HealthVerity

A 340-million-person map of American health, with the names torn off.

Somewhere in a cloud region, a request runs. A life sciences researcher wants every American who started a GLP-1 drug, then later showed up in a lab result, then filed a medical claim. She gets 19 million matched journeys back. She never learns a single name. She never needs to.

That trade - total connection, zero identity - is the whole business of HealthVerity. The company sits between more than thirty national data suppliers and the pharmaceutical companies, insurers, and government agencies that want to study them. It bills its Marketplace as the nation's largest privacy-protected healthcare and consumer data ecosystem: 340 million de-identified individuals, stitched into longitudinal patient journeys, all without violating a privacy rule.

It is not a flashy product. There is no app to download, no patient who signs up. HealthVerity is plumbing - the kind of infrastructure that is invisible right up until you realize an enormous amount of modern medicine runs through it.

Healthcare's data wasn't missing. It was unmatchable.

For decades the industry told itself a comforting story: if only there were more data, the breakthroughs would come. The story was wrong. The data already existed - in pharmacy systems, claims warehouses, lab databases, electronic medical records. The problem was that none of it could be safely connected. A pharmacy record and a hospital record might belong to the same person, but proving it meant touching the one thing the law forbids you to touch: the patient's identity.

So researchers worked with fragments. A prescription here, an outcome there, never the line between them. You could see that a drug was filled. You could not reliably see what happened next. The patient journey - the actual story medicine cares about - kept breaking apart at exactly the moments that mattered.

This is the tension HealthVerity exists to resolve, and it runs through everything that follows: how do you link a person's records to each other without ever linking them back to the person? Get it wrong in one direction and the data is useless. Get it wrong in the other and it's a privacy breach. The space in between is narrow. HealthVerity decided to live there.

Two men named Andrew, and a problem one of them had solved before.

In 2014, Andrew Kress and Andrew Goldberg started HealthVerity. Kress was not new to the idea. He had already built one of the largest linked de-identified healthcare data assets in the country at SDI Health, growing it to more than 600 employees before IMS Health acquired it in 2011. He had seen what worked, and - more usefully - what was still broken.

Kress is, on paper, an unlikely data-infrastructure founder. He holds a B.A. in English from Yale. Perhaps that is the point: the company's product is essentially a question of identity and meaning - what counts as the same person, told across a hundred disconnected sources.

The bet was specific. Not "collect more data." Instead: build the matching layer first - a privacy-preserving way to assign a persistent token to an individual, so any record about them could be synchronized with any other, while the identity stayed sealed off. Solve identity and privacy as one problem, and the data ecosystem would assemble itself on top.

Caption: Yes, both founders are named Andrew. No, they will not be changing it for your convenience.

Four boring words doing very un-boring work: Identity, Privacy, Governance, Exchange.

HealthVerity calls its framework IPGE, and it is less a product than a discipline. Identity resolves which records belong to the same person. Privacy keeps that person de-identified throughout. Governance enforces who is allowed to see what. Exchange moves the result to whoever licensed it. The clever part is that all four run together - you cannot pull one out without the system collapsing into either uselessness or a lawsuit.

Privacy and progress were supposed to be a trade-off. HealthVerity refused the trade.

The mission, in the company's own words, is to power "a high-governance, privacy-compliant way for our partners to synchronize and exchange real-world data across the broadest ecosystem, enabling longitudinal patient journeys, frictionless access to data and better patient outcomes." Strip the jargon and it is a refusal: the refusal to accept that you must choose between protecting patients and learning from them.

That refusal has a competitive edge to it. HealthVerity operates in a crowded field - IQVIA, Komodo Health, Datavant, TriNetX and others all want to own the patient-tokenization layer. What HealthVerity sells is not just data but trust as infrastructure: governance baked so deeply into the product that compliance is the feature, not the friction.

The model that's winning is the one nobody can see.

Real-world evidence is projected to roughly double as a market between 2025 and 2030. Regulators increasingly accept it. AI models are hungry for exactly the kind of clean, linked, governed data HealthVerity assembles - and an AI trained on improperly sourced health data is a liability, not an asset. The companies that can prove their data is both connected and consented-clean will have something rarer than scale. They will have permission.

Which brings the story back to where it started. That researcher, running her query, pulling 19 million matched patient journeys out of the cloud, learning everything about a population and nothing about a person. A decade ago that query was impossible - not technically, but legally and ethically. The records existed; the bridge between them did not.

HealthVerity built the bridge. It is unglamorous, mostly invisible, and quietly load-bearing for a growing share of how new medicine gets studied. The names are still torn off. That was always the point.