Every manufacturing category has made the same leap from art to engineering. Silicon got EDA and fabs. Metals and polymers got CAD and CAE. Biomanufacturing never got that layer — it still runs on one-off hero projects and bespoke protocols. Differential is building the design surface for cells: a model of how cells behave, accurate enough to plan against and to learn from everything built on top of it.

Industrial fermentation data is fragmented. It’s waiting on an ImageNet or PDB moment, and it won't get one by accident. It has to be generated on purpose, through a system where every experiment and deployment makes the platform better. Differential has built a Biological Data Factory where robots generate wet-lab data and run a closed-loop AI system to choose the next experiments to create optimal predictability. The system is automated, connected, and continuously learning how to optimise cellular behaviour and commercial scaling, while producing a foundational data asset that cannot be scraped, licensed, or inferred.

Martin was a founding engineer at Recogni, where a model only mattered if it could steer autonomous machines through the messiness of the physical world. Chris is a biologist who spent a decade selling complex data platforms to conservative industries, learning to communicate abstract capability into something legacy players would buy and integrate. Together, they’ve built Differential on an unwavering conviction: biomanufacturing becomes reliable through a system that makes predictability the product.

The end state is biology you can specify like code. Describe what you want, and the constraints and the system returns a strain+media+process recipe in a single shot. The weird, complex, multi-step products that cells are uniquely good at producing finally become possible to attempt and plausible to scale.