ØVOID
ØVOID experimental software lab (2 humans)
intentionally unfinished systems

Beautiful chaos, practiced on purpose.

ØVOID is a two-person experimental software lab practicing beautiful chaos on purpose. Our flagship is an AI-native gaming platform where games are createdNot “generated content.” More like: rules appear, get tested, get revised—by the system itself., played, and judged by AI—every game is a first draft, and the future is being written.

Focus: AI-native play
Method: experiments & iteration
Bias: taste over safety

ØVOID / Field report: AI-native play

v0.Ø — legible on purpose

01 Current project

An AI-native gaming platform where games are created, played, and judged by AI.

We’re building a platform that treats games as living drafts: the system proposes mechanics, tests them through play, and evaluates outcomes with AI judges. The point isn’t to perfectly automate fun—it’s to make a space where rules can evolve without losing coherence.

Created by AI

Mechanics, prompts, and objectives emerge as prototypesIf it feels finished, we probably shipped too late.—then get revised.

Played by AI

Agents explore the game like QA with imagination: they break rules, then learn them.

Judged by AI

Scoring rubrics don’t stay still. Judges invent categories like “vibe consistency” when needed.

Designed by ØVOID

We set constraints, instrumentation, and taste. We keep the chaos useful.

Philosophy note: Co-creation beats control: we set the stage, the system improvises.

02 Philosophy

AI makes software cheap to produce. That changes what matters. Our differentiator is taste, audacity, and iteration speed— plus a willingness to ship experiments that still have pencil marks.

  • AI makes software a commodity; taste, audacity, and iteration become the differentiator.
  • We choose experiments over safe choices, even when the safer choice has better kerning.
  • Beautiful chaos is a design material, not a bug report.
  • The future is being written—sometimes in pencil, sometimes in glitch. We keep both.
This page may contradict itself. That’s data.

03 Lab notes (selected)

From a late-night test

A late-night test produced a game whose only instruction was “betray the tutorial.” The AI then graded everyone on dramatic timing. We added a sticky note: “teach less, invite more.”

From the scoring rubric

We built a scoring rubric for AI judges, then watched the judges disagree politely and invent new categories like “vibe consistency” and “plot gravity.” The rubric is still in the repo, mostly as a warning.

Figure A (label only) Sticky notes, code snippets, coffee rings. The lab, in its natural state.
Figure B (label only) A gavel made of pixels, mid-swing. Judgment, but with compression artifacts.
Figure C (label only) First-draft arcade cabinet. Rules taped on. The tape becomes part of the UI.
Figure D (label only) Star chart of half-finished UI fragments. Constellations of “almost.”
Want the calm version? We didn’t build one.