Educational institutions generally face limited resources to address the growing demand for students’ emotional health support. This work presents the design and development of a conversational agent aimed at providing initial emotional support in university settings. The system offers personalized and anonymous assistance and, upon detecting risk situations—such as suicidal ideation, maltreatment, or abuse—requests user consent to collect data and refer the student in a timely manner to psychopedagogical support services. Additionally, it integrates with a web application that enables professionals to access the student’s interaction history for proper contextualization. For validation, two phases of testing were conducted: one involving psychopedagogical specialists and another with students, resulting in 185 interactions and 14 alerts. Preliminary results suggest that the agent can identify risk signals and respond in a differentiated manner, positioning it as a viable tool for emotional support in educational environments.