Real-World Outcomes of AI-Assisted Psychotherapy: A Mixed-Methods Study of the Psyhelp Premium User Cohort

The present mixed-methods study examines outcomes among premium users of the AI-assisted psychotherapy platform Psyhelp. From a pool of approximately 40,000 users, 10% (≈4,000) were premium subscribers (i.e. full access to journaling, interventions, self-monitoring). Using stratified random sampling via G*Power calculations, n = 200 premium users were selected for quantitative analysis of symptom change (depression, anxiety, emotion regulation, self-esteem, attachment/dependency) and engagement metrics (journals, minutes of use, intervention modules). Qualitative in-depth data (n = 20) explored user experiences, perceived mechanisms of change, and contextual factors influencing engagement. Results show statistically significant pre-to-post reductions in symptom severity (paired t tests, large effect sizes) with higher engagement associated with greater improvement. Qualitative themes include “awareness to action,” “digital habit formation,” and “relationship to the algorithm.” Discussion focuses on implications for digital therapy scalability, human-AI collaboration in mental health, and directions for future research. Limitations include sample representativeness, self-report bias, and lack of long-term follow-up. This study contributes to the growing evidence base for AI-augmented psychotherapy in real-world settings.