Aurea Imaging, a Dutch startup, focuses on remote sensing solutions for agriculture using NVIDIA Jetson devices. Their approach leverages a cloud-native model, deploying K3s Kubernetes on edge devices instead of the traditional embedded C and ROS stacks. A significant challenge they face is keeping the Jetson boards updated, particularly with software like JetPack and CUDA, which is essential for machine learning applications in the field. With devices often located in remote areas with unreliable network connections, managing updates and maintaining performance becomes complex. To address this, Aurea turned to Kairos, a CNCF Sandbox Project that allows for immutable OS configurations. Kairos simplifies Day 2 operations by enabling remote updates without the need for on-site technicians. This is critical when managing a global fleet of devices. By using Kairos, they eliminated inconsistencies in their OS and reduced troubleshooting headaches. The PoC they developed ensured a uniform fleet and streamlined provisioning processes, cutting down on installation time and minimizing human error. The upgrade process with Kairos is efficient and reliable, thanks to its design around Kubernetes. Devices automatically handle their own updates by pulling new images from a private registry. The A/B upgrade system ensures that if an update fails, the device can revert to a previous stable version. Maintaining these systems now involves using a Dockerfile to manage package installations, making firmware updates and fixes much easier to deploy across all devices simultaneously. Collaborating with the Kairos community not only improved their fleet's maintainability but also enabled Aurea to contribute enhancements back to the platform, benefiting the broader open-source ecosystem.