In this paper, I investigate a group of student-teachers' insights into an artificial intelligence (AI)-powered automatic question-detection tool designed for reflective practice: Question Bot (QBot). Six student-teachers (STs), who were enrolled in a secondary school teacher education program in Sweden, volunteered to use QBot to analyze their own classroom discourse and to reflect on their teaching. Findings based on the thematic analysis of interviews show that QBot promotes noticing and pattern identification. Furthermore, it facilitates reflection on and for action. STs reported that AI assistance is desirable, and some even argued that AI input should precede human feedback in post-observation feedback sessions. STs also reported a need to develop other AI features for detecting gaze, gestures, silence, feedback, voice pitch, and code-switching.