An AI Agent Model for Analyzing Recorded Calls at a Call Center in Peru

Zavalaga A.M.A.; Sullca G.J.B.; Alarcón J.S.D. · International Conference on Artificial Intelligence, Computer, Data Sciences, and Applications, ACDSA 2026 (2026) · DOI: 10.1109/acdsa67686.2026.11467865

Resumen

Quality monitoring in call centers is a critical but often inefficient process, where a high volume of interactions is not systematically evaluated, generating operational overload and service inconsistencies. Despite the existence of technological solutions, their adoption in the Peruvian environment is limited. In response to this problem, this study aims to propose and validate an automation model based on an artificial intelligence (AI) agent to analyze recorded calls in a collections contact center, adapting the analysis to the company's specific business rubrics. The methodology was developed in four phases, culminating in an experiment conducted in a real business environment. First, related works were reviewed and tools were selected through benchmarking. Then, an AI agent-based model was designed with Dialogflow and N8N integration. In the third phase, real data were collected and evaluation metrics were defined. Finally, a validation process was carried out by comparing the automated model with the manual process. The implemented algorithm employs intent recognition (Intent Classification) and entity extraction (Entity Extraction) models to analyze recorded calls in a collections center. Validation in a real environment demonstrated an average accuracy of 90.5% and an 83.8% reduction in call analysis time, significantly optimizing the operational workload of the quality area.

Instituciones: Universidad Peruana de Ciencias Aplicadas