Due to the risk of software defects and vulnerabilities, effective means for detecting and analyzing behavioral anomalies in device software is key for ensuring secure and reliable operation. Traditional logging is however not effective for large-scale deployments. Watch this presentation to learn about an alternative approach where anomalies are detected directly in the device and reported in a structured way to ensure nothing is missed. This allows for providing detailed information when it matters, while keeping the overall data uploads to a minimum. The presentation is concluded by a demonstration.