This project will develop datasets and software to deliver Functional Imagery Training (FIT) via mobile devices and, ultimately, via embodied agents (robots). As a person-centred intervention, FIT presents an interesting challenge: the practitioner or AI must tailor responses to the content of what the 'client' says, and guide mental imagery exercises based on that personal content. By building an interdisciplinary team of psychologists and machine learning experts this project will deliver real-world impact with broad implications for mental healthcare.