Background and aim. Mentalization-Based Therapy (MBT) is a manualized psychodynamic treatment that fosters reflective functioning and affect regulation. A key clinical challenge is sustaining patients' mentalization capacity in daily life, particularly in emotionally charged contexts where breakdowns are most likely to occur. The aim of this work is to critically describe an AI-assisted reflective tool (Mentora) designed to deliver ecological momentary interventions grounded in MBT principles and strengthen patients' capacity for mentalization outside sessions.
Methods and Results. Mentora operationalises MBT techniques in everyday contexts through: (1) reflective prompts to support self-other understanding, (2) feedback on recurrent emotional-linguistic patterns, (3) mood and event journaling integrated with daily routines, and (4) psychoeducational micro-interventions aligned with the MBT manual. Importantly, the system is configured by the therapist, who sets clinical objectives and calibrates functions, while patients regulate its intensity and scope. Mentora guarantees rigorous privacy, granting patients full control over what is shared with the therapist. Its design emphasises personalisation, scalability, and therapist-patient co-regulation. To ground the model, we conducted a systematic comparison of four large language models to establish criteria of adequacy for functioning as trainers of mentalization. In addition, simulated cases of Mentora use, developed under the supervision of a psychotherapist, show how reflective material can emerge between sessions and later be elaborated upon in therapy.
Discussion. This study demonstrates how ecological digital tools can expand the therapeutic scope of MBT while maintaining the centrality of the therapeutic relationship. Future directions include integrating wearable devices for psychophysiological monitoring (e.g., heart rate variability, sleep quality, stress markers), thereby enriching ecological momentary interventions and tailoring them to patients' mentalizing capacities and well-being.
Conclusions. Mentora illustrates how a manualized, theory-driven approach can guide the ethical and effective use of AI in psychotherapy, enhancing mentalization in real-world contexts while safeguarding patient agency and privacy.