Cognitive Behavioral Therapy (CBT) is a widely recognized and practical approach for treating various mental health issues, including depression, anxiety, and stress-related disorders. By focusing on identifying and restructuring negative thought patterns and behaviors, CBT empowers individuals to develop healthier coping mechanisms. Traditionally administered through face-to-face sessions with trained therapists, CBT has proven efficacy but faces barriers that limit its accessibility, such as limited availability of professionals, high costs, and geographical constraints. In this chapter, we address these limitations by presenting an adaptive Embodied Conversational Agent (ECA) that adjusts its questioning strategy based on the user’s real-time psychological distress. By integrating a distress detection model analyzing users’ speech transcriptions, the ECA personalizes the therapeutic interaction to ensure that interventions are neither overwhelming nor insufficient. We conducted an experiment with 49 participants, comparing an adaptive condition—where the number of Socratic questions was modified according to detected distress—with a random condition. Results demonstrated that the adaptive approach yielded superior outcomes in reducing psychological distress and negative moods. These findings highlight the potential of ECAs to offer CBT sessions for a broader population.