London-based startup Plumerai is challenging the smart home security status quo by introducing advanced on-device AI. By shifting facial recognition and people detection from remote cloud servers directly to the hardware, the company aims to eliminate the privacy risks and high operational costs associated with modern connected cameras.
The Privacy and Cost Crisis in Smart Home Security
Public scrutiny regarding surveillance technology has intensified this week following reports of Harvard students modifying Ray-Ban Meta glasses for facial recognition. As smart cameras become ubiquitous, the reliance on remote servers—often managed by major corporations like Amazon—has created significant security vulnerabilities and privacy concerns.
Tony Fadell, the creator of the iPod and co-founder of Nest, is an early investor in Plumerai. He argues that the current industry model is fundamentally flawed. “We’d have to worry so much just about the storage cost and the data transmission costs,” Fadell explains. “We’re taking full frames. It’s a ton of stuff that we’re recording, but not recording on-camera. I felt the weight of this all the time.”
Efficiency Through “Tiny AI”
Unlike the massive Large Language Models (LLMs) that power services like ChatGPT, Plumerai focuses on “tiny AI.” These models are optimized to perform complex tasks on low-power consumer electronics without needing the massive compute power typical of large-scale cloud infrastructure.
Fadell draws a parallel between this approach and the evolution of the iPod. “The only reason the iPhone could exist is because we started small with the iPod. Usually you can grow things up, you can’t make big things small,” he notes. “Microsoft tried to take Windows and make Windows Mobile on a phone. They took this big thing and it never worked. You have to start small.”
Real-World Application: Chamberlain Group Partnership
The technology is already moving from concept to consumer hardware. The Chamberlain Group, parent company of brands like myQ and LiftMaster, has partnered with Plumerai to integrate its AI directly into upcoming smart cameras, starting with an outdoor model.
Plumerai CEO Roeland Nusselder emphasizes the performance benefits of this local processing. “Our tiny AI is more accurate and runs on lower cost, lower power chips than anything else that’s out there—especially in the smart home camera market,” Nusselder says. “All of the AI features are from Plumerai, running locally on the camera.”
The Advantage of a Lean Approach
For companies like Ring, high operational costs are often passed down to the consumer, a trend highlighted by Amazon’s recent decision to increase professional monitoring fees. Plumerai, operating as a leaner startup, avoids the bloat associated with massive corporate tech conglomerates.
Fadell believes this focus is exactly what the industry needs to disrupt the current landscape. “I have learned that small teams—in the 10s, the 50s—can really do a lot when it’s the right set of expertise around the table,” Fadell says. “I like to be at the tip of the spear on disruptive technologies. It’s small teams with the right idea.”