Nicholas Archer joined AdventHealth in Orlando in 2004 as a financial analyst. Over the next fourteen years he held every job between that one and CEO of Project Fulcrum, the system's innovation arm. He did not learn how a 550-facility health system works from a consulting engagement. He learned it from the inside, year by year, floor by floor.
Hippocratic AI recruited him for its founding go-to-market team. He helped build the commercial playbook as the company grew from a fifty-million-dollar seed round to a $1.6 billion Series B, deploying AI agents into clinical workflows at a pace that exposed exactly where the governance breaks. He is now VP of Innovation Ventures at Cincinnati Children's Hospital, running commercialization and tech transfer at one of the country's top pediatric research institutions.
But the work he founded CARIVIVA to do is the work that brought him to CANONIC. When a patient in Jamaica or Trinidad or Guyana speaks to an AI clinical documentation system, every major commercial transcription platform fails. Not on accent — on clinical significance. A drug name becomes a different drug name. A symptom description becomes a different symptom. The errors are silent and they are dangerous.
Archer built the Bridge engine to fix this: 196 scripted clinical recordings across three Caribbean territories, a phonological coverage map, a dialect-to-clinical-English glossary, and a drug-name passthrough pipeline. In testing, Bridge corrected thirty-six clinically significant errors with zero hallucinations. Three commercial platforms and the bare foundation model scored zero out of thirty-six. The result is independent corroboration of the CANONIC thesis: when the pipeline is governed, the outputs are trustworthy. When it isn't, they aren't.
36/36
clinical corrections
0
hallucinations
196
recordings