The idea of trans-diagnostic and modular treatments highlighted in this Editorial points the way beyond self-contained ‘branded’ therapies to a more flexible intervention science responsive to advances in basic science and relevant also to personalisation of therapies. For such approaches to be rigorous, there need to be empirically identified trans-diagnostic target mechanisms. The Editorial reviews two studies of potential mechanisms of this kind in the developmental neurobiology of stress response in children at risk from prenatal maternal stress and post natal neglect and maltreatment. A number of treatment trials are also reviewed illustrating the common method of digitally-aided therapy; a common theme is the difficult in sustaining adherence to such treatment delivery methods. Finally it reviews a useful independent investigation of a commonly used autism screening instrument, suggesting improvements for greater specificity.