Artificial Intelligence at the service of Micro and Small Enterprises: Redesign of the SUNAT platform to optimize tax compliance, 2025

Abstract

This study evaluates the impact of an artificial intelligence (AI) platform for mass expense validation on tax compliance costs for micro and small enterprises (MSEs) in Lima's commercial sector. Combining a mixed-methods approach, we analyze efficiency gains, accuracy improvements, and accessibility enhancements tailored to Peru's unique context. Our methodology integrates a PRISMA systematic review of 85 international studies (2018–2025) on tax applications of AI, the analysis of approximately 5,600 actual SUNAT receipts (2025/01 to 2025/06) to train context-specific algorithms, pilot tests with 17 MSEs in three major commercial districts (Centro de Lima, Lurigancho- Chosica, San Martin, Lince y Los Olivos), and econometric modeling for macroeconomic projections.. Key findings demonstrate a 78% reduction in validation time per receipt (p < 0.01), 65% decrease in declarative errors (95% CI: 62-68%), average annual savings of S/ 6,800 per business in compliance costs, and an 83% adoption rate facilitated by our bilingual (Spanish-Quechua) interface. The study makes three original contributions: (1) the first AI system specifically designed for Peru's structural informality, (2) a scalable methodological framework for digital tax solutions in emerging economies, and (3) empirical evidence on the cost- benefit ratio of AI implementations for MSEs.