Dr Kola Tytler (MBBS MBA MRCGP) | 15 July 2026 | 10 min read
Neko's most visible product is the physical assessment: a 60-minute visit using proprietary sensors to capture skin imaging, cardiovascular and circulatory measurements, and blood biomarkers, reviewed with a clinician on the day. But the strategically important asset behind that visit may not be the scanner itself. It may be the proprietary, repeatedly collected multimodal dataset the scanner exists to generate, which is a materially harder thing for a competitor to reproduce than the hardware alone.
What Neko actually collects
Each assessment captures high-resolution skin imaging across the body, cardiovascular and circulatory measurements, blood biomarkers covering metabolic and pre-diabetes risk, body composition data, and, since a recent product update, clinician-reviewed information imported from members' own wearable devices. The company describes each scan as generating millions of individual health data points. Layered on top of a member base with a reported 75 percent rebooking rate, that is not a single dataset but a growing longitudinal one, repeated at roughly annual intervals for a meaningful share of the company's user base.
Why proprietary data is becoming the harder moat to build
General-purpose AI capability, the underlying models used to process images, signals and text, is becoming more accessible and more commoditised by the month, available to well-resourced competitors and even to well-funded new entrants relatively quickly. Proprietary, consistently collected clinical data is a different kind of asset entirely. It cannot be licensed from a foundation model provider or replicated by better prompting. Building it requires years of consistent data capture, a working clinical protocol, a growing user base willing to return, and a stable sensing hardware platform that produces comparable measurements over time. Neko's vertical control over its sensing hardware, protocol, data format, clinician review process and patient interface means it is, in principle, positioned to control most of that pipeline itself, rather than depending on third-party device suppliers whose roadmaps it cannot influence.
Why repeated assessments compound in value
A single scan can only compare an individual against population reference ranges, which is useful but relatively coarse. A second scan, a year later, using the same protocol and broadly comparable hardware, allows a genuinely different kind of comparison: an individual against their own prior baseline. Subtle change over time, a slowly rising blood pressure trend, a gradually shifting metabolic marker, may eventually prove more clinically useful than a single abnormal threshold crossed once, because it can flag a trajectory before any single measurement looks alarming in isolation. Annual reassessment therefore does two things simultaneously for Neko: it generates recurring subscription-like revenue, and it generates the specific kind of repeated, structured training data that a one-off assessment company, however large its total user base, cannot easily produce.
Comparing the different data moats across the category
Each company competing in this space is, in effect, building a different kind of data asset. Neko is accumulating proprietary physiological and imaging data captured on its own hardware. Prenuvo and Ezra are accumulating imaging data specifically, largely on third-party MRI infrastructure. Function Health is accumulating repeated biomarker data across a very large member base at high testing frequency. Q Bio is combining imaging, genetic and wearable data within a single longitudinal interface. OpenEvidence, in a different part of the stack entirely, is accumulating medical literature usage patterns and clinical query data from a very large physician user base. Ambient scribe companies are accumulating consultation audio, transcripts and clinical workflow data. iatroX, in its own niche, is accumulating clinician information-seeking behaviour and evidence-retrieval patterns. None of these datasets is directly comparable to another, but each represents a genuine attempt to build the kind of asset that is hard to buy and hard to copy.
The open question this raises
The central strategic question for the whole category, not just Neko, is whether the winning medical AI companies over the next five years will be the ones with the best underlying models, which is likely to become a shrinking source of advantage as model capability commoditises, or the ones with the most defensible, clinically meaningful, genuinely proprietary systems for generating data in the first place. Neko's $700 million round is, in that light, best read as a bet on the second answer rather than the first.