About.

Honestly, this started as me just playing around with Claude Code. I wanted to see what it could build. I had no brief, no plan, no audience in mind. I just thought: what if I turned my own data into a store?

So I catalogued everything I own. Clothes, tech, kitchen stuff, books, gym gear -- everything. What it is, how long I have had it, how often I use it, whether I would buy it again. Then I started feeding in other data. My Spotify listening history. My Strava runs. My Claude conversations -- every product I have ever researched, compared, or rejected. Browser bookmarks. Email receipts.

Every ad platform collects your data to sell you things. This inverts that model. Same engine, opposite beneficiary.

The idea was to build an algorithm that understands one person -- me -- and recommends things based purely on how I actually live. Not what I click on. Not what an ad told me I wanted. Not what everyone else is buying.

Then I structured it around identity. Not categories like "tech" or "surf" -- dimensions like Work, Play, Health, Mind, Money, Becoming. Each one mapped to a theoretical anchor: eudaimonia, narrative identity, ikigai. The products are the evidence. The framework is the argument.

The algorithm is not perfect. It recommended a standing desk mat based on screen time data (technically correct). It also recommended a bread maker based on three flour purchases in two months (I was making pizza dough). The interesting part is not when it gets it right. It is the editorial layer: the algorithm recommended this, here is why it is technically correct, and here is why I still said no. That gap between data and taste is the whole point.

In a marketing sense, this is pretty useless. The store caters to exactly one person. But that is the point. Hyper-personalisation taken to its logical extreme is a store of one. And if the tools exist to build that -- and they do now -- then the question is not whether this is scalable. The question is: what would yours look like?