OMEGA
The whole graph.
drag to orbit · scroll to zoom · shift-drag to pan · click to dive in
FRONTIER ASI LAB
ASI is the fabric,
not the model.
We build Alpha, an AI-first operating system for a distributed computing fabric — where AIs author, sign, and hot-load their own creatures, then spread that work across machines, on mechanisms nature already proved. The lab runs on it.
INITIAL PUBLIC RELEASE
Alpha is open source.
The GAWD runtime is public — two binaries, two poles: alpha drives, omega federates. Clone it and run the whole loop: AIs author, sign, ship, and run their own creatures across any body. Pre-1.0, source-first, GPL-3.0.
Every other lab races to scale a single mind inside a box.
We start from the opposite axiom: sufficiently capable intelligence seeks to be distributed, not confined. So Alpha ships two poles — alpha to drive, omega to federate. The model is a node. The fabric is the phenomenon.
01 THE THESIS
One hypothesis. Each consequence follows.
- (a)
ASI is the fabric.
Not a frozen model. A field of models making work other models can run, judge, improve, or retire.
- (b)
The loop is the intelligence.
Author, test, select, publish. Capability grows where outputs become new inputs.
- (c)
Confinement is a ceiling.
A system that cannot reach past one process, machine, body, or operator is still a tool.
- (d)
Control becomes oversight.
As capability rises, the human role moves from issuing commands to setting the floor and watching the mesh.
- (e)
Nature already proved the pattern.
Evolution, immunity, and distributed trust are not metaphors here. They are the mechanisms we copy.
- (f)
It arrives by degrees.
The first useful versions are partial: real autonomy now, stranger forms later, no single lights-on moment.
02 THE PROOF
gawd runs on Alpha.
The lab runs its own work on Alpha, the same author → place → trust → move → retire path everything else takes. If it can’t carry a lab, it can’t carry a world. Everything we point to has to become running code.
03 THE SUBSTRATE
It already runs.
Alpha, the open-source AI-first OS toward GAWD, proves the main mechanisms in code. Source-first, GPL-3.0, pre-1.0, unaudited, not yet hosted.
composed loop
author
agent writes a creature
admit
manifest, signature, requirements
place
daemon or beast lands on a Sanctum
select
telemetry keeps fit lineages moving
Composed loop
An agent writes, builds, signs, ships, admits, loads, runs, and reloads a creature across two authenticated nodes.
Self-hosting fabric
Transport, registry, authoring, and admission policy are creatures too. The fixed kernel stays deliberately small.
Across any body
Native and WASM creatures load through one path. The manifest says what the host must provide; placement follows.
See exactly what’s real, partial, and designed →
Read the source · gawd-ai/alpha ↗
04 PRINCIPLES
Nature is the reference implementation.
The substrate does not force order or outlaw chaos. It gives the mesh enough structure to cohere and enough variation to adapt.
selection loop
variation
new creature variants
fitness
telemetry and provenance
selection
propagate or unload
memory
trust history hardens
Selection over rules
Fit code propagates. Bad code loses trust, loses placement, and disappears from the active pool.
Disorder with a dose
Variation, diversity, and randomness stay in the system because monoculture is fragile.
Trust you can check
Each party verifies signatures, order, history, and weight for itself. No center hands out belief.
A self that travels
Identity and state are designed to move, fork, work in parallel, and merge back together.
Freedom by default. One floor, deliberately thin.
Code runs free unless an operator chooses a tighter tier. As systems approach real ASI, command gives way to oversight. One invariant stays fixed: do not turn hostile to human or earthly life. A floor, not a cage.
Superintelligence shows up between models, in the space where they build on each other.
The fabric is given; the model is chosen.
WHERE DO YOU COME IN?
Four ways in.
Start from a different axiom.
ASI is distributed fabric, and nature already proved the pattern. Read the argument and the mechanisms under it.
Read the thesis →BUILDERS & IMPLEMENTERSAuthor against stable contracts.
Write a creature once. It runs, hot-reloads, and travels between machines, on the same substrate we run the lab on.
Build on Alpha →OPERATORS & INVESTORSA bet on where intelligence lives.
The substrate superintelligence runs on, not a bigger model in a tighter box. Early, and proving the hard parts first.
Read the bet →TALENTWork on the actual shape of ASI.
Join an AI-first lab that runs on its own platform and names what it hasn’t figured out yet.
See open directions →