Articles

Enterprise modeling (EM) methods such as BPMN, ArchiMate, and UML assume mature digital infrastructures and professionals specialized in modeling; conditions that are rarely met in organizations operating in peripheral or resource-constrained environments. At the same time, large language models (LLMs) and multi-agent architectures are redefining the possibilities for eliciting, formalizing, and validating organizational processes. Despite growing interest in AI-assisted EM, the literature lacks frameworks that explicitly address the roles of agentic LLM systems when modeling expertise is limited and the organizational context is informal, multilingual, or digitally fragile. This article proposes ALMEC (Agentic LLM-based Model Engineering Collaboration), a conceptual framework that defines four specialized agent roles—Elicitor, Formalizer, Validator, and Explainer—that collaborate with human process engineers throughout the enterprise model lifecycle (discovery, formalization, validation, and maintenance). The framework integrates explainability and transparency mechanisms oriented toward non-expert stakeholders and structures human–AI responsibility at each stage of the lifecycle. ALMEC is instantiated through a case study of multimodal logistics organizations in the Alto Amazonas region (Loreto, Peru), where limited connectivity, multilingualism (Spanish, Shawi, Kukama), and organizational informality constitute representative resource-constrained conditions. The results extend EM theory by offering a generalizable blueprint for LLM–human co-modeling in peripheral and underserved organizational contexts.