Overview

PatientBank was a San Francisco-based health technology company that automated the process of requesting, digitizing, and managing medical records from any U.S. doctor or hospital.Founded in 2015 by four Yale alumni and backed by General Catalyst, Khosla Ventures, and Y Combinator, the company built genuine operational infrastructure — a database of 2.5 million healthcare contacts, multi-channel submission systems, and a HIPAA-compliant records platform — and scaled to serve over 150,000 users before shutting down in January 2018.

The core thesis of failure was not a product problem or a demand problem.PatientBank's consumer-facing, per-record fee model could not generate sufficient revenue to offset the high operational costs of manually bridging a healthcare system still overwhelmingly dependent on fax and paper.

The company solved a real problem at real scale, but the unit economics of acting as a human-powered middleware layer between patients and an analog hospital infrastructure proved structurally unsustainable.

Founding Story

The idea for PatientBank originated from a moment of personal absurdity. Paul Fletcher-Hill, a Yale Computer Science and Economics graduate, wanted to build a website that visualized his own medical data.^[1] He assumed accessing his records would be straightforward. Instead, he discovered he had to submit a fax request — and still did not receive his records within 30 days.^[2] That experience, which would be familiar to most Americans who have ever tried to transfer their own health history, became the founding insight.

Fletcher-Hill began building a solution in 2014.^[3] He assembled a team from his Yale network: Feridun Mert Celebi, Kevin Grassi (an MD who brought clinical credibility), and Graham Kaemmer. The combination of technical, medical, and business backgrounds gave the founding team a credible profile for a healthcare-adjacent problem that required navigating both software development and institutional healthcare bureaucracy.^[4]

The company was formally incorporated as Health Exchange Technologies, Inc. and began operating publicly as PatientBank in 2015.^[5] Its first institutional partner was the Yale-New Haven Health System, and that early relationship proved instructive. The team discovered that the records access problem extended far beyond individual patients — hospitals, researchers, and private practices all struggled with the same fragmented, paper-dependent infrastructure.^[6] This was an early signal of potential B2B scope that the company would later deprioritize.

The team applied to Y Combinator and was accepted into the Summer 2016 batch. That acceptance triggered a deliberate strategic pivot. Fletcher-Hill explained the reasoning: "We decided consumers represented the best way to expand quickly, as individuals often end up being the vehicle for collecting or sharing records between their doctors or other providers."^[7] The dual hospital-staff and patient focus was narrowed to a pure consumer play.

The YC period brought institutional validation and forced the team to sharpen its growth thesis. By Demo Day in August 2016, PatientBank had already processed 10,000 records from 2,300 hospitals — a concrete operational proof point that distinguished it from most early-stage healthcare startups still in the concept phase.^[8]