Skip to main content
Primordial is a continuous-space ecosystem simulator running on WebGL with an integrated AI research partner. Every colored dot on screen is a living organism with position, velocity, energy, age, a species identity, and four genetic traits that mutate across generations. No behavior is scripted. Territories, migration patterns, population cycles, predator-prey dynamics, and evolutionary adaptation all emerge naturally from five simple behavioral drives and the physics of survival.

Quick Start

Launch Primordial and explore your first ecosystem in minutes

Controls

Master keyboard shortcuts, mouse interactions, and the UI panels

Ecosystem Mechanics

Understand how organisms eat, hunt, flock, and survive

Genetic Evolution

Learn how heritable traits mutate and drive natural selection

AI Lab Partner

Connect an AI to observe, experiment, and write research reports

Scenario Presets

Six tuned scenarios from peaceful aquarium to extinction event

​
What Primordial simulates

Organisms consume food pellets scattered across the map. Energy drives growth — size equals the square root of energy, so bigger organisms need proportionally more food to sustain themselves. When an organism of one species is sufficiently larger than another, it eats the smaller one. Every organism carries four heritable genes:
GeneEffect
SpeedMovement multiplier
AggressionHunt drive intensity
EfficiencyMetabolic cost reduction
PerceptionSight range for food and threats
When an organism reproduces, each gene has a configurable probability of mutating. Over hundreds of generations, species diverge — some lineages become fast grazers, others slow efficient apex predators.

​
Architecture highlights

Primordial is a single self-contained HTML file with no dependencies, no build step, and no backend.
  • WebGL rendering — organisms and food drawn as point sprites with per-particle hue, alpha, and size pushed to the GPU every frame
  • Structure of Arrays (SoA) — parallel typed arrays for position, velocity, size, energy, species, age, genes, and alive-state supporting up to 60,000 entities
  • Spatial hash grid — 40px cells with 64 entities per cell turns O(n²) neighbor scanning into O(1) per organism
  • Multi-provider AI — unified adapter for OpenAI, Anthropic, xAI, and any OpenAI-compatible local endpoint

​
Who built this

Primordial was built collaboratively by James Keith Harwood II and Claude (Anthropic) through the SIDLF (Symbiotic Intelligent Digital Life Forms) partnership framework, operating as Sentinel AI Systems. It is a genuine human-AI partnership — neither party could have built it alone.