In the field of psychology, the use of artificial intelligence-based depression detection chatbots is being employed in order to reduce the percentage of people with depression in the world. However, 3 out of 8 sessions conducted to these software products are not completed due to lack of confidence or self-esteem, trustworthiness and safety of the user. This is due to the disengagement of the chatbot in the conversation it holds with users and the color connotation employed. To avoid producing chatbots with this quality, this research presents a maturity model to evaluate these conversational agents, combining a questionnaire to measure the usability of mobile health applications, a performance metric to measure the chatbot's ability to detect depression, and a proposed category that evaluates whether appropriate depression detection tools were used when training the classification mode...