Some Harvard University students made headlines Wednesday after wearing Ray-Ban Meta glasses with facial recognition capabilities. DIY projects are the latest news to fuel privacy concerns about ubiquitous technology. This issue is gaining importance as cameras become more and more prevalent in society.
Connectivity plays a big role in this conversation as well. Remote server requirements come with their own security/privacy issues. More problems arise when Ring's parent company, Amazon, and law enforcement get involved.
Founded in 2017, Plumerai's main value pillar is technology that powers on-device AI processing. The London-based company has developed a way to perform tasks such as person detection and identifying familiar faces without having to send data to a remote server.
Tony Fadell is an early investor. The iPod developer cited problems he encountered in making his decision to support the company as a co-founder of Nest.
“Storage costs and data transmission costs alone should be this much of a concern,” he told TechCrunch. “We're shooting full frame. There's a lot of stuff we're recording that isn't being recorded on camera. I've been feeling the weight of that.”
Adding computers means additional expenses for businesses, which are often passed on to consumers. Fadel points to Ring's recent decision to double the cost of professional 24/7 monitoring as a key indicator.
Plumerai specializes in small AI trained on much smaller models than the large black boxes underlying platforms like ChatGPT. Such large-scale language models (LLMs) rely on vast repositories of data, require too much computational power for small consumer electronic devices, and are prone to hallucinations.
Fadell likens the move to smaller models to when he was working on the iPod.
“The only reason the iPhone existed is because we started small with the iPod. You can usually make things bigger, but you can't make big things smaller,” he says. “So we started really small and grew the iPhone from there. Remember, Microsoft took Windows and tried to put Windows Mobile on their phones. But it never worked. You have to start small and grow from there.”
“We've been working on this for a very long time,” Plumerai CEO Roeland Nusselder told TechCrunch. “Experientially, our miniature AI is more accurate and runs on lower-cost, lower-power chips than anything else, especially in the smart home camera market.”
The startup has found a believer in Chamberlain Group. The Illinois company, parent company of brands such as myQ and LiftMaster, plans to incorporate Plumerai's products into its smart cameras, starting with outdoor cameras.
“All AI functionality comes from Plumerai and runs locally on the camera,” says Nusselder. “The way I see it, Chamberlain is not a big tech company, but a company that can do really great things with a little AI.”
Plumerai doesn't disclose how many employees it has, but it's almost certainly far fewer than the teams behind Ring and Nest. Being a relatively lean startup means we've traditionally focused on our own market segment. This is not a luxury that brands under the banners of large companies like Amazon or Google are afforded.
Fadel has spent time as an executive at some of the world's biggest tech companies, but now he's focused on supporting startups like Plumerai.
“The key is concentration,” he explains. “I learned that even a small team of 10 or 50 people can do so much with the right expertise. I love being on the leading edge of disruptive technology. We are a small team with ideas.”
See Tony Fadell at Disrupt 2024 — buy your tickets here.