Collector
A private catalog for the things I own: fragrances, spices, clothes, gadgets. Photograph it, scan it or paste a link, and it becomes a record I can find again.
At a glance
- Automation suggests, I confirm
- Have, Need, Test, Had
- Local first, cloud second
The question
I kept forgetting what I already had. Which fragrance I wore last week, whether there's still smoked paprika in the cupboard, when a warranty ends. Could one calm app hold all of that, quick enough to use with the object still in my hand?
What I built
A phone-first app where every kind of object gets its own collection with its own fields: notes and longevity for a fragrance, origin and heat for a spice, size and fabric for a jacket. Items can be added from a photo, a barcode, a Fragrantica page or by hand, and each one keeps a log of when I used it.
Everything lives on the device first and syncs to a private cloud account when I'm signed in, so search, browsing and logging keep working with no signal. Each collection also reads back patterns: favourite houses, most worn, what I've never used.
Inside
How it looks in use.
- Role
- Sole designer, developer and owner
- Built with
- Next.js, IndexedDB, Supabase, Gemini
- Runs
- Phone first, installable, works offline
- Status
- In beta testing





Up close
The details that carry it.



Decisions that mattered
- Automation suggests, I confirmPhoto recognition and enrichment fill in the record, but it always lands on a review screen first, with where each value came from.
- Have, Need, Test, HadOne simple switch covers owning something, wanting it, trying a sample and having used it up, so the wish list and the history live in the same place as the shelf.
- Local first, cloud secondWriting to the device before the network means logging a spray or adding a note is instant, and the sync can catch up later.
What I learned
The hard part of a personal archive isn't storing things, it's trusting what the machine filled in. Showing the source next to every suggested value did more for that trust than better recognition.
Something similar in mind?