A medicine for every person.

In 1928, Alexander Fleming came back from a summer holiday to a bench piled with culture plates he'd been meaning to clean. One had grown a mould, and around the mould the bacteria were dead. That was penicillin.

It took thirteen years before the medicine reached its first patient. In 1941 an Oxford policeman, Albert Alexander, was dying from an infection that had started with a scratch on his face. Within days he was sitting up and eating.

Then it ran out — they had been recovering the penicillin from his urine, and even that wasn't enough. The infection came back, and it killed him.

The medicine worked. They just couldn't make enough to save one man.

In 1990, a four-year-old named Ashanthi de Silva had no working immune system. One broken gene, and a cold could have killed her. Doctors put a working copy of that gene into her cells. She grew up.

In 2023, Casgevy became the first FDA-approved CRISPR gene-editing treatment for sickle cell disease. It works by editing a patient's own cells outside the body, then returning them. In all of 2025, only sixty-four people received Casgevy infusions.

The treatments exist. But a patient receives one only after it has been made the same way it was tested, checked, recorded, released, and delivered in time. That work is real, and you cannot wish it away.

But watch how that work actually gets done. Every step is performed once, then recorded all over again by hand — on paper, after the fact. Space-age medicine, run on a clipboard.

And it is about to get much harder. The newest treatments are grown from a single patient's own cells, a batch of one. Every patient becomes their own manufacturing run, and no one can do that by hand, millions of times over.

So it can only be done in software — the whole process in one system, run by an ever-evolving civilization of AI scientists, engineers, and quality experts, every batch under their watch at once: catching the one that will fail before it is made, tracing a fault to its true cause across a thousand past runs, testing each change in simulation instead of in a patient. The decade after discovery becomes days.

That is how the treatment that reached sixty-four people reaches a sixty-fifth — a patient waiting on a batch with her name on it, made in time. And then the next, and the next. Another patient, another batch, until there is a medicine for every person.