Recommendation systems (RSs) are very closely embedded in our lives, and among the numerous information, they selectively present personalized information to users. Such systems help users make innumerable decisions in the choices they face. However, the current system provides recommendations based on the user's behavioral data, which only accepts what is given and does not help them think about what they want. For such user limitations, this present work explores the possibility of designing RSs that can engage reflective experience. It allows users to build a system in a user-centered way rather than passively using an RS and will enable users to mutually benefit from intelligence, providing a better interaction for both the system and the user. Through historical photo-description data collection, designer workshops, and user interviews, we offer a reflective RS experience and revealed findings focusing on new user experiences that arise compared to traditional RSs. Based on this, we find that a reflective RS is possible, and we conclude the paper with discussions for designing RSs that engage the user in reflection.