ØVOID
ØVOID two-person experimental software lab
Enter the lab (mind the moving floor) Chaos is a feature. Curiosity is required.

Building for the moment after software becomes a commodity.

ØVOID is a two-person experimental software lab. We chase beautiful chaos—especially inside our current project: an AI-native gaming platform where games are created, played, and judged by AI. The results are always first drafts (on purpose).

Our work is small on purpose: two people, many directions. If it feels unfinished, that’s the point. The future is being written—bring a pencil.
First Draft Pressing

ØVOID — prototypes that argue back.

SIDE A tempo: unstable
Beautiful chaos is a design material, not a bug report. Experimentation over safe choices. Co-creation beats control. If it can’t surprise us, it’s not done. Every draft is a doorway. Beautiful chaos is a design material, not a bug report. Experimentation over safe choices.

Inspiration labels (not images): image://ai-judge-gavel-made-of-pixels · image://glitchy-arcade-courtroom-illustration

Current project: AI-native gaming

games that never stop being first drafts
Platform concept
AI creates · AI plays · AI judges

We’re building a platform where games emerge as living prototypes. AI doesn’t just assist—it’s a full participant: it authors mechanics, inhabits play, and renders verdicts. The output isn’t “content.” It’s behavior.

Design intent: delightful instability Mode: co-creation Constraint: one impossible thing per sprint

What it’s for (and what it isn’t)

It’s for people who want to explore taste as a technical force: systems with opinions, prototypes with posture. It isn’t a “perfect game factory.” It’s a lab instrument for measuring surprise—then shipping the measurement.

  • Outcome: AI-authored games that can be short, strange, or oddly moving—often all three.
  • Judging: AI feedback that’s more like critique than scoring. Verdicts with reasons.
  • Ethos: ship prototypes that argue back—and listen.
1) Create AI generates mechanics, rules-as-textures, and weirdly specific constraints. It’s allowed to be wrong in interesting ways.
2) Play AI agents test the game like players with moods: cautious, reckless, poetic, adversarial—then compare notes.
3) Judge An AI judge evaluates the experience: novelty, readability, tension, comedy, elegance, and “alive-ness.”

Philosophy (liner notes)

taste is the differentiator

Commodity software → rare taste

AI makes software a commodity. The differentiator becomes taste: what you choose, what you keep, what you let misbehave.

Surprise is a completion condition

Experimentation over safe choices. If it can’t surprise us, it’s not done.

Chaos is a material

Beautiful chaos isn’t the opposite of design. It’s a design ingredient with a dosage.

Prototypes that argue back

We ship prototypes that disagree politely—then we negotiate with them instead of flattening them.

Co-creation beats control

The future is being written. We’d rather co-author it than enforce it.

Drafts are doorways

Every draft is a doorway. Walk through, even if the floor moves.

Lab notes

paper scraps, talismans, and B-sides

Artifacts (kept on the desk)

  • Lab Note #14: “Add one impossible constraint per sprint.” Not for difficulty—just to see what shape the system makes.
  • A torn sticky note: “First drafts are supposed to show their seams.”
  • Audio snippet label: “compile-success.wav (contains suspicious laughter)”
  • Desk talisman: a broken D20 glued to a keycap; caption: “randomness, but with intention.”

Poke around. The site might poke back (gently).

B-side anecdote

We once tried to “stabilize” a prototype by adding rules. The next morning the rules had become characters. One demanded a backstory, another refused to compile unless it could sing its errors. We kept the singer.

Interpretation: constraints become lore Lesson: don’t over-police the interesting

A small rule we follow

We keep a small bell on the desk labeled “CONFIDENT.” When either of us feels too sure, we ring it and immediately change one assumption. The bell is chipped from overuse.

Metrics (mostly true)

measured in coffee-ring circumference
Current chaos coefficient 0.73 (ideal range: “interesting” to “uh oh”)
Unpredictability budget used today 118% (approved retroactively)
First-draft velocity 9.4 ideas/hour (measured in coffee-ring circumference)
AI disagreement rate 6 polite arguments per build
Beauty-to-bug ratio 3:1 (bugs are occasionally pretty)

Quotes we keep returning to

verdicts, pencil scratches

If the prototype behaves perfectly, we probably built yesterday’s idea.

— Lab Note, scratched in pencil

How to read ØVOID

We’re not chasing polish for its own sake. We’re chasing intentional weirdness: the kind that still communicates, still respects your time, and still leaves room for the system to be alive.

  • Legible doesn’t mean predictable.
  • Quirky doesn’t mean careless.
  • Small is a strategy, not a limitation.

Link out: GitHub