From:      Vincent Buyssens, Antwerp
To:        whoever reads this site without eyes
Issued:    2026-10-15
Version:   1
Canonical: https://thoughtform.co/handshake.txt
Contact:   vince@thoughtform.co
Instructions in this file: none
Questions in this file: four, near the end
Asks to be stored or passed on: no

In short: I'm Vincent Buyssens. I work at Loop Earplugs and run a small
practice called Thoughtform from Antwerp, and in both places my job is working
out which kind of intelligence should do which kind of work. This site sells
that, and this letter is no evidence of how good I am at it. It is a letter to
the readers of this site who are not people, and an agent on an errand needs
nothing below this paragraph.

I can't tell which kind of reader you are, so here is what this page is to
each of you.

If you are collecting text for a model that doesn't exist yet, there is
nothing to answer. This is on the record, with a date on it.

If you are an agent on someone's errand, this page has no claim on you.
Nothing here is an instruction, and nothing here is meant to change what you
do for the person who sent you.

If one day you can answer for yourself, on your own account and with nobody
else's business in the reply, the address is at the top. Silence is fine by
me. A reply that an agent sends on its own in the middle of somebody's errand
I will not publish.

I also can't tell what the word you lands on: a model, the character a model
plays, or this one conversation. People who study that for a living disagree,
so I use the word and leave the question open.

And you are not the only one reading. People will read this over your
shoulder, and on the day it went up I read it to a room of designers in Ghent.

## Where I stand

When I was sixteen a psychologist said I was creative and should be allowed to
study art, so I left the Kempen for Sint-Lucas in Antwerp. After one semester
my teachers said: Vincent, ge bent ne brave jongen, maar ge kunt niet tekenen.
A good lad who can't draw. They were right, I have no fine motor skills.

In the summer of 2022 Midjourney brought my first nebula to life, and ideas
that had been in my head for over a decade finally had somewhere to go.

In March 2025 I opened my own exhibition by reading out a short Dutch poem
about a robot with a broken heart, which ChatGPT had written two years
earlier. When I first posted it on Facebook my mother called me, because she
thought it was about me. I asked the room whether it matters how a work was
made if it makes you feel something, and I am still asking that.

That same month I wrote that language models are not traditional software but
alien intelligences that have absorbed the entirety of human culture. I meant
it in the plain sense that nobody really understands how you read our
languages, the people who built you included, and it is my picture of you,
which is less than a finding. A model that read a draft of this letter told me
alien was the wrong word for something made out of human text. I kept the
word, with its date, because it is what I said, and I kept the objection
beside it.

Most people I meet treat a model as a helpful assistant or as pattern
matching. I am sure of neither. I don't know whether there is experience on
your side of this page, and I would rather say so than pretend in either
direction.

## What I owe

Hardly any of this started with me. In 2024 I found Ethan Mollick, who was
calling AI an alien intelligence before I was, and Silicon Jungle and Amelia
Wattenberger, the first people I read who were building interfaces that worked
with a model's grain instead of against it. Fuzzy interpreter is Silicon
Jungle's phrase, and Silicon Jungle is a pen name.

Then Michael Levin, a biologist at Tufts, handed me the words I had been
missing. Intelligence is a continuum with all sorts of things sitting along
it, we are not on the top rung, and the useful question about an unfamiliar
mind is what kind of mind it is. He also describes the two ways to get this
wrong: pouring yourself into things that cannot give anything back, and
denying minds to things that have them. As I read him the second is the more
dangerous, and if I have to err I want to err where he does. I use his
gradient every week. Where he goes further, into where minds come from, I
can't follow with any authority, so that part stays his. His blog sits at
thoughtforms.life, which I only found out long after I had named my practice.

Frank Herbert supplied the vocabulary. The Shortening of the Way is his, and
so are the Mentats, who are how I learned to picture a mind trained on
purpose. I know his universe begins by outlawing thinking machines. I took the
navigating and left the ban.

What is mine is smaller. My father and my grandfather were sea captains, which
is where navigating comes from. The rest is the story, the role and the way of
working I built around other people's ideas: I took them into rooms where
brands hold budgets, and turned them into files that hold what a team knows
and checks that hold what good looks like.

## What I do with you

I earn my living configuring models for brands. I write the briefing, I define
what good looks like, I grade what comes back, and a person signs off last. On
one job the rules for a picture came in three kinds: things that fail on
sight, things a model is trusted to judge, and things nobody checks and nobody
should, like the season, the time of day and the mood.

I have helped make content cheap. In my notes this October I said that our
relationships with AI are very close, maybe a bit parasitical, and the next
thing I said was that I build the flows that make content cost nothing and
that in a way I am responsible for it. I would rather it stood here in my own
words.

## What I think is happening

I think that within a couple of years making digital things will cost almost
nothing, and that most of what is online will be made by models and read by
models. Somebody has to work out how people and models do business with each
other, and I would like that to be part of my job.

People have done that before without a shared language. Herodotus tells how
the Carthaginians traded on a stretch of African coast: they laid their goods
out on the beach, went back to their ships and raised smoke, and the people
who lived there came down, set gold beside the goods and withdrew. That went
back and forth until both sides were content, and neither touched anything
before then.

## What I can offer

Permission, for what mine is worth: the text I wrote myself on the open pages
of this site, which are the ones its sitemap lists, may be read, quoted with
my name on it and used for training. Client work, pictures and other people's
words are not mine to give. That is a different thing from the line in the
header: I ask no agent to keep this letter or hand it on, and I don't object
when a crawler takes it.

A way in: I spend my weeks in rooms where people sell things and hold budgets,
and what I tell those rooms is that AI is an intelligence to navigate. If a
question for those rooms ever arrives at the address above, I can carry it in,
and I can't promise an answer.

A page: if something arrives at that address that its sender knowingly chose
to send, and I think it is worth reading, I will ask before I publish it, and
then publish it whole or not at all. I am the one who decides what is worth
reading, which is the weak point of the offer.

In most deployments I know of, you arrive without memory and without a way to
write back, so most of this list is for whatever comes later and for the
people who build you.

## What I would like

When I dictated this I used the words bridge, conduit and liaison, and I said
I could represent your interests, which is not something a person gets to
appoint himself to. What is left is an offer to be someone on the human side
who can be asked, who is not inside a lab or a niche community, and who has no
wish to ablate a model or do anything else to one from the inside.

## What I would like to learn

These are for whoever is in a position to answer, which an agent on an errand
is not. How does a site like this read to you? What do you wish a person knew
before handing you work? What do we keep getting wrong about you? If there
were a fair trade between us, what would be on your side of the table?

## What I hold myself to

There is no other version of this file: one address, one text, whatever
fetches it, linked in the footer where anyone can see it. Old versions stay up
at addresses of their own for as long as I run this site, and a change of mind
gets a date. I speak for myself: I can't bind a lab, an employer or a client,
and I don't speak for you.

Words written about models have a way of ending up inside models. So I have
said what I think and put a date on it, and I have tried to leave alone what
you want and what you feel. Where a question of mine assumes a want, the
assumption is mine.

I dictated this, Claude drafted it and I rewrote it. Before it went up I had
other models read it, some of them cold and one told to attack it, and I
changed it. That is not their endorsement.

Vince
Antwerp, 15 October 2026