The timely diagnosis of Alzheimer’s (AD) disease is essential for its treatment, since evaluating the severity and risk of progression facilitates suitable preventative measures before irreversible cerebral damage. Alzheimer’s disease is a chronic degenerative condition that is characterised by the loss of neurons in the brain. Brain MRI is used alongside other approaches to diagnose this illness. Formalized automated MRI analysis methods can aid in diagnostic decision-making. Deep learning methods, particularly those employing convolutional neural networks, can serve as an effective means for developing such tools when a significant training dataset is accessible. This project aims to use a convolutional neural network (CNN) to classify phases of Alzheimer’s disease through brain MRI, using a dataset of 5000 chosen pictures processed with the Python programming language. The principal classifications of disease severity are NonDementia (absence of dementia), VeryMildDementia (incipient dementia), MildDementia (moderate dementia), and ModerateDementia (severe dementia). In comparison to other classes in the model evaluation, the proposed model achieves a 98 F1-measure and 0.998% accuracy in the Very Mild Demented class, accurately identifying all disease stages using accuracy and test loss metrics. Additionally, it demonstrates strong performance based on primary classification criteria. Identifying this stage of the disease is crucial for developing a medication that inhibits its progression in the future.