// culture · households · relationships · lore · minds · closure

NOTOYAI

These minds are not toys
Procedural Intelligence Generation
The Problem

Every interactive world
is frozen.

Procedural tools pour out continents overnight — terrain, cities, forests, props. But the people inside them are shells. NPCs reset the moment you look away. Crowds loop. Nothing remembers you.

A decade of prompt tricks and state machines hasn't fixed it. Because the gap was never content. The gap is cognition.

The Idea

PCG built the world.
Now PIG builds the minds.

Procedural Content Generation turned world-building into a graph problem: data in, nodes, content out. Procedural Intelligence Generation does the same for the cognitive world.

Point a graph at your level. Set a seed. Generate — and get a population that holds together.

The Substrate

First, we gave them minds.

Not one model. Not a prompt trick. A system: memory that shapes perception, identities that hold under pressure, fear, fatigue, trust, grudges.

Minds that carry their history with them — frame by frame, day by day. Not simulated. Lived.

Live Signal
Agent #4217 — Tick 3,841
"She checks the door again. Not because she heard something — because she can't stop thinking about the last time she didn't."

No API call. No prompt saying "act scared." The fear is real — stored, persistent, inherited from three hours ago.

Why This Is Hard

A mind is not a model. A mind is what sits around a model — the structure that gives it continuity, a body, a social world, a history it cannot escape.

Hundreds of interlocking systems. Years of work mapping human cognition into machine-readable form. The kind of thing you can't reverse-engineer from an output, and you can't prompt-engineer your way around.

// depth is the moat
The breakthrough isn't the models.
It's what sits between them.
The Pipeline

From terrain
to society.

Step one. The engine already generates the physical world.

01 — Unreal PCG
It already builds the world.
Terrain, buildings, roads, forests, props — poured out by the same procedural graphs studios already run. PIG reads those generated structures as its input. Nothing new to author by hand.
Terrain Buildings Roads Forests Props
The Pipeline

From terrain
to society.

Step two. PIG generates the cognitive world.

02 — NotoyAI PIG
A graph from world data to living society.
Cultures, households, professions, relationships, lore, minds — authored the way you author biomes, deterministic from a single seed. Closure is validated before export: no orphans, no dangling relationships, no loose ends.
Generate Culture Households Relationships Lore Validate Closure
The Pipeline

From terrain
to society.

Step three. The runtime brings them to life.

03 — Runtime → Engine
Hundreds of minds, back into the engine.
The NotoyAI runtime simulates the population frame by frame, then hands it back as actors, spawn points, bindings and animation intent. One sentence in — a whole society out.
Persistent Frame by frame Runtime-ready export
The Economics

Runs where a chatbot can't.

100s
Minds per world
1
Consumer GPU
$0
Per-token cost

Not cheaper AI.
A different cost model entirely.

The age of living worlds
starts here.

Building a game, an interactive film, a robotics platform? Early access is open for studios and partners.

We read every message. Please include your studio and what you're building.

First game in development on NotoyAI minds

Ruhnu JooksikIgikelts Studio

RUHNU ISLAND, 1944–1950 · UNREAL ENGINE 5 · STEAM NEXT FEST OCT 2026

Private access — studios & partners
Scroll to awaken