Evolution has spent millions of years writing chemistry into DNA. Hexagon Bio learned to read it - and turn it into medicine.
In a Menlo Park lab, a computer is reading the genome of a fungus most people would scrape off bread. It is looking for a molecule no one has ever made, and the protein in a human cell that molecule was, in effect, designed to grab.
That is Hexagon Bio on an ordinary Tuesday. The company does not screen millions of random compounds and pray. It treats DNA as a catalog of finished inventions - small molecules that microbes evolved over millions of years to do something very specific to a living cell - and uses software to find the useful ones before anyone picks up a pipette.
Today that focus is sharp: novel cytotoxic payloads for antibody-drug conjugates, the guided missiles of modern oncology. The premise is almost suspiciously simple. Nature already built potent cell-killers. The trick was never making them. The trick was finding them.
Some of medicine's greatest hits came from microbes - penicillin from a mold, statins from a fungus, a long list of antibiotics from soil-dwelling bacteria. The inconvenient truth is that nearly all of them were found by accident. For decades, "natural product discovery" meant grinding up organisms, testing the soup, and hoping something interesting fell out.
It was slow, it was random, and the pharmaceutical industry quietly walked away from it. Combinatorial chemistry was tidier. You could patent it. You did not have to go foraging. The only problem was that nature remained, by a wide margin, the better inventor - and most of its catalog stayed unread, locked inside genomes nobody had the tools to interpret.
Caption: A fungus is not trying to cure your cancer. It is trying to win a turf war with its neighbors. The molecules it deploys just happen to be ruthless - which, for an ADC payload, is the entire job description.
Hexagon Bio was founded in 2017 by four scientists who had spent years at the intersection of computation and chemistry. CEO Maureen Hillenmeyer had directed the Genomes to Natural Products program at Stanford. Yi Tang, a chancellor's professor at UCLA, brought deep expertise in fungal biosynthesis. Brian Naughton came from the data side as the founding CTO, and Colin Harvey built out synthetic biology.
Their bet was that DNA sequencing had finally gotten cheap and machine learning finally good enough to flip the old process on its head. Instead of finding a molecule and reverse-engineering it, you could read the genome, predict what molecule its gene clusters encode, and - the genuinely hard part - predict the human protein that molecule is shaped to hit. Discovery before synthesis. Biology before the bench.
Co-Founder & CEO. Former director of Stanford's Genomes to Natural Products program.
Co-Founder. Chancellor's professor of chemical and biomolecular engineering at UCLA.
Co-Founder and former CTO. The data-and-genomics half of the founding idea.
Co-Founder and VP, Synthetic Biology. Turns predicted molecules into real ones.
The engine is a computational discovery platform that braids together four disciplines that rarely share a hallway: machine learning, genomics, chemistry, and synthetic biology, with automation stitching them into a loop. Feed it a microbial genome and it predicts the small molecules encoded there, ranks them, and - critically - proposes which human proteins they are likely to engage.
That last step is the moat. Plenty of groups can sequence a fungus. Connecting a predicted molecule to a disease-relevant target, before you have ever made it, is what compresses years of guesswork into a shortlist. Synthetic biology then expresses the promising candidates so chemists can confirm what the software suspected.
Fuel matters too. Hexagon Bio has built a proprietary database it describes as roughly ten times larger than all public databases combined, and says it adds thousands of new genomes every month. A discovery platform is only as good as the data it reads, and the company has spent years making sure it reads more than anyone.
A platform's promises are cheap; here is what backs them. Hexagon Bio has raised roughly $272.9M across three rounds, capped by a $77.3M Series B in February 2023 that brought in the Canada Pension Plan Investment Board alongside returning backers The Column Group, Two Sigma Ventures, 8VC and Nextech.
The hiring tells its own story. The 2023 round came with a leadership upgrade aimed squarely at turning discoveries into drugs: Tara Arvedson, formerly of Amgen, stepped up as Chief Scientific Officer, and Victor Cee - who helped deliver the KRAS inhibitor LUMAKRAS at Amgen - joined to lead drug discovery. You do not recruit that bench to run a science project.
Then came the validation that mattered most, because it came from outside pharma. In December 2025, agriculture giant Corteva formed a multi-million-dollar joint venture with Hexagon Bio to hunt for nature-inspired crop protection. It was Hexagon's first venture in agriculture and Corteva's first in pharma-style discovery - a sign the platform travels beyond the disease it was built for.
Reported financing amounts (USD millions). Series A extension figure is cumulative.
Caption: The bars climb, but the more interesting number is the one outside this chart - a JV partner from agriculture, betting the same software that hunts cancer payloads can defend a corn field.
The stated goal is unglamorous and enormous at once: develop treatments for diseases of unmet need, starting with small molecules pulled from microbial genomes. The first front is oncology, where novel ADC payloads could help overcome the resistance that blunts today's cancer drugs. Infectious disease has been part of the conversation from the early rounds.
Underneath sits a bigger idea. If nature's chemistry can be read directly from DNA, then genomes become a renewable engine for discovery - not a one-time lottery. The Corteva deal hints at where that leads. The same platform that proposes a cancer payload can, in principle, propose a fungicide. Medicine first, but not medicine only.
Return to the Menlo Park lab, and the Tuesday-morning scene from the start. The computer is still reading a genome. But the picture around it has shifted. What was once a lucky accident - the mold on the petri dish - is becoming a process you can fund, staff, and repeat.
The skeptic's question is fair: can software really predict, in advance, which molecule will hit which target? Hexagon Bio is spending nine figures and a roster of veteran drug-makers to answer it.
If the bet pays off, the implications run past any single drug. A renewable, genome-fed pipeline of bioactive molecules would reshape how we find medicines - and, the Corteva deal suggests, how we protect crops too. If it does not, it will still have been one of the more elegant attempts to industrialize luck. Either way, the fungus on the bench is no longer waiting to be discovered by accident. Someone is finally reading it on purpose.