public key analysis -- 497,178 BTC-ETH key overlaps across 2.1B keys. |
| Start with | What you'll find |
|---|---|
| Daugherty_Engine_Benchmarks | Problem generators, solution validators, and claim verification scripts. MIT license. pip install -r requirements.txt and go. |
| What to verify | How to do it | Data source |
|---|---|---|
| 1M-spin Ising claims | Generate problems with scripts/generate_problems.py, solve with any solver, compare to our published timings |
Benchmarks repo -- data/benchmark_claims.json |
| Spectral predictions | Download data/lehmer_predictions.csv (28,160 rows), test against Riemann zeta zero databases (LMFDB, Odlyzko) |
DSC-1-Spectral-Unity -- data/ |
| Moonshine peaks | Check 15 peaks in data/moonshine_peaks.csv against Monster group representation dimensions |
Same repo -- scripts/validate_data.py runs 30+ integrity checks |
| Blockchain timestamps | Verify any claim via BSV transaction ID -- all results are publicly anchored | Transaction links in each repo |
All research data is CC-BY-4.0. All scripts are MIT. No account required. No API key needed for verification.
PHYSICS ENGINE APPLICATIONS
─────── ────── ────────────
Ising Model ──> DSC-1 Spectral ──> Optimization (1M Spins)
Hamiltonian Dynamics Engine Drug Discovery (BioPrime)
GUE / Random Matrices (Simulated Genomics (TopoGrammar)
Spectral Theory Bifurcation) Voice Synthesis (ORIGIN)
Agentic Commerce (ACO)
Physics-based GPU-accelerated optimization that solves problems conventional computers take 14 months to crack -- in under an hour.
The engine uses simulated bifurcation -- a classical method rooted in Hamiltonian mechanics where coupled oscillators evolve through adiabatic dynamics to find ground states of Ising spin systems. No quantum hardware required. The physics does the work.
| Capability | Performance |
|---|---|
| Ising Model Optimization | 1,000,000 variables in <40 minutes |
| 3-SAT Solving | 99.2% solution rate at phase transition |
| Scale | Full connectivity -- no embedding overhead |
| Verification | All claims blockchain-anchored (BSV) with public API |
The engine outscales specialized quantum systems (D-Wave: 5,627 qubits) while running on commodity NVIDIA GPUs at $1.57/hour.
Hardware: NVIDIA RTX A6000 (48GB VRAM)
| Aspect | Policy |
|---|---|
| Authentication | Post-quantum cryptography (Dilithium/ML-DSA) |
| Rate limits | Per-key limits; documented in each repo |
| Data retention | Inputs are not stored. Outputs are blockchain-anchored if requested. |
| Vulnerability reporting | See SECURITY.md in each public repo |
Professional-grade AI voice platform with multi-engine support.
- Real-time streaming text-to-speech
- Voice cloning and custom voice creation
- Voice mixing and full-duplex dialogue
- Multiple output formats
- Free to use
Physics-based molecular docking platform with Three-Phase Funnelization:
Phase 1: Neighborhood → Burial scoring + H-bond networks
Phase 2: Key-in-Lock → Pivot residue identification + steric gating
Phase 3: Gasket → Induced fit + desolvation penalty
| Capability | Performance |
|---|---|
| v3.2 → v4.0 Improvement | 45% accuracy increase across 10 validated targets |
| Docking Correlation | R² = 0.73 vs experimental binding affinities |
| Validated Targets | HIV Protease, EGFR, COVID-19 Mpro, Estrogen Receptor, + 6 more |
Grammar-aware analysis of three-dimensional genome structure.
| Capability | Performance |
|---|---|
| VUS Reclassification | 93% confidence (ACMG guidelines) |
| TAD Detection | F1 = 0.91 (outperforms HiCCUPS) |
| Throughput | 600 mutations/second, genome-to-drug pipeline in <5s |
The definitive framework for optimizing products and services for AI agent-driven purchasing decisions. 7-layer stack from Schema.org to Vector Optimization. Includes the Benchmark Tool for instant AI agent readiness analysis.
| Repository | Description | License | CI |
|---|---|---|---|
| DSC-1-Spectral-Unity | Monster Group--Riemann Zeta spectral connection -- 28,160 predictions | CC-BY-4.0 / MIT |  |
| Daugherty_Engine_Benchmarks | Verified benchmarks, problem generators, solution validators | CC-BY-4.0 / MIT |  |
Research papers permanently anchored to the BSV blockchain -- immutable, timestamped, publicly verifiable.
Bryan W. Daugherty, Gregory Ward, Shawn Ryan -- Origin Neural
Five-paper investigation establishing that stagnation dynamics serve as the dominant predictor of optimization outcomes, exhibiting a universal three-regime structure invariant across solvers and problem classes.
| Paper | Title |
|---|---|
| 1 | Stagnation as the Dominant Predictor of Failure Mode |
| 2 | The Synaptic Scalpel Phenomenon |
| 3 | Flow Math of Becoming |
| 4 | The Complexity Paradox in Solver Ensembles |
| 5 | Solver Complementarity at Phase Transitions |
Bryan W. Daugherty -- Origin Neural
Five interconnected paradigms reconceptualizing the universe as an emergent informational process. Introduces the Complexity-Time Correspondence (dS/dt ∝ dC/dt).
Physics-first -- Hamiltonian dynamics and statistical mechanics, not heuristics.
Rigor over hype -- Null results reported alongside confirmations.
Open verification -- Research datasets public. Claims blockchain-anchored. Reproduction scripts ship with every result.
Falsifiability -- One of our 7 DSC-1 predictions was not supported. That's published too.
Security -- Vulnerability reporting via SECURITY.md in each public repo. Post-quantum auth on all APIs. No security through obscurity.
OriginNeural.ai | DSC-1 Engine | BioPrime | TopoGrammar | ACO Academy
Origin Neural AI -- Research + Engineering
Physics-based computation. Real systems. Open science.