The founder who turned the doctor's inbox into data
Every day, millions of people type a worry into a patient portal. "My fever is back." "The swelling hasn't gone down." Most of those messages sit in a queue. Anin Sayana looked at that queue and saw an early-warning system nobody had switched on.
Sayana is the co-founder and chief executive of Quadrant Health, a New York company that reads electronic health records and patient messages in real time and flags trouble before it becomes an emergency. The pitch is blunt: analyze the chart and the inbox together, predict patient harm earlier, and hand clinicians a heads-up instead of a post-mortem. The company came out of Y Combinator's Winter 2021 batch and is run by Sayana alongside co-founder and CTO Krishna Gorrepati.
What makes the bet interesting is that it is not really about chatbots. It is about timing. Healthcare spends enormous energy reacting - to the deterioration that already happened, the readmission already booked, the message already three days old. Quadrant Health's whole premise is to move the clock forward. Catch the signal in the language before the lab work confirms it.
By 2026 that thesis had grown teeth. In March, Quadrant Health teamed up with the EHR vendor medent to launch an AI Receptionist - software that handles incoming calls, scheduling and patient triage for medical practices. The receptionist is a clever wedge: the same language models that read a worried portal message can also pick up a ringing phone, figure out why someone is calling, and route them. The inbox and the front desk are the same problem wearing different clothes.
That 93.23% figure is the one the company likes to quote: a custom clinical model reported to outscore previous AIs on the United States Medical Licensing Examination. Whatever the leaderboard ends up saying, the number tells you where Sayana points his energy - at models that can hold their own inside a clinic, not just a demo.
A science-fair kid who emailed Stanford and got an answer
Rewind to 2012. Sayana was an incoming high-school freshman at Bellarmine College Preparatory in the Bay Area, fresh out of Challenger School in Sunnyvale. He was also, improbably, a top-10 finalist in the Discovery 3M Young Scientist Challenge - one of the country's marquee science competitions for middle schoolers.
His project was not a baking-soda volcano. It used carbon nanotubes to deliver treatment aimed at chemotherapy-resistant cancer stem cells, the stubborn cells linked to relapse. He needed expertise he did not have, so he did the thing most 13-year-olds never try: he cold-emailed Stanford professors and asked for help. They wrote back. They mentored him. He spent a summer working with a 3M inventor to push the project further.
Read his career backward and the through-line is obvious. The nanotube project was about catching the cells that come back. Quadrant Health is about catching the harm before it lands. Same instinct - get ahead of the thing - just traded a wet lab for a language model. At 13 he told a reporter he wanted to study neuroscience. He went to Stanford and studied computer science instead, and somehow ended up back at the same problem from a different door.
What Quadrant Health actually does
Reads the record live
Continuously analyzes EHR data and patient messages to spot deterioration, infection or harm earlier and more reliably than a queue ever could.
Sorts the inbox
A custom large language model categorizes patient self-reported symptoms and needs - work tied to a collaboration with Stanford Health Care and published in npj Digital Medicine.
Answers the phone
The 2026 AI Receptionist with medent handles calls, scheduling and triage - the same intelligence applied to the practice's front door.
Built to pass the test
The team obsesses over models that perform inside medicine, citing a reported 93.23% on the US Medical Licensing Exam as proof the bar is clinical, not cosmetic.
From nanotube to neural net
The quirks worth keeping
There is a particular kind of founder who never really changes their question, only their tools. Sayana is one of them. The boy who studied relapse is now the CEO studying recurrence of a different sort - the readmission, the missed message, the call that goes to voicemail.
He is also proof that the cold email works. A 13-year-old asked working scientists for help and they said yes. That small act of nerve - assume the expert might answer - is the same nerve it takes to start a company that tells hospitals their inbox is a diagnostic instrument.
It is a humble surface hiding a big claim. Nobody romanticizes the patient portal. It is the dull, overlooked plumbing of modern care. Sayana's wager is that the most valuable signal in medicine is hiding in the most boring place - the words patients already type, the calls they already make - and that whoever reads them first, fastest, gets to change the ending.