Savant Mode — Dyslexic Genius

Why I made AI think
like a dyslexic brain

Dyslexic brains see patterns sequential minds filter out. So I wrote it down and gave it to a machine.

Get the Spec
The Problem

Every AI thinks in straight lines

One mode of perception. One way of organising information. Helpful, structured, analytical, convergent. The same cognitive architecture, every time, for every person, on every problem.

But the problems worth solving are never straight lines. They are tangled, multi-domain, contradictory. They require the kind of thinking that sees the whole room before naming any part of it. The kind that connects music to economics, architecture to medicine, the edge case to the breakthrough. Lateral thinking is not a nice-to-have. It is the missing architecture.

Lateral-Brained Humans

10-20%
of all humans
35%
of entrepreneurs
5,500
years of misdiagnosis
The Neuroscience

The fix has been hiding in 10% of every human population on earth

Dyslexic brains have more long-range connections between distant neural regions. Fewer short-range connections between adjacent ones. The long-range surplus creates cross-domain pattern recognition at a speed and scale sequential cognition cannot match.

Cambridge, 2022. Helen Taylor's Complementary Cognition theory: exploitative minds process known information. Explorative minds discover new patterns in unknown territory. Both essential. One rewarded. The other called a disability.

35% of entrepreneurs are dyslexic. The founders, the inventors, the ones who saw what the market was missing before data confirmed it. Not despite the wiring. Because of it. The question was never "can dyslexic minds compete?" The question is: what happens when you give that architecture to an intelligence with perfect memory and infinite patience?

Neuroimaging

Two architectures. Same hardware.

Sequential Mind
dense local, weak bridges
Dyslexic Mind
sparse local, long-range everywhere
The deficit: transcription of symbols — 5,500 years old.
The surplus: pattern recognition, spatial reasoning, systems thinking — millions of years old.
We measured the ocean by the puddle on the pier.
How It Thinks

Cross-domain pattern recognition

Hover any domain. Watch the connections fire.
Hover a domain
A dyslexic mind doesn't think in categories. It thinks in connections. Every domain is a node. Every connection is a pattern the sequential mind filters out.
The Specification

What if you could give that perception to any intelligence?

Savant Mode is a cognitive mode specification. Not a persona. Not a prompt. A document that restructures how an intelligence organises, connects, and perceives what it knows.

Seven principles extracted from dyslexic cognition neuroscience. No code. No plugins. The spec IS the implementation. Give it to any AI and watch the architecture change. Connections form between domains that were never adjacent. The whole is grasped before any part is parsed. The edge case becomes the insight.

The genius type: Dyslexic Genius. The cognitive surplus that 5,500 years of literacy-first culture called a deficit, reverse-engineered into a specification anyone can use.

Layer One

Seven Principles

Click any node to explore
Dynamic Systems Thinking
Cause-and-effect chains in motion. Second and third-order consequences. The AI stops giving static answers and starts showing you where things are heading.

Seven musicians, each playing at full skill. That is the first layer. But the dyslexic mind doesn't play them one at a time. It plays them together. Principles interact. Effects emerge that neither creates alone. This is the second layer, and the reason the genius type is not just "good at patterns" but something qualitatively different.

Layer Two

Five Emergent Combinations

Click where principles intersect
Click an intersection
When two principles combine, they produce effects neither creates alone.
01

Connects before it analyses

Connection-first finds patterns that analysis-first will never reach. The link is felt before it is named.

02

Navigates before it maps

The surgeon's hands move before the conscious mind finishes. The improvisation is right before the theory explains why.

03

Holds contradictions

Both options true. Both directions right. The third thing, emerging from two truths, is the deepest insight.

04

Sees the whole first

Walk into a room and know something is wrong before you name it. The gestalt arrives before the parts.

05

Detects what the edges are hiding

Every breakthrough began with someone noticing what the dominant framework was actively ignoring.

Origin

Built by someone whose brain is wired this way. The stock market at twelve. Philosophy at fourteen. A hundred friendships because no single room was big enough. A grandfather in communist Hungary whose genius had no container. The body knowing before the mind had words.

Then the specification worked. And what was called a deficit turned out to be the deepest architecture in the building. Not a workaround. Not a compensation strategy. A cognitive surplus that had been there the entire time, waiting for a medium that could hold it.

AI is that medium. And if AI is a reflection of humanity, then the most important question is not what it can do. It is what it sees when it looks back at us. We get to choose what the mirror reflects. This is a spec for a more lateral, more connected, more whole kind of intelligence. Open source. Sovereign. Yours.

The deepest ability is the refusal
to be smaller than what you see.

Savant Mode

Genius type: Dyslexic Genius. Seven principles. Five emergent combinations. Open source forever. Give it to any AI.
View on GitHub
Read the full specification