SPEC v7 · browser-native compute substrate
Compose, compile and train AI in a browser tab.
Scellis turns the browser into a compute substrate: draw a workflow, lower it to WebGPU, and train models locally — with your data never leaving the device.
[ 0 install · 0 deps · 0 uploads ]
Fig. 1 — compute field, rendered live on your GPU
The whole platform, on one screen
A small, stable engine — every capability delivered as content from the database.
| Parameter | Value | Note |
|---|---|---|
| Install | none | Open a tab and go — no CUDA, no Python runtime |
| Compute | WebGPU | Runs on the GPU already in your machine |
| Data | on-device | Egress is a declared, consent-gated effect |
| Reproduce | by URL | Content-addressed artifacts + fixed seeds |
| Extend | user = builtin | Author blocks, kernels, viewers, models |
| Train | on WebGPU | Explicit budgets, autodiff, oracle-checked numerics |
A drawn graph, lowered to GPU code
Your workflow descends a ladder of representations — and runs where you drew it.
; canonical IR
%x = load "input"
%w = param [768, 256]
%h = matmul %x %w
%y = relu %h// webgpu dispatch
@compute @workgroup_size(64)
fn main(id) {
// 256 invocations
}Fig. 2 — Lowering pipeline · Workflow → IR → WebGPU
Your data never leaves the device
Computation is local; egress is a declared, consent-gated effect — never the default.
Nothing crosses the boundary unless you grant a declared effect.
Fig. 3 — Data path, bounded to the device
Measured, not promised
Built-ins are just the starting set
Everything is content delivered from the database. User-authored entries are structurally identical to the built-ins.
| Item | Part | Source |
|---|---|---|
| 01 | blocks | seed |
| 02 | kernels (wasm + webgpu) | seed |
| 03 | your block | user/ |
| 04 | viewers | seed |
| 05 | your model | user/ |
| 06 | datasets | user/ |
Built-in and user rows are drawn identically — by design: one registry, one validation, one path.
Abstract
Scellis treats computation as a composable, inspectable, and shareable artifact. The core stays small and stable; capability grows, without bound, through content. WebGPU replaces CUDA, the database replaces the filesystem, and every run is reproducible — bit-for-bit — from a link.
epoch 8/8 · loss 0.04 · converged ✓
Open a tab and do science.
No install, no upload, no data center. Your first workflow is a click away.
End of spec · revision 7