Chemistry is not an art.
It is computation.

For centuries, an art

Chemistry has largely been taught and practiced as a craft: thousands of facts to memorize, reactions to recognize by name, tables to consult, intuition earned only over years at the bench. It has been extraordinarily productive — and, for most people, extraordinarily opaque. The knowledge lived in experience and in reference works, learned by recall more than by reason.

Underneath, it was always physics

But structure and reactivity are not arbitrary. How atoms bond, which shapes they take, whether a reaction runs — these follow from a small set of physical principles. Given the atoms and the rules, the outcome is determined. It does not have to be looked up in a catalog of what someone has already seen. It can be derived. Chemistry, at bottom, is deterministic — closer to mathematics than to art.

A different kind of AI

Aufbau is an AI system, but not a language model. The models in the headlines predict what looks right — they can be confidently wrong, and no two runs need agree. Because Aufbau derives its answers from physical law instead, the same atoms always give the same result: deterministic, reproducible to the last bond, and traceable to the energy that produced it. When the physics can't decide, it says so rather than invent an answer. It doesn't guess.

Aufbau — a new way to study it

Aufbau exists to make that shift visible. Rather than recalling results, you watch them emerge: molecules assembling from atoms, reactions resolving to their products, each following from principle rather than from a stored answer. It is a way to understand chemistry — to see the reasoning, not just the outcome — that hasn't really been possible before.

The app is free, and it will stay free. The goal is not to sell a tool; it is to change how chemistry is learned — to let anyone, not only the specialist, see that the subject is knowable from the ground up.

The larger idea

If chemistry can be computed from first principles rather than retrieved from a database of the already-known, then the same reasoning that teaches in the app can, in time, help answer questions no database can — reaching combinations that have never been recorded. That is the direction we are building toward: chemistry you compute, not chemistry you catalog.

We are early on that road and honest about it. But the premise is settled, and it is the reason for everything here.

The next step: chemistry in silicon

Determinism has a consequence most people miss. Because each element's behavior follows a small, fixed rule — not billions of learned parameters — that rule can be built directly into hardware: a fixed-function circuit for chemistry, massively parallel, running on-premise and even air-gapped. A language model can't be reduced to a tiny circuit per element; it needs general-purpose machines. Determinism is the one property that lets chemistry move out of software and into silicon.

We call the direction a chemistry co-processor: the same reasoning Aufbau uses today, realized as an array of simple, verifiable cells whose results are identical to the software — in space, in time, and on any machine. And it need not be silicon alone; the same fixed logic can live in electronic, photonic, or ionic hardware.

This is a direction, not a product yet — early, and patent pending. But it follows from the same premise as everything else here: chemistry you can compute, and therefore build.

Working with us

Researchers, educators, and organizations who share this view — and who would like to build on the computation behind Aufbau — are welcome to get in touch.