NINAD SREEBACK TO NINAD.APP ↗

THE VERSION LOG

Ten versions of this site.
Eighty-one tries.

I rebuilt my portfolio with AI, over and over, to find out what it's good at and what it still needs a person for.

Here's every version that mattered: what I tried, what I learned, and the old ones you can still walk around in. Brands are swapped for placeholders, everything else is as it was.

VERSION 1

Experimenting.

I TRIED
A green pixel field, six projects and a hidden ping pong game.
I LEARNED
AI is fast at building. It has no idea what matters.

CATTO: AI was fast at the building, clueless about what mattered.

Walk around V1 →

VERSION 2

Doubling down.

I TRIED
More of everything: talking people, night mode, a patient fisherman.
I LEARNED
Doubling down only works if you double the right thing.

CATTO: More of everything. Even the fish were confused.

Walk around V2 →

VERSION 3

Trying on a personality.

I TRIED
I swapped pixels for a dark vinyl shelf. I don't even listen to vinyl.
I LEARNED
AI can try on ten styles in an afternoon. It can't tell you which one is yours.

CATTO: A whole shelf of records. He doesn't listen to vinyl.

Walk around V3 →

VERSION 4

Learning to say no.

I TRIED
Back to pixels. Fewer jokes, and real numbers instead of “∞ files”.
I LEARNED
AI says yes to everything. Taste is knowing what to throw away.

CATTO: He said no a lot. I respect that.

Walk around V4 →

VERSION 5

Building a tower.

I TRIED
A tower you climb floor by floor, with a full case study on every floor.
I LEARNED
I designed for the visitor I wanted, not the one who actually shows up.

CATTO: Nine floors of reading. I napped on floor two.

Walk around V5 →

VERSION 6

Shifting gears.

I TRIED
I studied other people's work, then flipped the tower upside down.
I LEARNED
Good references make better prompts. The taste goes in before the prompt does.

CATTO: He flipped the tower. I landed on my feet.

Walk around V6 →

VERSION 7

Making it easy.

I TRIED
Every floor answers three questions: how long, what I owned, what I solved.
I LEARNED
AI writes long. Clarity means asking for less.

CATTO: Clear. Quick. Useful. Bit boring. Needed a cat.

VERSION 8

Easter eggs.

I TRIED
A cat by the lamp, a moon that cracks and a rod you can actually fish with.
I LEARNED
Delight is a choice. AI doesn't wonder what would make you smile.

CATTO: I'm in it now. Tap me thirteen times. I dare you.

VERSION 9

Getting side-tracked.

I TRIED
Three small games under the sea, and a track I played myself.
I LEARNED
The side-tracks taught me the most about building with AI.

CATTO: Three games, one song. The portfolio waited.

Go to the arcade →

VERSION 10

It's live.

I TRIED
ninad.app, with SEO and a sitemap. Google can find it now.
I LEARNED
AI built most of it. Deciding what was good stayed my job.

CATTO: The sun's up. There's a dock with your name on it.

81OUTPUTS, V1 TO LIVE

What I brought to every one of them, that AI couldn't

AI did most of the making. These are the parts that decided whether any of it was good.

  1. Taste. What looks right, not just what looks impressive.
  2. Saying no. To vinyl, to extra features, to good ideas that weren't mine.
  3. References. Knowing whose work to learn from.
  4. The truth. No made-up numbers, no claims I can't back up.
  5. The details. Like a cat that says “OK. STOP.” on the thirteenth tap.

THERE'S A DOCK WITH YOUR NAME ON IT.

GO TO NINAD.APP →
V1 · 1/6