AI systems architect Alexandria Hamilton aims to create products that make a positive impact.
Hamilton has seven years of experience across the Electronic Health Records (EHR) sector, working at Oracle Cerner and a pediatric health system, where she led data and AI products. As noted on her LinkedIn, she built AI systems for healthcare and other regulated industries. She is also going back to school for nursing to become more clinically grounded.
“I really just want to deepen my clinician knowledge, continue to build things, and just build AI products that are for good,” she told AFROTECH™ in an interview.
Orinyx
Hamilton built Orinyx using her own API. It is a product described as an “independent safety layer for clinical AI” that she put out under her AI advisory and systems design practice, charmthirteen. It targets hospital systems and clinical medical information officers, built on her conviction that AI governance should not come last.
Hamilton said Orinyx will focus on medication errors and will check a hospital’s clinical AI results against Food and Drug Administration (FDA) labeling and clinician guidelines already used by clinicians. It can detect hallucinations and other potentially incorrect information.
“Part of the AI implementation of the education part gets lost. It just gets thrown out the window every single time. If you were to tell me that there is this product that can not only help me work closer with my patients but reduce my administrative burden, and be 80% accurate, I will take you up on that,” she noted. “I don’t blame anybody. I blame marketers, if anything, because they’re missing that part where they’re saying even though it’s extremely accurate… there’s still 20% to 10% of encounters that are getting lost in that.”
Hamilton added, “If there’s nothing that tells you to go back and check something, you may have missed something critical. It’s the patient at the end of the day that pays for that. … Clinicians are simply just practicing medicine. We’re all practitioners of AI. It’s still new. To have Orinyx live inside of your system, if it catches something, it doesn’t change anything. All it does is let the provider know, like, ‘Hey, go take a second look.’ Think of it like a spell check, except for it’s like for clinical documentation.”
Hamilton shares that she lost a family member due to a physician’s misdiagnosis. The family member was then told to return home and died there. Although this was before AI, Hamilton believes AI may lead to more incidents like this because practicing medicine already includes trial and error.
“It still kind of sticks with me … It just makes me think if something like this had existed during that, [how would it have impacted that process],” she reflected.
She notes there are currently two versions of the product. The first functions as a “testing ground” for AI agents, and the second lives inside the EHR and checks continuously.
Orinyx is currently in beta.
Separately, Hamilton told AFROTECH™ over email that she is enrolled in a pathology class studying diseases while looking at them differently — especially learning how some diseases can be missed in specific patient populations. This helps so her software can pick up on these patterns and what AI may miss.
“I’ve been building Orinyx from a product and engineering background, so I could think through how an AI hallucinates, but not really how disease shows up differently across patients. This class is giving me that lens for the first time, and it’s a big part of why I can actually name this gap now instead of just talking around it,” Hamilton said in a statement over email.
