Bexorg makes a dead brain go back to work.

The brain on the table is not thinking. It is, however, busy.

In a renovated building in downtown New Haven, a human brain sits in a sealed circuit. It arrived hours after its owner died, with no electrical activity and no obvious future. Then Bexorg connected it to BrainEx, pushed a custom artificial blood through its vasculature, and the cells did something inconvenient for our intuitions about death: they started metabolizing again.

This is the company as it exists today - five machines running, a sixth being built, and a small team of neuroscientists, engineers, and surgeons treating postmortem organs as instruments rather than relics. Bexorg is not trying to bring anyone back. It is trying to read the brain while it is briefly legible, and to sell that legibility to an industry that has been guessing for decades.

You cannot test a brain drug on a brain you do not have.

Here is the quiet scandal of central-nervous-system medicine: roughly 95% of CNS drugs that enter clinical trials fail. For Alzheimer's and Parkinson's, the figure creeps toward 99%. Companies spend a decade and a fortune, and the molecule that looked perfect in a mouse does nothing in a person - or does something nobody predicted.

The reason is unglamorous. Most preclinical brain research happens in rodents, primates, dishes of cells, or pea-sized organoids. None of them is a human brain. The blood-brain barrier - the bouncer that keeps most drugs out - behaves differently in each. By the time a therapy meets an actual human brain, it is meeting it for the first time, in a trial, with patients attached.

Two Yale neuroscientists wagered that "postmortem" was negotiable.

In 2019, a lab run by Nenad Sestan at Yale published a Nature paper that made ethicists reach for their coffee. Four hours after death, pig brains - perfused with a bespoke fluid the team called BrainEx - showed restored cellular and metabolic function. Not consciousness. Not thought. But not nothing, either. The result was careful, controversial, and impossible to un-see.

Zvonimir Vrselja, a physician-neuroscientist who helped invent the perfusion technology in that lab, made a bet: the same machinery that revealed something about death could be pointed at disease. In 2021 he and Sestan founded Bexorg to turn an academic marvel into a drug-discovery platform. Vrselja runs it as CEO. Sestan stayed close as a scientific advisor. The wager was that pharma would pay for the one thing it never had - a human brain that responds.

One platform reads the brain. The other learns from it.

Bexorg's offering has two halves that need each other. The wet lab generates data no one else can. The AI makes that data worth more with every run. It is a flywheel disguised as a biology experiment.

The data is the moat. The partners are the receipts.

A reactivated brain is a remarkable demo. A signed pharma collaboration is a business. In June 2025, publicly traded Biohaven committed to a multi-program partnership, putting Bexorg's platform behind preclinical programs that included compounds targeting brain metabolism and Parkinson's. Gene-therapy work tied to the University of Oxford and the UK Medical Research Council followed. Investors noticed too: Engine Ventures led a $23M Series A, alongside Amplify Partners, Starbloom Capital, Connecticut Innovations, and E1 Ventures.

Move the first human brain a drug meets from the trial to the lab.

Bexorg's business model is platform-as-a-service: it runs drug-testing programs for biopharma and academia on BrainEx, and licenses access to one of the most comprehensive human CNS datasets being assembled anywhere. Pricing details stay private. The logic does not. Every program that validates in human tissue before a trial is a program that fails cheaper, earlier, and with no patients in the room.

The competition is everything Bexorg is trying to retire - rodent studies, primate models, organoids, iPSC-derived neurons, static tissue banks. Each is a useful approximation. None responds. Bexorg's claim is narrow and large at once: an intact, metabolically reactivated human brain is a better preclinical model than a guess, and the industry has been running on guesses.

If it works, the failures get cheaper. If it scales, the cures get closer.

Bexorg plans to run on the order of 1,000 whole-brain experiments a year. Each one feeds XO Digital, which makes the next experiment smarter and the dataset more valuable. That is the part venture capital underwrote: not a single brain on a table, but a system that gets better the longer it runs. The questions that remain are the hard ones - how far human-brain data predicts human-patient outcomes, how the ethics hold up under scale, whether pharma rewires decades of habit. None has a finished answer.

Return, then, to the brain in the sealed circuit. Hours ago it belonged to someone who, in many cases, had a disease Bexorg now studies - dementia, Parkinson's, Alzheimer's. It will never think again. But for about a day, its chemistry answers questions, and those answers may spare a future patient a drug that was never going to work. Bexorg did not bring the brain back. It gave it one more job. That is the bet, running five machines at a time, in New Haven.