The claim doing the work here belongs to Faro. The company says a proprietary ontology converts the scientific, medical, regulatory and operational relationships inside a protocol into structured intent that software can reason over [7]. Document generation has spread across enterprise AI products generally [14], so the question is not whether Faro can draft protocol text. It is whether the model underneath holds. Outside corroboration is narrower than the framing suggests: the Clinical Data Interchange Standards Consortium's statement concerns the format in which Faro Study Designer authors a protocol, not what an agent later does with that model [9].
The reason a schema matters more than a prompt: protocol decisions are linked, so moving an endpoint can propagate into eligibility rules, site procedures, data collection, statistical analysis and submission materials [6]. A language model can write the paragraph describing that change. Configuring the systems that implement it needs something that remembers the links [6]. The most concrete evidence Faro already has some of that is the dullest line in its product list, automated electronic data capture builds [8], which is a write into an operational system rather than another document.
The money is less interesting than the signature. Eight investors are named against $37.3 million, an average under $5 million each [12], and no valuation was disclosed [3], so nothing in the announcement indicates what the equity cost. What Merck Global Health Innovation Fund adds is the buyer's side of the desk: its parent is the kind of regulated organisation where agents must clear procurement review, data-governance controls and clinical oversight before they go near a study [10]. Merck protocols also already appear in Scott Chetham's published work, a 2024 peer-reviewed paper on redesigning schedules of assessment that Faro cites on its own homepage [11]. The relationship predates the check.
Chetham's history reads as the problem statement. He led clinical research operations and data management at Verily Life Sciences [4], and General Catalyst credited him in 2023 with the development of more than 100 clinical-trial protocols across 54 programs with biopharma sponsors [5], an average of close to two protocols per program [13]. He and Ross Jaffe founded the company in 2019 on the premise that a protocol should be structured and machine-readable from its first draft, after watching timelines slip while the document stayed static and disconnected from the operational choices it governed [4][15].
The uncomfortable part of the structure is that the standards-aligned half can be inspected and the ontology cannot. If the agents survive a sponsor's oversight regime, the reason will sit in the layer no reviewer outside Faro can read [7][9].
