My dad's family tree, made readable

Why I wanted it

My dad has been doing genealogy for longer than I have been doing most things. His database holds 19,628 people and 4,124 families, with gigabytes of scanned documents, photographs, and, because he drove to the cemeteries himself, hundreds of headstone pictures. It runs on a genealogy website he set up years ago and still maintains.

It is a remarkable thing, and it reads like a database. Every ancestor is a row. I wanted each one to read like a magazine profile: a photo, a life span, a few paragraphs of story with the sources right there, a timeline, the family around them, and a place for the rest of us to leave memories. I also wanted to be able to just ask the tree a question.

What happened

The first rule I wrote down was: don’t touch Dad’s site. Everything I build sits beside it, reading a copy of his data, and his site keeps running exactly as it always has.

Over two days I laid the foundation. The data comes across in one direction, into a clean structure I control. Every AI capability, researching a person, reading a scanned document, writing a narrative with citations, starts life as a command I run by hand and read the output of. I made that choice on purpose: the pipeline exists, the automation does not, until I trust it.

The center of gravity is research, not narration. An AI can write a nice paragraph about anyone. What makes a family site worth having is durable, well-cited knowledge underneath the paragraph.

Then I stopped. The foundation is done and tested. The part you would actually see, the profile pages, is not started.

What it looks like

Nothing yet. That is the honest answer. There is a design for the profile page that I like a lot, and no screenshot of it.

What’s next

Build the profile page. It is the part that would make my dad’s decades of work come alive, and it is the part I have been circling. When it exists, this page will show it.