The science behind the platform.
Today’s strongest AI systems are trained on data that is abundant on the internet: text, code, images. Biological data has no equivalent abundance — most of it is expensive to generate, siloed inside individual labs and clinics, or simply doesn’t exist yet, because no one has measured that system often enough to see how it changes.
Our research program exists to close that gap: to make the cost of a biological measurement low enough that repeated, longitudinal testing becomes normal rather than exceptional, and to build the modelling layer that turns that stream of measurements into something predictive.
That work spans lab automation, assay design, and the machine learning that reads the results — described in the four focus areas below, and put into practice in Robolab and Biopen.
Where the research effort is concentrated.
Closed-loop automation
Removing labour from the wet-lab loop — sample intake, prep, assay and readout run staffless, continuously, inside Robolab.
High-breadth biomarker panels
Reading hundreds of markers from a single sample, because the signal that predicts disease is in how markers move together, not any one value in isolation.
Digital twin modelling
Turning repeated, structured measurements into a living model of a biological system — one that updates with every new reading instead of going stale after a single test.
Field-grade pathogen detection
Compressing lab-grade pathogen assays into a single-motion, no-training device, so surveillance doesn't require a lab, a courier, or a wait.