About

I built the data layer I kept wishing existed.

Verisim is synthetic healthcare data, built by an actuary and calibrated to public benchmarks, so you can build before you can get the real thing. Medicare Advantage is live today, with more coming on the same engine. It's early, and mostly a solo project.

Why it exists

Most people who need healthcare data can't get it.

Building anything in healthcare runs on data: claims, eligibility, labs, revenue. The real thing sits behind privacy law, licensing, and cost, so the people who could build with it spend their time waiting on access.

Startups stall before they can prototype, researchers can't test a hypothesis, and payer and provider teams can't benchmark, because the data stays out of reach.

Verisim is meant to fill that gap: synthetic data that's credible enough to build on, priced so a small team can start. Not a replacement for real claims, a way to get moving before you have them.

What goes into it

Built across the angles that matter.

I've worked across these areas over the last decade, and the data is built to hold up to each of them. I'm keeping my name off it for now.

Actuarial science

Calibrated to the benchmarks an actuary checks first: risk scores, MLR, utilization, and trend.

Data science & analytics

Built for the workflows the data lives in: risk models, backtests, reproducible pipelines.

Payer–provider ecosystem

Modeled on how claims actually move between plans and providers: coding, adjustments, lag, benefit phases.

Value-based care

Covers what VBC teams work in: attribution, episodes, population risk over time.

Healthcare operations

Grounded in how care is actually delivered and recorded, down to the operational detail.

Data engineering

Linked, longitudinal, and clean enough to load and join on the first try.

AI / ML

Generates the people and their journeys first, then every record they produce.

The actuarial side and the AI side pressure-test each other on every release, and the data is better for it.

What I believe

Four things I hold to.

Measured fidelity

Fidelity, checked and shown.

I check fidelity against published benchmarks and ship the audit with every dataset, citations included. If a number is off, you see it before you buy.

Honesty

Clear about what's synthetic.

I say plainly what's calibrated today and what's on the roadmap, where synthetic data is the right tool, and where it isn't.

Access

Access over gatekeeping.

Real claims sit behind privacy law, licensing, and six-figure cost. I price it so a small team can use it, not just the institutions that already could.

Lived experience

Built by someone who's hit the wall.

I've calibrated models, waited on data deals, and shipped without the data I needed. This is the thing I kept wishing existed.

Why now

Why I built it now.

AI got good enough in the last couple of years to generate synthetic people and journeys that hold together, not just plausible-looking rows.

Meanwhile the data-access problem keeps getting worse. The gap between who needs healthcare data and who can get it widens every year.

A new capability met an old problem, so I started building.

Get in touch

I'd love your feedback.

I especially want to hear from actuaries, data scientists, and anyone building on this, the people who'll poke holes in it and tell me where it falls short. Reach me at hello@verisimhealth.com.