Purpose: Longitudinal coaching programs are increasingly common, yet student experience often diverges from program intent. At Geisel School of Medicine, a mandatory program pairs each matriculating student longitudinally with a physician faculty coach, most without formal coaching training. Using a framework that conceptualizes student-faculty relationships as encompassing reflective coaching, advising, mentoring, psychosocial support, professional identity formation, and/or community, this study examined what students found most beneficial through both inductive and deductive qualitative analysis, and compared human analysis to AI analysis using Claude and MAXQDA. Methods: Open-ended responses from 262 M1–M3 students during 2021–22 were analyzed deductively against the six program frameworks and inductively using Braun and Clarke's reflexive thematic analysis. Both AI platforms received identical prompts for each task. Results: Deductive analysis showed psychosocial support, mentoring, and advising as the primary benefits, with minimal attribution to reflective coaching, professional identity formation, and community. Inductive analysis produced comparable but richer findings, identifying the program as a "safe space" and identifying negative themes, particularly around mandatory group sessions and concerns about coach preparation, that deductive coding did not capture. Differences were observed between human and AI analyses, with smaller differences noted between the AI platforms than between either AI and the human analysis. Conclusion: The program addresses important needs but is not delivering its intended coaching function. Inductive analysis surfaced dimensions of student experience, including critical perspectives that deductive framework coding missed, suggesting the two approaches are complementary. The pattern of human-AI differences warrants further investigation, particularly regarding AI capacity for inductive versus deductive qualitative analysis.