2026-09-30 · 5 min read

I Built My Own Longevity Clinic

The concierge longevity clinic is one of the great luxury bundles of the moment. Pay somewhere north of five thousand dollars a year and you get four things: a comprehensive blood panel with a biological-age readout, a coach who keeps you accountable, a supplement protocol, and a long-range plan. It's a good bundle. It's also, underneath, software and judgment wearing an expensive coat.

So I unbundled it. One at a time, I built four tools, each taking one department of that clinic and rebuilding it as software I own: running on my own data, priced as one payment instead of a monthly rent, and with every number auditable instead of trusted on faith.

I'm not a doctor. I'm someone who got tired of renting access to his own body.

The clinic, department by department

The lab: LabViz. A blood draw is a single frame of a movie. The clinic's value isn't the draw; it's the trend across years, and your lab portal shows you one frame against ranges built for "not sick" rather than "aging well." LabViz takes bloodwork from any source, trends every marker over time, and computes a biological age from it. The signal is in the trend, and the trend lives nowhere.

The coach: Hugh. The accountability part of a clinic is presence: someone who remembers your reasons and pushes at the right moment. Hugh is a single-thread AI coach built to be a hard-ass when it matters. It quotes your own stated reasons back to you when your motivation dips, in the moment of a slip, never to gloat over a win. What software adds is availability at 9pm on a bad day. What it can't be is a person you'd hate to let down.

The pharmacy: PillPal. The protocol is the pills, the consistency, and the quietly uncomfortable question of what it all costs. PillPal shows what to take right now, tracks the streak honestly, and adds up the real monthly spend from per-dose costs: the number you avoid looking at, in one place.

The plan: Beyond. Most longevity advice starts from what's easy to measure and arrives as an uncitable firehose. Beyond starts from the other end, the life you want to be living at ninety, picks the capabilities you refuse to lose, does the decline math, and builds a five-pillar plan around them.

Four apps. One reassembled clinic, owned by the patient.

The layer below: owning your data means doing the math

This is the part I'm most willing to be judged on.

The biological-age number in LabViz is real math, not a vibe. It implements PhenoAge, the Levine 2018 model, with the published mortality coefficients, plus a second method (Klemera-Doubal) built on NHANES coefficients, and it does the unit conversions so the inputs mean what the papers meant. Every "wellness app" will hand you a made-up score. Almost none will compute a number a researcher would recognize, from your own labs, with the arithmetic exposed. That difference is the difference between a toy and an instrument.

PillPal lives or dies on a date bug most trackers get wrong. Take a supplement at 9pm in California, let the app record it against server time in UTC, and it lands on tomorrow. Your streak is quietly wrong forever. So every check-off is recorded against your own local calendar day, and the adherence math treats a day with nothing due as neutral rather than a miss, so it won't break your streak for a today you haven't finished. It's unglamorous. It's also the entire product working correctly.

LabViz has to read a lab it's never seen. Results arrive as a PDF from one lab, a portal export from another, a photo of a printout from a third. So it uses vision OCR for paper reports, a model fallback when the rule-based column matcher can't map an odd format, and it can stream a half-gigabyte Apple Health export without falling over: 164 canonical markers with their aliases underneath.

I didn't derive the Levine model; researchers did, and I read their work. I didn't invent the accounting; I knew, as someone who has tracked his own health, that a streak broken by a timezone is a lie. The judgment (which model, which markers, that a 9pm dose is a today dose) was mine. The implementation was cheap.

One payment, not a monthly rent

There's a conviction under all four, and it's a business rule as much as a belief: I won't build, or pay for, a subscription that charges monthly for access to your own health data. Beyond's coach and LabViz's advanced features unlock for a single payment, and then they're yours. The category runs on subscriptions that quietly bill you for motivation you stopped using in March. A tool for your own body should earn one honest payment and then get out of the way.

Where the seams show

Health software that overclaims isn't just tacky. It's unsafe. So here's the audit, including the parts that embarrass me.

• Beyond's coach doesn't cite yet. The library it's meant to draw on is real and clinician-reviewed, but the retrieval layer that would let the coach quote it isn't wired in. Its instructions tell it the library exists but isn't indexed, and never to fabricate a citation. That's the right failure mode, refuse to invent sources, but the honest headline today is "a coach that won't lie about its sources," not "a coach that cites them."
• Beyond's coach mostly runs on my machine. The deployed version leans on a development auth instance, so the streaming coach works on localhost more reliably than in the wild. A prototype I use, not a product you can.
• Hugh is a local-storage prototype. Its memory is the browser. The accounts, server-side history and reminders that would make it a real coach are planned, not shipped.
• These are tools for an audience of one. I built them for myself. I can't tell you anyone else uses them, and I'd rather not claim daily use of all four when the truth is that some have earned their place and some are still waiting to.

What would make this whole project a bad idea? If I trusted my own instrument more than it deserves. A biological-age number I computed is still a model, not a verdict, and a supplement streak is not the same as being healthier. The tools are worth building only if they sharpen my questions for an actual doctor, not if they let me play one. Owning the math is not the same as understanding the body. It's a start.

Close

The clinic was never magic. It was a lab report, a coach's memory, a pill organizer and a plan: bundled, marked up, and rented back to you a month at a time. I took the bundle apart and rebuilt it as four things I own, on my own data.

Build the instrument, then go see the doctor. Own the numbers, and use them to ask better questions.