Robotic-assisted surgery (RAS) involves training processes and challenges that differ from open or laparoscopic surgery, particularly regarding the possibilities of observation and embodied guidance. The video recording and the dual-console system creates a potential opportunity for participation. Our research, conducted within the department of visceral surgery of a big Swiss, public, academic hospital, uses a methodology based on the co-analysis of video recordings with surgeons in self-confrontation interviews, to investigate the teaching activity of the lead surgeon supervising a surgeon in training at the dual console. Three short sequences have been selected for the paper. Our analysis highlights the skills-in-construction of the surgeon in training regarding communication with the operating team, fluency of working with three hands, and awareness of the whole operating site. It also shows the divergent necessities of enabling verbalization for professional training, while ensuring a quiet and efficient environment for medical performance. To balance these requirements, we argue that dedicated briefing and debriefing sessions may be particularly effective; we also suggest that the self-confrontation video technique may be valuable to support the verbalization on both the mentor's and the trainee's side during such debriefing, and to enhance the mentor's reflexivity regarding didactic choices.