Diseño e Implementación de un Software Web para el Control de Ingresos y Gastos con Enfoque Tributario, con Asistencia de un Agente Inteligente

Abstract

For self-employed individuals in Ecuador registered under the RUC, meeting tax obligations represents a considerable administrative burden. Most taxpayers rely on manual spreadsheets, which frequently result in calculation errors, missing vouchers, and fines from the Servicio de Rentas Internas (SRI) that could have been prevented. This work describes the development of a web platform aimed at individual taxpayers not required to maintain formal accounting records, as well as the accountants who support them. Rather than replicating existing corporate accounting software, the system addresses the specific needs of this segment by tracking sales, purchases, withholdings, and credit notes; computing VAT per period; and categorizing expenses according to the deductible classifications established under Ecuadorian tax legislation. A conversational agent powered by OpenAI's GPT-4.1responds to queries in plain language, enabling users to consult their own tax records without requiring specialized legal knowledge. The platform was developed using Next.js, Supabase, and Vercel. Testing with real taxpayer data demonstrated that the monthly tax cycle decreased from 590 minutes to 17 a 97.1% reduction while eliminating the recurring errors associated with manual processing. These findings support the practical viability of the proposed solution for this segment of Ecuadorian taxpayers.