Background: Artificial intelligence (AI) is increasingly incorporated into higher education. However, most empirical studies focus on technological adoption or learner satisfaction rather than on how pedagogical design, perceived learning impact, and student experience interact within AI-mediated learning environments. Understanding these relationships is essential to determine whether AI supports higher-order reasoning processes rather than merely increasing technological engagement. Objective: This preliminary study aimed to develop and evaluate a theory-driven AI-mediated pedagogical framework and examine relationships between pedagogical design, perceived learning impact, and student satisfaction in a university learning context. Methods: An observational educational evaluation was conducted during implementation of an AI-mediated instructional framework in an undergraduate physiotherapy course. The full academic cohort (n = 22) completed a 24-item questionnaire assessing seven pedagogical domains on a 5-point Likert scale. Descriptive statistics, Pearson correlations, exploratory regression modeling, and factor analysis were used to examine relationships among domains. Academic performance indicators were summarized descriptively. Results: Students reported high evaluations across all domains (means > 4.5/5). The strongest association with satisfaction was perceived learning impact (r = 0.79, p < 0.001). Moderate correlations were found for usability (r = 0.66), AI content quality (r = 0.61), pedagogical coherence (r = 0.58), critical thinking (r = 0.52), and ethical integration (r = 0.47). Academic pass rates exceeded 90%. Conclusions: Perceived learning impact emerged as the central mechanism linking AI-mediated instructional design to student satisfaction, suggesting that the educational value of AI depends on alignment with cognitively demanding learning processes.