Invasive mammals pose a significant threat to native island ecosystems. However, detecting and removing mammals at low densities remains a costly and challenging endeavor. While camera traps (CTs) are effective detection tools they must be manually visited to collect data for analysis, confirmation of target animal detection, and subsequent management action. Deploying CTs in remote, inaccessible, or dangerous areas leads to delayed data collection and processing, hindering timely management efforts. To decrease time, costs, and risks associated with CTs Conservation X Labs developed a prototype automated monitoring system called the Sentinel. The Sentinel plugs directly into CTs, processes sequences of collected images to detect if animals are present using on-board machine learning models. Within several hours of CT the first trigger event, the Sentinel transmits text-based metadata of animal detections, termed insights, via low bandwidth satellite communication protocols to an online dashboard that automatically notifies field staff of events of interest. To further advance the utility of this device in disconnected environments or where CT access is dangerous access, we expanded the transmission capabilities of the Sentinel to include remote upload of CT data to a nearby small Uncrewed Aircraft Systems (sUAS). Here, we report on the first field-test of the prototype Sentinel units for invasive mammal management. We conducted this test on Kaho‘olawe Island, Hawai’i where island managers face significant challenges in managing feral cats due lack of connectivity, steep terrain, and unexploded ordinance. Our results confirm that remote data retrieval from Sentinels using sUAS effectively addresses current limitations by reducing physical retrieval time by 300-400%. Further, we found that although the deployed ML model achieved an accuracy of >90% in detecting the species of interest, generated insights were limited to text or highly compressed images hindering direct validation of automated satellite-delivered detections by end-users. In use-cases like invasive mammal management, high confidence in data accuracy is required to guide subsequent management decisions, meaning that even with satellite-delivered insights end-users had a strong desire to regularly collect all original CT data to confirm model predictions. We found that combining Sentinel’s close-to-real-time species detection and data transmission capabilities with sUAS remote upload capabilities holds significant promise as an end-to-end rapid monitoring system, especially where human verification of raw data is required and physical access to cameras is costly, dangerous, or could negatively impact animal behavior.