This paper presents the design and implementation of an intelligent, domain-specific chatbot for answering Alzheimer’s disease-related questions. Leveraging a Retrieval-Augmented Generation architecture, the system integrates biomedical embeddings derived from PubMedBERT. The model has generative capabilities of the Mistral-7B-Instruct model to deliver precise, and evidence-based responses grounded in the scientific literature. To ensure contextual relevance and mitigate hallucination risks, a Facebook AI Similarity Search vector database was constructed from a curated set of academic PDFs. It facilitates retrieval of authentic information. Computational challenges associated with deploying large language models on CPU-limited hardware were addressed through careful optimization of model parameters and embedding procedures. The chatbot is deployed via Chainlit, providing an interactive and accessible interface for users such as students, researchers, and caregivers seeking reliable information on Alzheimer’s disease. The topic specific answering potential of the chatbot increases its application in the real world.