AI Agent Architectures: a Comparative Study of Workflows versus A2A (Agent to Agent) and their Application in the Education, Industry and Services Sectorsagénticas de IA: U

• Autores: Bustillos Ortega, Olda, Murillo Gamboa, Jorge, Mena Bocker, Daniel, De la O Fonseca, Carlos, Aguilar Mora, Carlos
• Formato: artículo
• Fecha de Publicación: 2025
• Institución: Universidad de Lima
• Repositorio: Revistas - Universidad de Lima
• Lenguaje: español
• OAI Identifier: oai:ojs.pkp.sfu.ca:article/8430
• Enlace del recurso: https://revistas.ulima.edu.pe/index.php/Interfases/article/view/8430
• Nivel de acceso: acceso abierto
• Materia: agents; artificial intelligence; education; technology; workflows; agentes; educación; flujos de trabajo; inteligencia artificial; tecnología

The development of agentic architectures for artificial intelligence (AI) has ledto new models of interaction and automation, most notably AI agent-based workflows andagent-to-agent (A2A) communication. This paper presents a comparative study of botharchitectures, analyzing their design principles, human-AI coordination capabilities, scalabilityand adaptability in different environments, as well as obstacles and challenges. The analysisshows that while agentic workflows provide efficiency and control in stable environmentswith well-defined processes, A2A architectures excel in distributed and heterogeneouscontexts, offering greater flexibility and autonomy. Based on the premise that technologyshould strengthen, not replace, human capabilities, this study aims to generate a significantimpact in the areas where it is most needed. Representative examples from the education,healthcare, and industrial sectors are examined, demonstrating how these architecturestransform key processes. Comparative figures and tables are presented, integrating variousapproaches between the two agentic AI architectures. Finally, the technical and ethicalchallenges associated with their implementation are discussed, and future lines of researchare proposed for a responsible and effective adoption of these technologies.