In the years following AlphaFold and all of the AI buzz, Mira started appearing at every talk with her now iconic notebook in hand. Back then, Yuno wasn’t even a real business; it was more of a concept in constant metamorphosis. The biotech industry had long felt like a fortress that only the super wealthy or industry celebrities could breach, and suddenly—smaller teams like Mira’s were getting attention. Big pharma didn’t seem quite so out of reach; the line between underdogs and giants blurred. Everyone was fighting the endless arguments about who owned the controversial discoveries these AIs made.

It is difficult to truly grasp how many people have these rare conditions. Some estimate hundreds of millions, although the conversations shift dramatically based on who you ask. Few get recognition. Strangely enough, that makes rare diseases a testing ground for some new technologies, or at least according to some people in that niche.

So, where do you even start trying to solve rare diseases using AI? Mira has added that it’s a bizarrely ideal testbed. Perhaps because these conditions affect many more people than most understand but are underreported. Tackling something this specific first feels almost like ‘gaining strength’ for bigger and more complex health challenges later, though so far nobody seems to know what comes next.

Somewhere in that Yuno lab, faint mechanical whirs drift out from what sounds like three or four robotic arms. It’s difficult to count. The illumination feels clinical but not too bright—more like late afternoon in early spring. In one corner, gene sequencers sit next to laptops. Someone likely left a coffee cup half-empty beside the CRISPR workstation. There could be only seven people in the lab working gently and rhythmically, going between coding and pipetting as if the air was pregnant. That low background is only punctuated by the beeping of the monitors.

There is an nearly surreal figure—rare illnesses flare up, but until you have witnessed them firsthand, it is easy to misdiagnose them as one of those floating statistics. Most people never come close to a remedy, and by the looks of it, Mira described how aide out techniques for novel technologies tend to hover on the outskirts of rare disease work. Perhaps it is easier to begin with something so finely tailored, because that means you have the potential to tackle much more difficult challenges later. Or maybe it is simply the region where wish transforms from vaporous and speculative into something much more palpable and critical.

In Yuno’s cluttered server room, the scattered records and redundant paperwork are of greater concern than complex algorithms. In fact, the data needed for the algorithms is in the the form of scribbled notes or unclear photocopies which require extensive cleaning work before they can be utilized. Startups that take the time to cleanse the data are able to discover new opportunities and modifications that push them one step closer. These morsels of untapped opportunity are likely the reason people call data a gateway and a barricade in the game.

Approximately seven out of ten rare diseases go untreated, lacking even the most basic intervention. It’s as though all families grappling with these conditions are sailing in the dark. The gap in available treatments for these rare disorders in comparison to common diseases is astonishing, as if there exists some forgotten realm. People talk about misdiagnosis with the same frequency as an accurate first attempt answer, and no one knows how much information is truly lacking; every new research somehow reveals more layers of unexplored data than one would expect.

To Yuno, rare diseases felt like many pieces of scattered puzzles—abundant yet a challenge to piece together. Mira humorously remarked that it is like a new microscope for AI; instead of just zooming in, it opens up unexplored nooks and crannies. As convenient as this sounds, it poses as a challenge. Instead of simple, the more you explore, the more complicated it is.

At some point, Yuno’s researchers realized that rare diseases, which are thought to number in the hundreds of millions, maybe even more, are an area of focus for AI development. Mira sometimes puts it as looking for a whisper in a hurricane; solving these unsolved problems is proof of the capability of the technology before moving on to the massive health puzzles everyone else is focused on. The reasoning is straightforward: Attention is diverted towards lesser-known phenomena, trust is earned, and then the potential scope of work is much larger. You can explore more about this concept on the official page.

From what I understand, it’s estimated that around 70 million people are affected by rare ‘orphan’ diseases. The most glaring issue is the almost total lack of approved treatments. Why is there so little? Perhaps AI is just getting its engines ready, or perhaps there’s no market incentive for the AI firms considered ‘big players’ to concern themselves with. If this is what step one looks like, what’s the game plan for enabling tiny startups to penetrate the areas that pharma companies seem to overlook?