Today, chatbots are commonly used for music, mainly targeting young people. These chatbots are tailored to the preferences of younger users, but they often overlook the usability challenges experienced by older adults. There is little research on adapting them for older adults. To address this, we conducted a survey with 100 older adults to understand their music preferences and requirements for a music chatbot. Then, we developed the music chatbot based on the human-centred design (HCD) principles while also incorporating LLMs and prompt engineering. We tested the music chatbot prototype with five older adults, who generally had positive feedback although trust and reliability remain a challenge. Moving forward, we aim to explore ways to make the music chatbot more inclusive based on unique preferences and needs of older adults.