Tony Fadell says AI gadgets failed by chasing tech, not trust

When Tony Fadell took the stage at the inaugural MIT Future Fest, he put up a slide with three images of once-hyped AI devices that have already been discontinued: the Rabbit R1, the Humane Ai Pin, and the Limitless pendant. “These were kind of the ‘Gen 1’ AI products, and because of my background, these companies called me and asked me to help them,” Fadell said. “I didn’t.”

It’s no surprise those companies wanted his input. Fadell is credited as the “father of the iPod,” a co-creator of the iPhone, and the founder of Nest, the smart-thermostat maker Google later acquired. He said each of the three failed devices missed the same basic test: they didn’t meet a real need — interesting technology for enthusiasts, in his words, rather than something that applied to most people’s lives.

Part of the problem, Fadell argued, is that these products promised a personal assistant without most people having experienced what that relationship actually looks like. Fewer than 0.01% of the world’s population has ever had a human assistant do anything for them, he said, so when companies say they want to build “an assistant,” it’s often people who already have assistants projecting that need onto everyone else. It took him a couple of years, he said, to learn how to use an assistant well, trust them with sensitive data, and let them act as his agent — scheduling meetings, dealing with his bank.

Fadell pointed to Meta’s recent launch of Muse, its all-purpose AI assistant, as proof that trust is a harder problem with AI than with a person: a security researcher quickly found a serious vulnerability in it after launch, and a report from 404 Media said Meta employees had already flagged security issues that triggered a rushed, overtime fix before launch.

Fadell argued that the winning AI assistant will likely need to run primarily on-device, both to protect privacy and to stay lightweight, rather than depend on massive data centers. He floated Apple as the company best positioned to pull this off, given its hardware, chips, and the trust it has built around on-device data like Face ID — though he noted Apple still lacks a leading AI model of its own, leaning instead on custom versions of Google’s Gemini for the new Siri. Rivals like Meta and OpenAI, he suggested, are pushing into gadgets partly because they lack the vast base of devices — and the sensor access that comes with them — that Apple already has.

Asked about the role of luck in finding product-market fit, Fadell said startups have far less room for error than giants like Apple — a startup gets one shot, unlike Apple’s ability to absorb something like the Vision Pro.

Why it matters

Fadell’s critique suggests the next wave of AI hardware will be judged less on raw capability and more on whether companies can earn the kind of trust usually reserved for a bank or a doctor — a bar that security stumbles like Meta’s Muse show the industry hasn’t cleared yet.