LUCENT
LUCENT defines a new computational primitive where computation occurs via event-driven state transitions, not continuous inference, coordination emerges from local interactions, not global optimization, and system behavior is adaptive and resource-aware by design. While prior work has studied these mechanisms independently, LUCENT is the first framework to integrate them into a single operational system. This approach is uniquely grounded in insect neuroscience and physiology, including well-characterized mechanisms such as pulse-coupled synchronization and looming-sensitive neurons and biologically inspired engineering paradigms, such as soft robotics. The model is directly informed by domain expertise in insect neural systems, enabling a level of biological grounding that is typically abstracted away in AI models.
RAYS
RAYS is Glowhopper’s human-in-the-loop AI data verification engine that continuously monitors, scores, and validates community health infrastructure data, escalating only uncertain cases to regional data verification managers to ensure real-world accuracy at scale. RAYS incorporates a multi-agent AI system that continuously monitors naloxone access sites, predicts data decay, and dynamically orchestrates human re-verification, ensuring a self-updating, high-fidelity public health dataset. This pipeline directly powers Glowhopper Go, an interactive deployment platform that surfaces verified naloxone access points in real time, and feeds into LUCENT for continuous model improvement.
Glowhopper Go
Glowhopper Go is a digital platform built directly from our foundational dataset of naloxone distribution sites. It translates verified, community-updated information into real-time access for individuals seeking life-saving resources. Powered by RAYS, Glowhopper Go is designed to surface the most reliable, actionable information available while continuously adapting as new data comes in. Glowhopper Go is live in New Haven and in the process of expanding to New York. Learn more View pilot results