Architectures of knowing: toward a theory of AI-augmented observability in enterprise knowledge systems

Purpose This study aims to propose a new approach to enterprise architecture (EA) that integrates artificial intelligence (AI) and knowledge management (KM) to shift from static models to dynamic and reflexive systems of organizational knowledge. Design/methodology/approach Building on insights from EA, KM, AI and infrastructure, this study developed a knowledge architecture that integrates four interconnected layers: tacit, explicit, behavioral and cognitive. After analysis, these layers presented four evolutionary paths, namely, externalization, combination, validation and internalization. Findings showed that AI technologies, such as large language models, search-assisted generation and semantic graphs, mediate and structure knowledge flow. And two concrete examples are presented here as evidence. The first is a conceptual model that positions EA as a means for KM, and the second is a design science artifact that demonstrates AI-enabled observability across system layers. Findings By accounting for epistemic observability, this study realizes the ability of organizations to rely on AI to reveal, verify and exploit architectural knowledge, so as to interact with system behaviors and artifacts in real time. The findings show that EA can be reframed as a flexible, reflexive system that fosters continuous learning, adaptability and better decision-making in dynamic environments. Practical implications The proposed model and tools enhance organizations’ capacity for knowledge sharing, decision support and adaptive governance. They also provide practical avenues for applying AI technologies in EA and KM to strengthen digital infrastructures. Originality/value This study redefines EA as a dynamic knowledge system, opening up new avenues for research in information systems, hybrid thinking and digital governance. It also shows AI’s potential to expand EA’s scope and impact.