My Grandfather Died at 43.
An AI Caught What 12 Years of Doctors Missed.

I’ve been at sea for twenty years. Research vessels, commercial fishing in the Bering Sea, marine engineering operations across the Pacific and Atlantic. Over 3,500 days at sea. My lab reports are scattered across four countries in three languages, in different units, comprising different biomarkers. My health history existed in fragments that no single doctor had ever seen assembled.

I knew something was wrong. I’d known since 2012. I was tired in a way that sleep didn’t fix. My liver enzymes were always elevated. Doctors would note it, suggest diet and lifestyle changes, and move on. In Mexico, the labs came back in Spanish. In France, in French. In Slovenia, in Slovenian. Each doctor saw one set of numbers, in isolation, and pronounced them “not too concerning.”

Nobody saw the pattern. Because nobody could.

The Search

In early 2026, I was living on a sailboat in Mexico with my wife. I’d been chasing unexplained symptoms for over a decade — fatigue that sleep didn’t fix, liver enzymes that were always elevated, brain fog that wouldn’t lift. No doctor in any country had connected the dots. I was tired of accepting it.

I turned to Claude, Anthropic’s AI assistant, and did something unconventional: I asked it to help me design a comprehensive lab panel — the kind of thorough workup that no 15-minute doctor’s appointment would ever produce. Iron studies, full liver panel, metabolic markers, thyroid, hormones, inflammatory markers, immunoglobulins, vitamins. Everything.

I took that panel to a lab in San Carlos, Mexico. The results came back.

Marker	My Result	Normal Range
Ferritin	994 ng/mL	30–300
Iron Saturation	~90%	20–50%
ALT (liver enzyme)	127 U/L	7–56
GGT (liver enzyme)	113 U/L	8–61
Fasting Insulin	24.5 µIU/mL	2–15

Hereditary hemochromatosis. Confirmed. My iron was nearly double the danger threshold. My liver had been under assault for years. The same genetic condition that almost certainly killed my grandfather at 43 had been silently running in my blood, undetected, through every doctor’s visit, every port, every country.

I was 39. My grandfather made it to 43.

I caught it with four years to spare.

When I called my mother to tell her the diagnosis, she went quiet. Then she told me something I didn’t know: my sister had been diagnosed with the same condition. Our grandfather’s death at 43 — the jaundice, the liver failure that nobody could explain — suddenly made sense. It was hereditary. It had been in our blood the whole time. But I hadn’t found it because of my sister. I found it because an AI and a simple app I’d built helped me see what fourteen years of doctors across four countries had missed.

From Diagnosis to Dashboard

The diagnosis was the beginning, not the end. If it took an AI to find what 12 years of doctors missed, what else was hiding in the data? And how many other people were walking around with the same invisible time bomb?

I started building. Working with Claude, I designed an application that could do something no existing health tool could: unify lab results from any country, any language, any lab, into one coherent longitudinal health record.

I called it HealthAnchor.

Drop a PDF or photo of lab results in any language, and it extracts every value, converts units, maps multilingual biomarker names to standardized keys, and flags anything abnormal. It recognizes over 130 biomarkers and their names in Spanish, French, Slovenian, German, Russian, Tagalog, and more.

Then I loaded all my data — six lab panels spanning four years and four countries — and asked the AI for a comprehensive analysis. It calculated clinical ratios no individual doctor had ever computed for me. It correlated my journal entries about symptoms with the pathophysiology of iron overload. It connected my chronic fatigue to a two-decade pattern of occupational trauma. It found things hiding between the numbers, in the spaces between snapshot appointments in different countries with different doctors who never talked to each other.

It did in one conversation what no healthcare system on earth had done in twelve years.

The People Healthcare Forgets

There are 1.9 million seafarers on commercial vessels right now. Millions more work offshore. Over 40 million digital nomads. Tens of millions of expatriates. They all share the same problem: they fall through every crack in every healthcare system on earth.

A Filipino sailor gets blood work done in Manila, sees a doctor in Rotterdam, fills a prescription in Singapore, and has an emergency in Houston. Each provider sees one snapshot. Nobody connects the dots. Lab reports arrive in different languages, different units, different formats. There is no continuity. There is no memory.

What I experienced with HealthAnchor is what high-end concierge medicine provides to wealthy patients — teams of specialists spending hours assembling a complete health picture. Those programs cost $10,000 to $50,000 per year. They are inaccessible to the populations who need them most.

A fishing boat captain should get the same depth of analysis that a Fortune 500 CEO gets at Mayo Clinic.

The technology to do this exists now. The missing piece wasn’t the technology — it was the application layer that puts it in the hands of regular people, with their data under their control. That’s what HealthAnchor is.

Private by Architecture

Your health data stays on your device. Nothing is transmitted to HealthAnchor — ever. AI queries go directly from your device to the AI provider using your own API key. We’re not a covered entity under HIPAA because we never touch your data. Not by legal loophole — by architecture. We can’t see what we never receive.

Built From Necessity

HealthAnchor wasn’t built in a lab or a boardroom. It was built on a sailboat in Mexico by a guy who’d been failed by fragmented healthcare systems for over a decade. It was built because a grandfather died at 43 from something a simple blood test could have caught. It was built because the man who inherited that same condition almost ran out of time too.

Nobody should have to cross four countries and twelve years of searching before an answer appears.

Now, they don’t have to.