Hospitals can't reliably get their own protocols to the bedside. Dr. Ido Zamberg of C8 Health on the 17-year evidence gap and fixing clinical knowledge chaos.
Episode Resources:
- Official Website for C8 Health
Every clinician knows the feeling: you need a protocol, and you have no idea where it lives. Maybe it's a PDF on a shared drive. Maybe it's an intranet page nobody's updated since 2019. Maybe it's a printout taped to a bulletin board with a phone number for a coordinator who left last year. Medicine is the most knowledge-intensive profession there is, and yet hospitals — good at care delivery, billing, regulatory compliance, and training residents — are not built to be knowledge organizations. The EMR doesn't solve it either. It's excellent at patient-level data and transactions, and largely silent on institutional know-how.
This week, we dig into C8 Health, an AI-powered best practices implementation platform tackling exactly that gap, and talk with co-founder and Chief Medical Officer Dr. Ido Zamberg. Zamberg's path is unusual: a senior developer at HP, Mercury Interactive, and Autodesk — where his specialty was optimizing access to information across large enterprises — who became an anesthesiologist and medical educator. He traces the idea back to a specific moment on geriatric rounds, watching colleagues each build their own ad hoc checklists for the same admission, all slightly differently, all because the hospital's actual protocols were impossible to find. The tool he built for his own department spread across five hospitals.
The conversation gets into where generative AI genuinely fits here — a retrieval-grounded assistant answering from vetted institutional content, with citations down to the paragraph, rather than a general-purpose chatbot improvising clinical advice. Zamberg also makes a pointed argument about quality improvement: that hospitals spend heavily on extracting quality data, then lock it in dashboards that individual providers never see, and pair it with one-time interventions that decay almost immediately. With new evidence taking an average of 17 years to reach practice, ERAS compliance hovering near 50% nationally, and more than 70% of quality interventions failing to sustain, the case that this is fundamentally a distribution problem is a hard one to argue with.