Plumerai Raises Stakes for On-Device AI in Home Security Cameras
Plumerai brings highly efficient, on-device artificial intelligence to home security cameras, eliminating the need for cloud processing. Backed by Tony Fadell, the company aims to reduce costs and privacy risks for consumers.
Recent facial recognition projects involving smart glasses highlight growing privacy concerns surrounding ubiquitous cameras. In response, London-based startup Plumerai focuses on enhancing on-device AI processing for home security systems. The company develops technology that performs people detection and familiar face identification directly on the camera, removing the need to send sensitive data to remote servers.
Tony Fadell, the co-creator of the iPod and early investor in Plumerai, cites his struggles with cloud storage and data transmission costs during his time at Nest as a key reason for backing the firm. He points out that relying on remote servers forces companies to spend heavily on additional computing power, a cost that inevitably gets passed down to consumers through higher subscription fees like those recently implemented by Ring.
Plumerai specializes in tiny AI models that operate efficiently on low-cost, low-power chips rather than relying on massive, compute-heavy systems like large language models. Fadell compares this approach to the development of the iPhone, noting that starting small and scaling up proves far more effective than attempting to shrink massive, complex systems. Plumerai's CEO claims this targeted method currently delivers industry-leading accuracy while running on significantly cheaper hardware.