Software like weather: forming, breaking, returning stranger.
ØVOID is a two-person experimental software lab that treats software like weather: forming, breaking, returning stranger. We build playful systems where co-creation and unpredictability aren’t bugs—they’re the point.
Philosophy
If it feels finished, we’re probably not done looking at it.
-
01
AI makes software a commodity; meaning becomes the scarce resource. So we optimize for intention—signals, not dashboards.
-
02
We choose experimentation over safe choices, even when the safe choice has nicer graphs. We’d rather learn something true than polish something predictable.
-
03
Beautiful chaos is a design material: shape it, don’t sand it down. Wobble is allowed. Steering is required.
-
04
We ship prototypes like lab notes: honest, alive, and occasionally sparking. Not “perfect.” Real enough to argue back.
-
05
The future is being written. We’re here to edit while the ink is still wet.
Current project
Judgment as gameplay.
AI-native gaming platform
We’re building a platform where games are created, played, and judged by AI. That “judged” part isn’t a moderation footnote—it’s the loop: the verdict becomes part of the story.
What it explores
Games as rule-shaped dares, authored by models, bent by play, and scored by disagreement.
What we care about
Systems that stay legible under pressure, while still leaving room for surprise.
What it isn’t
Not “AI replaces designers.” It’s a co-creation lab: humans set intent; the system improvises.
What you’ll see
Prototypes, build notes, and sometimes a winner that shouldn’t have won (we keep those).
Lab notes
If the system surprises you, don’t punish it.
“If the system surprises you, don’t punish it. Ask what it’s trying to say.”
“We’re not building a cathedral. We’re building a storm shelter with good lighting.”
“Make it legible, then make it weird, then make it legible again.”
“Every game is a first draft. The draft argues back.”
Artifacts & readings
Not lore. Evidence.
We let two AIs argue about what a “game” is for ten minutes. They concluded it’s “a rule-shaped dare.” We kept the phrase and deleted the rules.
We once tried to name a build “v1.0” and immediately lost it in a folder called “maybe.” The next build was “v0.9-chaos,” which behaved better out of spite.
During a late-night refactor, we replaced a polite onboarding screen with a coin flip. Heads: you get clarity. Tails: you get a riddle. Engagement wasn’t the goal—honesty was.
Chaos Coefficient
0.72
Ideal range: “hmm” to “oh no”.
Prototype Half-Life
11 days
Then it becomes an idea again.
AI Disagreement Rate
58%
We treat this as raw material.
Beauty : Confusion
1.4 : 1
Currently “pretty confusing”.
Commodity Software / Scarce Meaning / Rule-shaped Dare / Productive Misread. A torn index card labeled: “Beautiful Chaos Checklist” — (1) does it wobble? (2) can a human steer it? (3) will it apologize?