For four years, the hottest idea in health tech was putting your medical records on a blockchain, and I was its traveling salesman. Hashed ran Distributed Health, a conference that drew over a thousand people from around the world, we held meetups on four continents, and I even co-authored a book on it. Of every use case we swore blockchain would fix, one drew more attention than all the rest combined: give people ownership of their medical records.
We never launched anything in this space during those days because we couldn't honestly validate it. There was no business model we believed in. Selling data to pharma was the obvious play, but we didn't think that was a business worth building. I still don't think that’s a viable approach. We had no real demand either, outside a niche of patients carrying a folder between offices because someone in the family was sick and the system was failing them. And we couldn't get the data. The clinical record sat locked inside systems that didn't talk to each other, held tight by people with no reason to let go. Plus, the 80% of your health that happens outside the clinic wasn't reachable at all.
We weren't the first to hit this wall. Google Health launched in 2008 and was gone by 2012. Microsoft HealthVault ran twelve years and shut off the lights in 2019. Dossia had Walmart, Intel, and BP behind it and still faded into irrelevance. All of them were free or employer-funded, all of them assumed the record would carry enough engagement or savings to pay for itself, and all are now bodies littering the road to a PHR.
The issues were not technical. When an idea fails that many times across that many stacks, maybe we should stop blaming the stack. The real problem was simpler. The record asked you to become a records clerk: collect your data, reconcile it, keep it current, maintain a filing cabinet for your own body, and someday, when you're sick and scared, it pays off. We called it empowerment. It felt like homework. Clayton Christensen spent a career on the idea that people don't buy products, they hire them to do a job. Nobody ever wanted to hire a health app. They wanted the job done. Healthcare has never had an app shortage. It has an effort surplus.
The ground has shifted since. Artificial intelligence is great at finding and organizing data and is the unlock for widespread value creation. The Cures Act rules finally have teeth. The exchange rails exist. Translation engines can turn one system's mess into another system’s input. The environmental, social, and personal data that comprises 80% of our well-being is finally reachable. And we have visibility into new business models that don't require selling out the patient. It seems the raw material and precious metals for the PHR are more available than they have ever been. This is starting to look like execution risk to me.
Players big and small are rushing back in search of the holy grail. In the past year OpenAI, Google, and Apple all shipped consumer health product. Anthropic's Claude for Healthcare allows users and medical professionals to securely connect, summarize, and analyze outside personal medical records and health data directly within the platform. The intelligence is real. A model that reads your labs and reasons across them is not the healthcare scrapbook people tried to ship a decade ago.
To be clear, challenges remain. These systems still begin the same: connect, link, import, upload. They made the assistant smarter and left much of the clerical work with the consumer. And to get the intelligence, you hand the most sensitive data you own to the largest data-collecting companies that have ever existed. It’s like asking Freddy Krueger to babysit your kids. “Secure,” “private,” “self-sovereign…” these words that meant everything in the world I came from are not what you get currently.
So the question was never whether patient-controlled health data is a good idea. It always was. The question is whether now is finally the window.
For the first time, I think it might be. The pieces are on the table: we can assemble the data, agents can do a variety of jobs, and a viable, exciting business model is coming into focus.
Plenty of challenges remain. The data still isn't yet liquid enough. Distribution is always hard to figure out. But risk of success is starting to tilt more toward execution. And the thing that always killed this was never technical… it was creating something people wanted.
