The integration of Large Language Models into recommendation systems holds significant potential for improving personalization and usability in academic contexts. This study evaluates the acceptability and usability of an LLM-based recommendation system implemented on the IdeaLize project platform of the Heilbronn University of Applied Sciences, which supports collaboration between students and faculty. Through a mixed-method approach, including surveys, interviews, and focus groups, user-oriented insights and practical feedback were gathered, while the key metrics for usability and acceptance were identified through a systematic literature analysis. The findings emphasize the intuitive usability of the system and the value of personalized recommendations while identifying challenges such as slow performance and data maintenance. Users suggested prompts and use cases that highlight the system’s versatility, from finding project ideas to networking with potential collaborators. The study contributes to the understanding of user needs and evaluation metrics for LLM-based systems and offers practical recommendations for their further development and integration into academic workflows.