The project
I am uploading myself
Everything I publish goes into a model. Not as a marketing trick and not as a demo, but as an experiment whose outcome I do not know.
Where it stands today
- 61
- texts in his head
- 35,020
- words
- 13 Jan 2026
- oldest text
- 12 Sept 2026
- last fed
Why there is no progress bar
A bar needs a denominator. What is one hundred percent of a person? I cannot answer that, and an invented number would paper over it. So there are only numbers that go up, and a list of what he cannot do yet. That list gets shorter. That is the meter.
Why
I write about software that acts on its own and about who is accountable when it does. At some point I wondered what happens when I point that at myself. Not an assistant that runs my calendar, but a version of me that keeps answering when I am not there.
The uncomfortable answer is that it simply works. No technical breakthrough required. All it needs is text, and I write that anyway.
So I am doing it in the open, with the counters on show. Then anyone can watch how complete a person gets from their own archive, and where it stops.
The upload log
Every time he learns something new, a line gets added. Only what is actually in there. No plans, no "coming soon".
- +2 new
Two pieces from early September
The piece about ads in ChatGPT and the piece about computer use and what that does to adoption. Ask him now what changes when software drives your computer itself, or why an ad inside a conversation feels different from an ad next to a search result, and he can answer with what I wrote about it myself instead of with a generality.
Could not do yet, at the time
- The exact publication dates, currently approximate
- My Copilot training sessions and the rooms I have stood in
- Why I started this project at all
- +2 new
The question that comes before all the others
Where do you start!? is in there, in both languages. It is the fifth instrument but it sits first, because it is about the question that almost always gets skipped: not which tool you buy, but which work you actually want to put into it. Ask him now why an AI project fails and he can name two things he did not know before: work you cannot check while a mistake is expensive, and knowledge that only sits in somebody's head.
Could not do yet, at the time
- The exact publication dates, currently approximate
- My Copilot training sessions and the rooms I have stood in
- Why I started this project at all
- +6 new
Four instruments, with the explanation
There are four: Tool choice, Model choice, the Compliance Check and the Credit calculator. Each now comes with an explanation for people who do not know the terms, and that explanation is in his head. Ask him what a credit is or why picking a model can become a budget line, and he can explain it in my words rather than a vendor's.
Could not do yet, at the time
- The exact publication dates, currently approximate
- My Copilot training sessions and the rooms I have stood in
- Why I started this project at all
10 · totaal
What he does not do
- He does not invent opinions. Everything he says comes from a text with a date, and he names that date.
- He makes no commitments. No prices, no agreements, nothing on behalf of me or my employer.
- He never denies being a model. Article 50 of the AI Act requires that, and I wrote about it the day it took effect. So it sits above the window, not in a footnote.
- And right now he is bad at this. He knows too little of me to sound like me. That changes; that is the whole idea.