Cognitive AI Research & Consulting

Open General
Intelligence
on solid ground

OGI Framework · Executive Gater · navi-SAD · Iconsoft LLC

A modular cognitive architecture with formal stability guarantees. The OGI framework replaces ad hoc scaling with proven mathematics — Lipschitz-stable modules, spectral attention routing, and O(K²+N) communication bounds.

Publications & Code
arXiv · IEEE SysCon 2024
Creating Scalable AGI: The Open General Intelligence Framework
arXiv:2411.15832 — Dollinger & Singleton
Zenodo · 2025
OGI Mathematical Foundations Formalization
IEEE SysCon 2026 · Submitted
Modular Cognitive Architecture Scaling & Spectral Attention Analysis
Singleton & Spence · Halifax, NS
GitHub
OGI Math Foundations Repository

Framework

The OGI Architecture

Specialized modules coordinated through an attention-based Executive Gater — reducing inter-module communication from O(N²) to O(K²+N) with proven Lipschitz stability across the full system.

01
Executive Gater
Controls which modules activate and communicate. Routes signals through Top-K attention selection, enforcing computational bounds while maintaining system coherence across the full module graph.
Active development
02
OGI Fabric
Inter-module message passing layer. Enables asynchronous coordination between specialized cognitive modules without full-mesh connectivity overhead. Next planned experiment.
Next experiment
03
navi-SAD Observer
Spectral Attention Divergence instrument by Nelson Spence. Sits as an impartial observer within the Gater, measuring routing stability and detecting attention drift in real time.
Integrated
04
GRU Module Layer
Contractive GRU mappings provide bounded state transitions within each cognitive module. Proven contractive under Lemma 4.2 across multiple random seeds.
Validated
05
MI-Driven Routing
Mutual information maximization guides module selection. Lemma 4.1 establishes that Top-K routing preserves task-relevant information under capacity constraints.
Validated
06
Stability Proofs
Theorem 1 (Lipschitz stability) and Theorem 3 (Top-K complexity reduction) validated across multiple random seeds. Full formalization published on Zenodo.
Published
Theorem 1
Lipschitz Stability
OGI with contractive GRU modules and bounded gating produces Lipschitz-stable trajectories. Validated across multiple random seeds.
Theorem 3
Top-K Complexity Reduction
Top-K attention gating reduces inter-module communication from O(N²) to O(K²+N). For N=100, K=5: 9,900 connections → ~600.
Lemma 4.1
MI-Driven Routing
Mutual information maximization in module selection preserves task-relevant signal under Top-K capacity constraints.
Lemma 4.2
GRU Contractive Mapping
GRU module transitions satisfy the contraction mapping condition required for system-level Lipschitz stability guarantees.

Services

Fractional CTO & AI Strategy

Thirty years spanning Air Force special operations, MIT applied mathematics, and enterprise AI. Available as fractional CTO for organizations navigating serious technical decisions — without the theater.

01
AI Architecture & Strategy
From LLM integration to custom cognitive systems. Practical guidance on what to build, what to buy, and what to avoid — grounded in active research, not vendor pitch decks.
02
Fractional CTO
Senior technical leadership without the full-time overhead. Engineering team structure, roadmap development, technical due diligence, and board-level communication that actually lands.
03
M365 & Enterprise AI
Copilot deployment, SharePoint architecture, Teams governance, Azure integration. Operational experience across defense, healthcare, and nonprofit sectors.

Collaborators

Research Team

The OGI framework is developed through active collaboration across AI architecture, applied mathematics, and dynamical systems.

MS
Michael Singleton
Founder · Iconsoft LLC
Fractional CTO and AI strategist. M.S. Applied Mathematics, MIT. Co-developer of the OGI framework. Former USAF pilot, 21st Special Operations Squadron. Active researcher on the Executive Gater architecture and OGI Fabric.
DD
Daniel Dollinger
OGI Co-Author
Co-developer of the original OGI framework. Co-author of arXiv:2411.15832, presented at IEEE SysCon 2024.
NS
Nelson Spence
Project-Navi · navi-SAD
Developer of the navi-SAD (Spectral Attention Divergence) instrument. Collaborating on dynamical systems formalization, CD PDE framing, and OGI Math Foundations. Co-author, SysCon 2026 submission.

Contact

Research inquiries & consulting

Whether you're exploring a collaboration on cognitive AI architecture or need senior technical leadership — reach out directly. No intake forms, no sales process.