The whole system

A digital model of an entire life, kept up to date daily with AI. I built the first one for myself.

The problem

Finances, admin, property, diary, contacts, health. Each one lives somewhere different: your head, a folder in your inbox, a spreadsheet last updated months ago, a drawer full of paper. And one person trying to carry all of it.

Everyone knows how that turns out: deadlines that appear out of nowhere, documents you look for three times over, decisions taken without the full picture because assembling the picture cost more than deciding. It is not a problem of personal untidiness. It is that no human can hold a system with that many pieces in their head.

What was built

A digital twin: an ordered model of what there is, with every fact in a single place, and an AI operating it daily. It is not one more app. It is the structure underneath all of them.

The model has three layers. The hard data: bank transactions that arrive on their own through an API, tasks, calendar, contacts, health measurements. The documents: more than 300, ordered and searchable in plain language. And on top, the AI that maintains the whole thing: when I tell it something, it updates every piece affected; when I ask it something, it answers with my data and cites where each thing comes from.

And one piece almost nobody puts in: an automatic audit that checks every day that nothing is lost. That every fact is where it should be, that no task has been orphaned, that the views match their source. A system nobody watches degrades; this one watches itself.

Everything runs locally. The data never leaves my machine.

How it works day to day

I write it one sentence: “the garage invoice came in, €240”. The system logs the expense on the car's record, closes the task about taking it in, puts Thursday's collection in the calendar and writes the day's entry in the log. I say what happened; it updates everything else. (A staged example with sample data, as in the demo.)

Finances are not typed in: transactions arrive on their own through an API and the console shows the month's spending by category and an 18-month cash projection. Grants and deadlines are watched and flagged, and the investment rules are written down and the system checks that they are being followed.

When I need a figure, I ask in plain language: “how much have I spent on the car this year?”. The answer comes with the breakdown and with citations to the records and the log each figure comes from. (Sample figures in the demo.) Without citations, an AI answer is an opinion; with them, it is a fact you can check.

The same goes for every other area: projects with their next step clear, tasks and calendar in sync, contacts up to date. In health, a training plan with automatic progression, body composition measured to a protocol, nutrition logged and rest from the wearable (WHOOP) connected since July 2026.

Result

This is not a prototype: it is my daily life, kept up to date with AI since June 2026, with an automatic audit checking every day that nothing is lost. The method is the same one I apply to a business: first the model of what is; then the AI working on top of it.