Conversational agents powered by large language models (LLMs) have advanced significantly, but they often lack adaptation to users’ emotional and cognitive profiles, limiting adaptive interactions. This article presents an adaptive conversational framework that combines user profiling, including medical, emotion, and cognitive characteristics, with contextual adaptation using retrieval-augmented generation. Using real-world user profiles paired with both original and adapted health-related questions, we build an automated pipeline to dynamically generate responses that adjust response tone and complexity. A subjective evaluation shows strong alignment between adapted responses and user needs. The proposed framework provides a scalable path toward more inclusive, user-centred conversational systems.