UpDoc, Inc.

See attached file(s)

UpDoc Public Commentary for FDA RFI
for docket FDA-2026-N-7874

UpDoc Comments on FDA Considerations for Generative AI-Enabled Medical Devices

UpDoc appreciates FDA's efforts to engage industry and other stakeholders as it considers an
appropriate regulatory framework for generative and agentic AI-enabled medical devices. As a
developer of FDA-cleared software and clinical AI technology deployed within health systems,
we have a direct interest in seeing that framework develop in a way that supports safe, wellgoverned innovation.
UpDoc believes Clinical AI should augment physicians and operate within appropriate clinical
boundaries and governance. As Clinical AI progresses from bounded functions toward
increasingly sophisticated systems capable of reasoning across complex clinical information
and performing broader clinical functions, safety, accountability, auditability, appropriate human
oversight, and integration into existing care delivery systems remain essential.
We recognize that emerging technologies will often advance more quickly than established
regulatory frameworks and evidence-generation methods can evolve. Addressing that gap is a
shared responsibility: developers have an important role in helping regulators understand new
technologies, their risks, and the controls available to address those risks, while regulators can
provide increasing clarity as experience and evidence develop. That shared responsibility also
includes developing clear and consistent terminology for describing clinical functions, levels of
autonomy, human oversight, and associated risks.
One area where continued clarity would be particularly valuable is the boundary between AI that
supports or communicates a clinician's existing treatment plan and AI that performs clinical
functions requiring independent clinical judgment. As generative and agentic capabilities
advance along this continuum, having a clear, consistent way to describe where that boundary
sits, and how it maps to existing device frameworks, would help developers anticipate regulatory
expectations earlier in development rather than resolving that question product by product.
As FDA develops its approach, we believe it would also be helpful to continue clarifying how the
specific risks associated with generative and agentic AI functions relate to corresponding
controls and evidence expectations. Evidence expectations should be proportional to the risks
presented by the intended clinical function and should account for the extent to which those
risks are addressed through design controls, software verification and validation, human factors,
clinical safeguards, human oversight, monitoring, and other established risk-management
measures. Over time, AI itself can also help address this challenge by making evidence
generation, monitoring, and regulatory evaluation more efficient, while preserving the rigor
necessary for sound regulatory decision-making.
Industry and regulators share the same objective: enabling beneficial technologies to reach
patients efficiently while maintaining appropriate standards for safety and effectiveness. We
welcome the opportunity to contribute to that shared effort.

UpDoc, Inc. - Confidential and Proprietary Page 1 of 1