When I started Zupermicro, the most advanced piece of technology in a typical records office was a fax machine with a jam-detection light. Fourteen years later, I spend a lot of my time thinking about what happens when a clinician can simply ask for the chart they need, in plain English, and get it in seconds.
I am genuinely optimistic about this. Not because the technology is magic (it isn’t), but because the problem is so concrete. A nurse in Boise should not have to wait three days to learn that a patient’s allergy list changed in Spokane. That delay is not a law of nature. It is a solvable engineering problem.
This year we opened our clinical records gateway to approved AI assistants. It speaks the Model Context Protocol, the emerging standard that lets assistants like Claude and ChatGPT use tools safely. An assistant can search our master patient index by name, verify a patient by name and date of birth, and open a chart with the same break-the-glass safeguards a human would face. Our developers wrote up the details on the developer page.
The numbers so far are encouraging. Average time-to-chart for our pilot clinics fell from 41 hours to under four minutes. The thing I did not expect: the clerks love it most. They spend less time hunting and more time on the hard cases that actually need a human.
There are real risks, and we take them seriously. Every disclosure is logged for the accounting-of-disclosures report, and every access is audited. But I would rather build this carefully than pretend the fax machine is the best we can do. If you run a clinic and want to be part of the next pilot, I would love to hear from you.