SaMD — Software as a Medical Device — is the FDA's term for software that performs a medical function on its own, such as diagnosing, treating, mitigating, or preventing disease, without being part of a hardware medical device. The defining test is intended use, not technology. The same algorithm can land in three very different places depending on what it claims to do and how it is used: an unregulated general-wellness tool, an exempt clinical decision support (CDS) tool, or a regulated device such as a Class II product that requires premarket review before it can ship.
The distinction that decides this for AI features is whether the software supports the clinician or substitutes for them. Clinical decision support that surfaces information, references, or options — while leaving the clinician able to independently review the basis for the recommendation and reach their own conclusion — generally stays outside device regulation. Software that delivers a conclusion the clinician is expected to rely on, especially in a time-critical or non-reviewable situation, moves toward being a regulated device.
For a telemedicine AI team, the practical SaMD question to ask of every feature is simple: does it inform the clinician, or decide for them? Marketing language is part of the analysis — claims like "detects" or "diagnoses" can pull a feature across the line even if the engineering is modest. The common, expensive pitfall is building and launching a feature as a productivity tool, then describing it in sales materials as if it makes diagnoses, which can retroactively trigger device obligations. Decide the intended-use claim early, write it down, and keep the product and the marketing consistent with it.
