In contrast to monolithic AGI approaches, this position paper introduces the Cognitive Agent Networks (CAN) paradigm, where general intelligence emerges as a property of interactions between multiple autonomous agents. CAN’s key components include distributed cognitive functions, coordination via shared ontologies, and synchronization through a mechanism we term ’cognitive resonance.’ We present a conceptual framework and initial formal steps to define CAN and experimentally demonstrate, using a simplified search task, that a hybrid network (AI + simulated human) surpasses a purely machine-based system in efficiency. This work also delineates the principal challenges and directions for future research toward the development of advanced AI through hybrid cognitive networks.