TYPE YOUR PROMPT.YOUR GPU RENDERS.INSTANTLY.

Natural language to WGSL shader. Real-time. On your GPU.

// CAPABILITIES

01

WebGPU Native

Direct GPU pipeline. No WebGL fallback, no abstraction layers between you and the silicon.

var<uniform> params: Params;

02

Real-time compilation

WGSL compiles in milliseconds. Every prompt, every iteration — instant visual feedback.

device.createShaderModule({ code });

03

Iterative prompting

Conversation-aware context. Refine your shader through natural dialogue without losing state.

messages.push({ role, content });

04

Dual AI models

Claude and GPT-4o switchable mid-session. Pick the model that thinks in your visual language.

model === "claude" ? anthropic : openai

// HOW IT WORKS

01

Type a prompt

Describe the visual effect you want in plain English. Be specific or be vague — the AI adapts to your intent.

02

AI writes WGSL

Claude or GPT-4o generates a complete WebGPU shader fragment from your description in seconds.

03

GPU renders live

Your GPU compiles and executes the shader at 60fps. Real-time visual feedback, every frame.

OPEN THE STUDIO_

Chrome 113+ or Edge 113+ required for WebGPU.