Creyon Bio

Here is a thing that is true about drug discovery, and that everyone in drug discovery would prefer you not think about too hard: most of it is guessing. Sophisticated, expensive, credentialed guessing, performed by very smart people in very clean rooms - but guessing. You make a molecule, you test it, it disappoints you, you make another one. Repeat for years. The industry has a polite name for this. It calls it "screening." Creyon Bio has a less polite view, which is that you shouldn't have to make thousands of things you already suspect won't work.

Creyon is a San Diego biotech founded in 2019 by two alumni of Ionis Pharmaceuticals - Chris Hart, who ran functional genomics there, and Swagatam Mukhopadhyay, who was a principal computational biologist. Ionis matters to this story because Ionis essentially invented the class of drugs Creyon works on: oligonucleotides. These are short strings of engineered genetic material - antisense oligonucleotides, siRNA, RNA-editing systems - that don't attack a disease's downstream symptoms but go upstream and quietly adjust the RNA instructions a cell is reading. It is an elegant idea. It has produced real, approved medicines. It has also historically been designed the slow, artisanal way.

So Hart and Mukhopadhyay left, and their pitch was roughly: what if you treated this like an engineering discipline instead of a craft? Aerospace does not build a thousand wings and see which ones don't fall off. Chip designers do not fabricate a million chips and keep the ones that boot. They simulate. They have rules. Creyon's bet is that oligonucleotide medicine can have rules too, if you gather enough of the right data and point enough of the right math at it.

That last point is worth dwelling on, because it explains what kind of company Creyon is trying to be. It is not purely a drug company, chasing one molecule toward one approval. It is not purely a software vendor, either. It is the increasingly popular hybrid: a platform that sells its capability to big partners while also aiming its own three wholly-owned programs at the clinic - a lead neuromuscular candidate the company has aimed toward clinical development, plus work spanning delivery tricks like a transferrin-receptor aptamer designed to sneak drugs across the blood-brain barrier. The logic of the hybrid is that the platform makes every future product cheaper to build, and the partnerships pay the bills while your own pipeline compounds in value. When it works, it is a very good business. When it doesn't, you are a services shop with a science-fair poster. The line between the two is execution.

It helps to be concrete about what the engine is actually for, because "AI drug design" has become a phrase that means everything and therefore nothing. Creyon's modalities are specific: antisense oligonucleotides that mask or redirect a stretch of RNA; siRNA that instructs a cell to silence a gene's message; and RNA-editing systems that rewrite the message directly. Each of these depends brutally on getting the chemistry and sequence exactly right - a few atoms in the wrong place and a promising drug becomes toxic, or inert, or lands in the wrong tissue. This is precisely the kind of high-dimensional, physics-constrained problem where a well-fed model can plausibly beat a human running experiments one at a time. Creyon didn't pick a single modality and specialize; it built a design layer meant to work across all of them, on the theory that the underlying rules are shared. Breadth is a strategy when depth has become table stakes.

What can a partner or a patient actually get out of that? For a drug company like Lilly, the deliverable is speed and shots on goal: hand the engine a validated target and get back optimized candidates without burning years on brute-force screening. For patients - eventually, if the clinic cooperates - the promise is medicines aimed at diseases that were previously too rare, too fast-moving, or too commercially awkward to justify a decade of trial-and-error. Creyon frames its focus around neuromuscular, central-nervous-system, and immunologic conditions, with delivery tricks aimed at the hardest addresses in the body, like the brain. That is the payoff it is selling. Whether it arrives is the one question no amount of quantum chemistry can yet answer.

There is also a management chapter, because there always is. In September 2024, Creyon brought veteran biopharma operator Serge Messerlian onto its board; by April 2025 - the same window as the Lilly announcement - he was Chief Executive Officer. Messerlian is not a lab founder. He ran Janssen Oncology, the cancer arm of Johnson & Johnson, and was CEO of Teon Therapeutics, and spent years around big-company M&A and multi-billion-dollar operations. Installing an operator of that profile at the top of a 31-person startup, right as it lands a pharma megadeal, is a fairly transparent signal about where the founders think the hard problems are now. The science got them the option. Turning options into approved medicines - and into more deals like Lilly's - is a different sport.

None of this is proven yet, and it would be dishonest to pretend otherwise. Creyon has not put a drug on the market. Its lead program is still marching toward the clinic, not through it. AI-for-biology is a crowded, hype-soaked field, and "we can predict the drug before we make it" is a claim the entire industry would love to be true and has repeatedly found harder than advertised. Quantum-level modeling is computationally expensive and famously unforgiving; machine learning is only as good as the data feeding it, and biological data is notoriously noisy. Creyon competes, directly or by analogy, with the RNA establishment - Ionis, Alnylam, Wave, Arrowhead - and with a growing pack of computational upstarts. Any of the usual biotech disappointments could arrive on schedule.

But the shape of the bet is clear, and it's a good bet to understand even if you don't know whether it pays. Creyon looked at an industry that treats molecular design as something between an art and a lottery, and decided the interesting innovation wasn't a new molecule at all. It was a new method for arriving at molecules - one where the expensive, humbling part of the experiment happens in a simulation, and the things that reach a human being have already survived a computer's skepticism. If they're right, they don't just make one drug. They make the factory that makes the drugs. Eli Lilly, which has seen more drug factories than almost anyone, apparently thought that was worth up to a billion dollars to find out.