Improving Access to Public Services through an Intelligent Assistant with RAG: Experimental Evidence in a Local Government of Lima

Collado J.; Tupac-Agüero K.; Delgado F.D.C.; Gamboa-Cruzado J.; Vasquez M.R.; Lea G.M.; Meza Á.N.; Luis P.D.P.C.; Martínez-Seis B.C. · Computacion y Sistemas (2026) · DOI: 10.13053/cys-30-3-6779

Resumen

Timely and reliable access to information on public services remains a challenge in municipalities where traditional channels predominate, with restrictions in capacity, availability, and consistency of responses. The objective of this study was to develop and implement an Intelligent Assistant based on Retrieval-Augmented Generation (RAG) to optimize access to information and the management of queries on public services in Carabayllo, Lima. An applied study was conducted using an experimental post-test design with a control group and an experimental group, considering 60 observations distributed equally. The system was developed using Scrum and evaluated through four indicators: number of queries, accuracy of responses, waiting time, and user satisfaction level. The analysis included descriptive statistics, the Anderson-Darling test, t-tests for independent samples, and the Mann-Whitney U test. The results revealed statistically significant differences in favor of the experimental group across the four indicators: greater service capacity, higher response accuracy, shorter waiting time, and greater citizen satisfaction. Taken together, the findings show that incorporating RAG can improve operational and perceptual dimensions of municipal service. It is concluded that this architecture constitutes an effective alternative for modernizing access to public information, although its performance depends on the quality of the corpus, the retrieval of context, and the implementation conditions.

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad Nacional Daniel Alcides Carrión; Universidad Nacional de Moquegua