Emerging infections like Ebola, HIV, flu and Covid‑19 often arise when animal pathogens cross into humans. At University of Cambridge, Professor Clare Bryant is applying DeepMind’s Co‑Scientist platform to pinpoint the molecular switches that trigger severe outcomes such as sepsis. The AI ingests grant summaries, spits out ranked hypotheses, and suggests experiments that align with existing lab work.

Bryant first fed the system a brief proposal on avian‑human flu transmission; the tool returned familiar ideas alongside unexpected leads. When the full, funded grant entered the engine, Co‑Scientist flagged a previously ignored protein linked to several signaling pathways. The loop also flagged off‑target effects. That “a‑ha” moment prompted Bryant to upload confidential unpublished data, iteratively narrowing hypotheses from candidate proteins to exact amino‑acid residues.

Armed with those precise targets, Bryant’s team is engineering cell lines that carry the identified mutations, a task that would normally consume two to three years of bench work. Early estimates suggest the Co‑Scientist‑driven workflow could deliver results within six months, dramatically accelerating discovery of interventions that block zoonotic jumps before they cause human outbreaks.