A new cell profiling technology combines high throughput imaging and machine learning to provide a rapid, cost-effective way to determine how specific compounds act to destroy the bacterium that causes tuberculosis. It could speed discovery of anti-TB drugs and be applied to other pathogens.