• 01 The Problem: Why AI Has Hit a Wall 3
• 02 From g to c: A New Theory of Intelligence 5
• 03 The Co-Evolution Model 7
• 04 Research Foundation & Methodology 9
• 05 Business Model & Competitive Advantage 11

The Scaling Crisis

For the past decade, AI progress followed a simple formula: more data, more compute, better results. This era delivered transformative capabilities—from language models to image generation. But the curve is bending.

The fundamental issue isn't computational. It's that current AI optimizes within a narrow framework based on Spearman's general intelligence (g) from 1904—a concept that captures only 40-50% of human cognitive variance.

The Data Everyone Is Missing

Most AI training data is "result-oriented"—final outputs, correct answers. What's absent is Process Data: how thought forms, reasoning traces, decision branches. This is cognitive "dark matter."

Industry Leaders Sound the Alarm

We're not alone in recognizing these limitations. The most influential voices in AI are calling for fundamental change:

This convergence of expert opinion points to a clear conclusion: the next frontier of AI isn't about bigger models. It's about understanding what makes human cognition irreplaceable.

Introducing Complementary Intelligence (c)

Spearman's general intelligence factor (g), proposed in 1904, shaped a century of cognitive science. It captures correlation across cognitive tests—people good at one mental task tend to be good at others.

The problem: g explains only 40-50% of cognitive variance. The remaining 50-60% represents dimensions where humans maintain qualitative advantages over AI.

What Makes c Different

We operationalize complementary intelligence (c) as cognitive primitives that are:

• Under-utilized in current AI training and evaluation
• Measurable through neuroscience-grounded methodologies
• Trainable in humans, generating high-fidelity Process Data
• Crucial for human-AI complementarity in real-world tasks

Current AI excels at pattern recognition and statistical inference—capabilities well-captured by g. But they lack embodied experience, social intuition, and genuine understanding that characterize human cognition.

From Competition to Co-Evolution

The prevailing discourse follows a competitive frame: as machines get smarter, humans become less relevant. This zero-sum thinking misreads history and misses the biggest opportunity of our time.

The Infinite Loop

Exhibit 5

The Co-Evolution Engine: A Continuous Cycle

As AI masters Level N, we discover the next frontier and upskill humans to Level N+1.

Process Data: The Missing Link

Most AI training data captures results—final outputs. What's missing is Process Data: how thought forms, reasoning traces, decision branches, uncertainty calibration.

Our human training programs generate this Process Data at scale. This dataset doesn't exist anywhere else—and it's key to training AI on cognitive dimensions that matter.

People resist tools that replace them—but embrace tools that upgrade them. When employees see AI as amplifying their capabilities rather than threatening livelihoods, adoption accelerates.

The co-evolution model isn't wishful thinking. It's how complex systems naturally evolve. Predators and prey, immune systems and pathogens—mutual pressure creates mutual advancement.

Behavioral Wind Tunnels

Like aerospace wind tunnels that test aircraft under controlled conditions, Behavioral Wind Tunnels create controlled, ecologically valid environments for mapping intelligence in real tasks.

Trait-State Decomposition

Our methodology separates stable individual differences (traits) from context-dependent variations (states). This allows us to:

• Identify which capabilities are trainable vs. fixed
• Measure intervention effectiveness with precision
• Predict individual learning trajectories
• Optimize training programs for maximum cognitive gain

Addressing Scientific Critiques

We believe transparency builds trust. Here are hard questions we've faced—and our responses:

Two Markets, One Flywheel

Our business model serves two distinct markets that reinforce each other, creating a flywheel effect that compounds over time.

Competitive Moat

Our defensibility comes from four interlocking advantages:

1. Source Code Advantage: We understand brain architecture, not just AI engineering. We define ground truth before tech companies know what to look for.
2. Proprietary Longitudinal Process Data: How thought forms, tracked over time. Generated through training programs; doesn't exist elsewhere.
3. Unique Methodology: Behavioral Wind Tunnels + trait-state decomposition + AI benchmarking—a vertically integrated pipeline.
4. Continuous Discovery: c is a moving target. We discover new primitives faster than competitors can copy.

Addressing Business Model Critiques

The Long View

The question isn't whether AI will transform work—it's whether that transformation will be zero-sum or positive-sum. We're building for a future where AI advancement drives human development, human development drives AI advancement, and the cycle continues.

This isn't utopian speculation. It's how complex systems evolve. Predators and prey, immune systems and pathogens. Mutual pressure creates mutual advancement.