An international consortium including the Garvan Institute developed AAnet, an artificial neural network leveraging single-cell gene expression data to resolve the diverse cellular populations within tumors at unprecedented resolution. This AI framework detects continuous gene expression patterns reflecting tumor heterogeneity that traditional clustering methods miss. The approach enables comprehensive characterization of cell subtypes driving therapeutic resistance and progression, advancing precision oncology by tailoring treatments to target all malignant components within a tumor, particularly in aggressive cancers like triple-negative breast cancer.