Orinyx: The Hallucination Guard for Clinical AI

Hey everyone, Lex here, founder of Orinyx. I spent years in healthcare, first on the clinical side, then in health IT, and I kept watching the same thing happen: hospitals are putting AI in front of doctors faster than anyone can check whether the AI is right. These tools hallucinate. One example that stuck with me: an AI recommended daily aspirin to prevent heart attacks and called it guideline-consistent. That guideline was reversed in 2022. Nobody caught it, and there was no record the AI ever said it. Orinyx is the layer that catches that. It sits between any clinical AI tool and the patient record, checks what the AI recommends against approved clinical sources in real time before a doctor signs, and logs every check. We're starting with medication safety, where the risk is most concrete, and building out from there. The part I care most about: verification only counts if it's independent. An AI vendor checking its own work has the same blind spots that caused the error. Orinyx makes no clinical AI of its own. That separation is the whole point. We're pre-launch and in beta, backed by AWS Startups. If this is something you've run into or thought about, I'd really value your take. Follow here or join the waitlist and you'll see it the day we launch. Thanks for taking a look. Lex