The sign before the meaning

Some logos are born from an animal, an initial or an abstract shape. Deraison.ai's comes from a slightly less expected place: an algebra book printed in Zurich in 1659.

Artificial intelligence didn't exist back then, obviously. Neither did computing. And yet mathematicians were already wrestling with a strikingly modern question: how do you give a visible shape to an abstract idea?

That's what interests me most. I hold a PhD in mathematics and deraison.ai is my artificial intelligence company. When I set out to redo the site and the visual identity, I didn't want to add yet another stylized brain, glowing neural network, or tech hexagon to the pile of AI logos already out there. I wanted to go further back, to the moment we started inventing the signs we think with.

1659: when an operation becomes a sign

In 1659, the Swiss mathematician Johann Heinrich Rahn published his Teutsche Algebra in Zurich. The book holds a fairly small place in the collective imagination. Yet it contains a remarkable milestone in the history of mathematical notation: the earliest known printed use of the ÷ sign to represent division. Rahn also uses the asterisk * for multiplication and arrangements of three dots, notably ∴, to express consequence ("therefore").

The passage devoted to division is almost disarmingly simple: Rahn presents ÷ as the main sign for dividing. The obelus existed before him as a typographic mark, but it's in this Teutsche Algebra that it first appears in print carrying this specific mathematical function.

This story has to be told with the caution mathematicians are fond of: Rahn was working at the time with the English scholar John Pell, who had tutored him in mathematics. Historians can't determine with certainty which of the two originated all of the book's notational innovations. The exact authorship of ÷ remains disputed.

That uncertainty, if anything, makes the story more interesting. Because what matters most may not be knowing who, on some afternoon in 1659, first put those three strokes of ink on a page. What matters is the gesture itself: turning a mental operation into a symbol.

A few dots and a line

A division is an idea. ÷ is an interface. In a few millimeters of ink, an entire operation becomes recognizable, transmissible and usable.

This is one of mathematics' great powers: it doesn't just discover relationships, it also invents languages compact enough for the mind to actually work with them. A bar. A few dots. And suddenly, something that used to require a whole sentence becomes a sign.

That's where the deraison.ai logo comes from. It isn't an archaeological reproduction of a character from the Teutsche Algebra. It borrows its grammar instead: the dot, the line, the punctuation, the relationship between discrete elements. A small typographic constellation. At first glance, almost nothing. But mathematics has taught us to be wary of simple shapes: =, +, ∞, ∫, ∴, ÷. Some of our civilization's most powerful ideas fit in a handful of pixels.

From algebra to artificial intelligence

Three and a half centuries later, the problem has changed scale, but not entirely in nature. Contemporary artificial intelligence also relies on our ability to turn the world into representations we can manipulate: words become tokens, objects become vectors, relationships become matrices, observations become numbers. Billions of parameters eventually produce a representation a machine can compute with, predict from or generate from.

Between the ÷ sign of 1659 and a model with billions of parameters, the technological gap is dizzying. But one thread connects them: inventing a representation that makes computable what previously wasn't.

That's an idea I wanted to build into deraison.ai's identity: an artificial intelligence that presents itself as neither magic nor an oracle, but as the continuation of a very long history of formalization, representation and computation.

Why "deraison"?

The name deraison.ai deliberately carries a tension. Mathematics is often presented as the territory of absolute reason: proofs, axioms, rigor, necessity. Contemporary artificial intelligence complicates that picture. We build systems using extremely precise mathematics and then get behaviors we can't always reduce to a simple chain of explanations. Everything is computed. And yet not everything is immediately intelligible.

That's exactly the space where the name deraison finds its meaning. Not against reason, but beyond obvious reason, where you have to experiment, measure, understand and sometimes accept that a complex phenomenon precedes the elegant explanation we'll eventually give it. As a mathematician working in artificial intelligence, I particularly enjoy this paradox: using the most rigorous tools we have to explore systems whose properties can still surprise us.

A logo as a program

So I wanted the new logo to say very little explicitly and a lot implicitly. No robot. No brain. No artificial spark. Only elementary shapes: dots, because before the curve there is data; a line, because between points we look for relationships; punctuation, because a thought becomes operative only once we find a way to write it down. And a nod to that fascinating period when modern mathematics was gradually inventing its own alphabet.

The deraison.ai logo is thus less a drawing than a small system of signs. It's also a way of asserting a certain view of artificial intelligence: beneath the apparent complexity of models, return to fundamental structures. Look for the right formalism. Find the right representation. Strip away unnecessary complexity until what remains is simple enough to become intelligible.

A score, an emotion

There is another story of signs that has stayed with me all my life: musical notation. A staff, a handful of black notes, rests, dynamics, ties. Almost nothing, to the eye. And yet, when a musician looks at it, they don't see symbols. They see a shiver, waiting.

It's the same gesture as in 1659: capturing something invisible, an intention, a feeling, a kind of vertigo, inside an object stable enough to survive time. A Bach score is three centuries old. It has outlived its composer, his instruments, his concert halls. And yet, the instant a bow touches a string, the emotion it holds wakes up intact, as if time had never passed at all. The sign carried it across.

A mathematical proof gives me a kindred emotion, tuned to a different key. Music opens straight onto feeling. Mathematics opens onto a victory. There's a moment I know well, that I go looking for, almost the way you go looking for the right note: the one where the hypotheses finally lock together and the conclusion can no longer be anything else. It doesn't just "work." It's a small, silent triumph: reality, briefly stubborn, has just yielded to an idea born inside a human skull.

That, I think, is where the root of my intellectual satisfaction in this profession actually lives, far from a model's score on a benchmark or the race for more compute. It lives in that moment when a vague intuition, almost too shy to put into words, becomes a clean result, when the whole world lets itself be held, for a fraction of a second, inside a sign. A well-built artificial intelligence occasionally gives me that same sensation: a vast, opaque system suddenly starts doing something right, something sturdy, something useful, something you can finally show, the way you show a proof, the way you finally play the score you'd been silently rehearsing in your head.

Symbols outlive machines

There's something dizzying about opening a 17th-century text today and immediately recognizing a sign. Machines have changed, the media that carry them have changed, and so have the theories behind them. Paper became screens, tables of calculations became GPUs, and the handful of symbols on a page of algebra became billions of operations per second. But we're still doing what Rahn, Pell and their contemporaries were doing: looking for better languages to think with.

That, in the end, is perhaps what the deraison.ai logo represents: a tribute to an old mathematical story, but above all a very current conviction: before you can compute something, you have to find a way to represent it. In 1659, that fit in a few dots and a line. Today, everything still starts with a sign.

Thanks to Audrey Dejoux, the artist who designed this website with a delicate tenderness, as close as her creative imagination could get to the flame I carry for science.

Sources:
MacTutor History of Mathematics — Johann Heinrich Rahn
Sotheby's — Teutsche Algebra by Johann Heinrich Rahn, 1659