It is unknown how the 16 billion neurons in the human cortex coordinate to integrate diverse streams of information into coherent perception, cognition, and action. Progress in identifying the neural activity underlying this process has been limited by the constraints of noninvasive recording methods in humans. Recent work leveraging rare intracranial recordings in humans has identified networks of high-frequency ‘ripple’ oscillations that co-occur and phase lock across distant cortical sites and appear to play a role in cortical integration. Two clear network architectures have emerged consistently across a variety of cognitive contexts, with high consistency and specificity. The first occurs early after stimulus onset, is highly selective for stimulus type, and follows a hub-and-spoke architecture with a stimulus-related brain region at the center that co-ripples with frontoparietal areas. This network architecture appears to be involved in broadcasting stimulus information to the cortex for further processing. The second network architecture is specific for correct performance and cognitive effort, it follows the radial network, occurs only in target conditions, and consists of distributed and densely interconnected co-rippling all across the frontoparietal cortex. This network is highly selective for cognitive effort and task performance, suggesting its involvement in general integration. These network architectures have now been observed across a wide array of contexts: across the visual and auditory domains and across semantic, pictorial, and face stimuli. They have also been shown to occur across both hemispheres and have demonstrated potential to be lateralized. Together, these findings suggest that cortical co-ripple networks are fast, spatially distributed, behaviorally-relevant and coordinate across hemispheres and several cognitive domains. These findings support the theory of co-ripple networks as a putative mechanism for domain-general cortical integration.