The rapid growth of Indonesia’s elderly population highlights the need for community programmes that not only extend lifespan but also improve healthspan. This study aims to design ageing activity topics based on participants’ suggestions using Natural Language Processing (NLP) and unsupervised clustering. Data were collected from 213 open-ended responses from participants of the Center for Aging Wellness (CAW), University of Surabaya, covering liked aspects, improvement areas, and future topic suggestions. Text data were pre-processed (lowercasing, removal of punctuation, digits, and stopwords, and spelling normalization), then transformed using TF-IDF and clustered with K-Means. Five values of k (2–6) were tested, with k = 6 selected based on the highest silhouette score and interpretability. Six thematic clusters emerged: knowledge literacy (12.2%), appreciation and learning (35.7%), leisure and social activities (8.9%), brain health and dementia (18.3%), physical and mental health (15.5%), and financial independence (9.4%). These clusters were mapped into broader need categories and aligned with the Indonesia Longitudinal Aging Survey (ILAS) 2023 domains. As an exploratory study, the results support thematic mapping and programme planning rather than prediction. The findings demonstrate that participant feedback can be transformed into an evidence-based roadmap for more targeted and context-sensitive ageing community programmes.