This study aimed to explore and explain the experiential characteristics of human interaction with the emerging phenomenon of AI psychological chatbots. Utilizing a qualitative approach and a grounded theory design, this research conducted an in-depth analysis of user experiences and perceptions of "Mana," a Persian AI psychological chatbot. The study participants included 16 volunteers aged 20 to 38, selected through voluntary and theoretical sampling from student social media groups in Tehran. Data were collected between 2024 and 2025 via semi-structured face-to-face interviews, and the analysis was performed based on the Strauss and Corbin strategy through open, axial, and selective coding. The findings revealed two main themes: "Functional Judgment" and "Psychosocial Perception of the AI Therapist," along with their respective categories and subcategories. The results indicate that the participants' experience in interacting with AI is the product of a dynamic process and a constant interaction between prior expectations, perceived quasi-therapeutic agency, the quality of emotional relatability, and continuous evaluation of system performance. Ultimately, these findings suggest that the human-AI therapeutic relationship is a multifaceted construct shaped by both technical efficacy and subjective psychological interpretations.