Tony Fadell just showed the world a graveyard. At the inaugural MIT Future Fest last week, the man who invented the iPod and co-invented the iPhone projected a slide with three images: the Rabbit R1, the Humane AI Pin, and the Limitless pendant. All discontinued. All dead within two years of launch.
His verdict was brutal: “You have to really understand what you’re trying to do, what pain you’re trying to solve. In each of these cases, each one of them didn’t meet any kind of need — it was just interesting technology for geeks, and you’re like, ‘Okay, that’s neat, but it doesn’t really apply to my life.'”
I’ve been thinking about that quote for days. Not because it’s controversial — it’s obvious, really — but because it cuts through so much of the noise we wade through in tech. We’ve been told for two years that AI is going to replace our phones, our apps, our workflows. And then the hardware showed up, and it was… a $199 orange box that couldn’t reliably order an Uber.
The Graveyard of Good Intentions
Let’s do a quick autopsy. The Humane AI Pin launched in April 2024 with a $699 price tag and a promise to replace your smartphone with a screenless, voice-driven assistant. Reviewers savaged it. MKBHD called it “the worst product I’ve ever reviewed.” It overheated, misheard commands, and burned through battery in hours. By February 28, 2025 — less than a year later — Humane had discontinued the Pin, shut down its cloud service, and sold its assets to HP for roughly $116 million. The company had raised over $230 million.
The Rabbit R1, designed by Teenage Engineering, launched the same month at $199. It was a cute little orange square with a scroll wheel and a camera. The pitch was a “Large Action Model” that would operate your apps for you. The reality was a sluggish device that ran Android apps in a custom wrapper and couldn’t do much your phone couldn’t. It’s still shipping — Rabbit has released software updates through 2025 and into 2026 — but it’s a niche curiosity, not a category-defining product.
The Limitless pendant, made by a company Meta acquired, was discontinued on December 5, 2025. Meta folded the technology into its broader wearable efforts rather than continuing it as a standalone product. Another $30 million in funding, gone.
Three devices. Hundreds of millions in funding. Two years. Zero survivors.
The Assistant Fallacy
Fadell’s critique goes deeper than “these products were badly executed.” He identified something more fundamental: most people don’t actually know what an assistant is.
“Most consumers don’t even know what an assistant is,” he said. Fewer than 0.01% of the global population has ever employed a human assistant. When tech executives talk about building the ultimate aide, they’re projecting their own experience with support staff onto everyday users who lack that frame of reference.
“It took me a couple of years first to understand how best to utilize [an assistant], and then to trust them with the most sensitive data, and have them be an agent where they’re setting up meetings with people and working with my bank,” he said. If it took the father of the iPod two years to trust an assistant with his bank, what chance does a $199 gadget have?
This is something I think about a lot as someone who manages an ICT division. We roll out tools to staff who didn’t ask for them, don’t understand them, and won’t use them until someone sits with them and shows why it matters. The technology is never the hard part. The trust is.
Trust Is the Feature
Fadell’s prescription for the next wave of AI gadgets is simple: trust and safety are going to be paramount. “Anything that handles personal messages, location data or financial details must earn confidence first.”
He predicts the winning AI agent will run entirely on-device. Not because cloud computing is bad — it’s incredibly powerful — but because on-device processing protects privacy, reduces latency, and keeps the system lightweight. Apple has spent years building this trust through features like Face ID, where your biometric data never leaves your phone.
“It’s because they don’t have access to the sensors on your phone,” Fadell said of cloud-only AI companies. “So what do they do? They go and make a device with all that stuff in it without a screen… and then they just Bluetooth or Wi-Fi into your phone and then back up to the web, or into a 5G network, so they can get all the sensor data.”
That’s the fundamental tension. Cloud AI companies need your data to function. On-device AI companies need your trust to succeed. Fadell is betting on trust.
What Actually Works
The interesting thing is that AI gadgets aren’t dead. They’re just narrowing. The devices succeeding right now share a common trait: they stopped pretending to replace your smartphone and started being very good at one specific job. It’s the same principle I covered in my piece on using AI text detectors without getting burned by false positives — narrow, reliable tools beat grand promises every time.
Take Plaud. The San Francisco company makes a card-sized AI meeting recorder — the Plaud Note — that attaches magnetically to your phone case, records conversations for up to 30 hours, and transcribes them across 112 languages. As of June 2026, Plaud has shipped over 2 million devices and hit $100 million in annual recurring revenue. They took roughly $6 million in total funding and built a business that actually works.
The difference is stark. Plaud didn’t try to be your assistant. It tried to be a better notepad. It solved a specific problem — capturing and transcribing meetings — for a specific audience — professionals who can’t afford to miss details. No grand promises about replacing your phone. No $699 price tag. Just a tool that does one thing well.
This is the pattern Fadell was pointing at. The most successful products are often the most boring from a technology perspective. They just work reliably and make people’s lives genuinely better.
My Take
I’ll be honest: I was excited about the AI Pin and the R1. I wanted them to work. The idea of a screenless assistant that could handle tasks without me pulling out my phone felt like the future. But Fadell is right — I didn’t need them. I have a phone that does everything they promised, but better, faster, and with a screen I can actually see.
What I do need is tools I can trust. And trust in AI is getting harder to establish — as I noted in my analysis of OpenAI’s new reasoning technique that made AI safety harder to see. As someone who manages IT systems and handles sensitive government data — and who recently wrote about managing Copilot’s file access and privacy settings — I can’t afford to hand my information to a device that might leak it, might mishear it, or might stop working when the company runs out of money. That’s not a tech problem. That’s a trust problem.
The next wave of AI gadgets will succeed or fail on one question: can I trust this with my most sensitive data? If the answer is yes, it doesn’t need to be flashy. It just needs to work.
Fadell built the iPod by asking a simple question: how do we carry a thousand songs in our pocket? Not “how do we build a screenless AI assistant?” Not “how do we replace the smartphone?” Just: what problem are we solving?
The graveyard is full of companies that forgot to ask that question. The next wave would do well to remember it. Even Microsoft, with its new Surface Laptop Ultra and agent security layer, is betting that on-device AI and trust are the way forward.