Universities face a growing demand for information during admission processes, leading to an administrative overload due to the large number of individual and in-person inquiries from applicants. This work proposes the development of a chatbot based on Natural Language Processing (NLP) to automate and optimize the management of inquiries in admission processes, taking the National University of Moquegua (UNAM), Peru, as a case study. A specific corpus was built from the UNAM General Admission Process Regulations, supplemented with real inquiries from institutional sources. The corpus was segmented and preprocessed to train five neural network architectures: RNN, CNN, LSTM, GRU, and BiLSTM. Each model was evaluated in terms of accuracy and generalization capability using a preprocessing pipeline that included tokenization, lemmatization, stopword removal, and conversion to numerical sequences. The proposed chatbot uses a neural network architecture with a 300-dimensional embedding layer, followed by different sequential processing layers depending on the selected architecture, and an output layer with softmax activation for multi-class classification. The models were trained using categorical cross-entropy loss and the Adam optimizer, with 60 training epochs. The results show that the architectures based on BiLSTM and RNN outperform the others in terms of accuracy and robustness, suggesting their suitability for implementation in university admission environments. The developed system has the potential to significantly reduce the workload of administrative staff and improve the experience of applicants by providing quick and accurate responses to their inquiries.