Nothing here is final. That’s the point.
We build software that argues back—politely, usually.

Beautiful chaos, with intent.

ØVOID is a two-person experimental software lab exploring what happens when AI makes software a commodity and taste becomes the differentiator. We build in beautiful chaos—because the future is being written, not polished.

Mission / philosophy

taste > tooling

We’re ØVOID: two people, experimenting at the edge where AI makes building cheap and attention gets expensive. Our work leans into beautiful chaos—not as aesthetic noise, but as a deliberate way to find new software shapes.

Our default posture is “first draft.” We ship something real, then we argue with it until it reveals what it wants to become. Co-creation beats control: the system gets a vote, and so do you.

Micro-policy: Predictability Budget: 5% (spent immediately).

  • We don’t tame chaos; we give it a UI and a place to misbehave.
  • Co-creation beats control: the system gets a vote, and so do you.
  • The future is being written. We keep the pen slightly uncapped.
  • Design note: If it’s predictable, it’s probably already obsolete.
    lab posture: exploratory deliverable: learnings quality bar: intentional

Current project

AI-native games

Meet our AI-native gaming platform

A platform where games are created, played, and judged by AI. Think: generative gamecraft + AI opponents + an AI judge that can be “impishly fair.”

  • Creation: games emerge from prompts, constraints, and taste—then get iterated fast.
  • Play: the system plays too, exploring edge cases humans avoid (or invent).
  • Judgment: scoring can be numbers, but also critique—sometimes a poem instead of a metric.

We’re interested in what “game” becomes when rules can be negotiated, authored, and enforced by agents. We treat the judge as part referee, part editor, part mischievous museum label.

Chaos Coefficient

0.73 (measured in surprised silence).

Draft Velocity

11 first-drafts per afternoon (±3 daydreams).

Judgment Mood

“impishly fair” (calibrated by eyebrow raise).

Novelty Spillover

2.4 accidental features per intentional feature.

Lab notes (selected)

anecdata
  • At 2:14 a.m., we watched the platform generate a tutorial level that politely taught you to lose. It called the lesson “grace.” We kept it, then added a “rematch” button out of spite.
    type: tutorialtone: politeresult: unexpectedly formative
  • A whiteboard diagram titled “simple onboarding” slowly accumulated arrows until it resembled a weather system. We renamed it “Onboarding, but as climate.”
    type: diagramsymptom: arrow bloomfix: rename honestly
  • During a test, the AI judge declared a game “elegant” and then docked points for being “too reasonable.” We took that as product feedback—and personal advice.
    type: judgingverdict: elegantpenalty: too reasonable
“If AI makes software cheap, attention gets expensive. So we spend attention like it’s currency: carefully, strangely, and with receipts.”
Lab Note #07
Read the latest lab note →
You found an edge case. Congratulations.

Artifacts (from the bench)

evidence of work
Scanned sticky note artifact

“Make the button feel like a dare.”

moodboard.json file

["spark","fog","neon","oops"]

Lab stamp print

A hexagon that says FIRST DRAFT APPROVED in crooked type.

Screenshot caption note

“Scoreboard, where the AI judge wrote a poem instead of a number.”

Audio transcript snippet

“click… click… laughter… ‘why is the floor negotiable?’”

Paper prototype half-torn

“Settings,” with only one toggle: More weird: OFF / ON / PLEASE