Is Limitless Pendant the Future of AI Devices?

Since Jony Ive and Sam Altman announced the existence of "io", the best guessing game in Silicon Valley is trying to figure out what these"AI devices" are that they are talking about.

Sometimes I feel like I’ve just seen the future and I want to own it. That’s the reaction I had when someone told me about the Limitless Pendant about a month ago. So I ordered one and started using it on April 20, a Sunday.

My Limitless Pendant, currently turned off which you “know” because there is no white light to indicate that it is listening.

In the interim, Jony Ive and Sam Altman revealed their bromance intended to produce “AI devices” without actually defining what that means. The Limitless AI Pendant is one such device. In a tech-only irony, the seed investor in the company that produced the AI device is Sam Altman. Maybe there is a connection between his fever dream of the perfect AI device and the company he seed-funded, which has now produced an example of one such AI device. Forthwith, a critique of this AI device with an eye toward how it helps us understand what that means for the future.

Bottom line: It’s a good product with some value and a lot of promise, but most of that promise is unrealized.

I was expecting an unobtrusive device that would keep track of everything I said and heard during the day and help me remember what I talked about or listened to help me be more effective. That’s the expectation the company sets: “Our vision is to free the human mind from its biological limitations”. That’s more aspirational than realized.

This is what I learned about using the Pendant after a month.

1. It doesn’t integrate with ear buds and headsets, so it can’t hear those conversations. Not a big problem if you are wearing them for consuming audio in music or video, but a real problem for joining real-time conversations. (To give credit, the device does a good job of picking up and distinguishing all the speakers in a busy environment, like a restaurant.)

2. Privacy is a usability issue. In theory, you are supposed to ask for permission in order for the device to record other people. I haven’t seen that, although I read that that “feature” is turned off by default. I guess I haven’t turned it on and I’ve never asked permission. Most people don’t even notice the Pendant, which is small and unobtrusive.

3. You can buy a docking station as an accessory. It’s sort of a BS accessory, implying that you need to give the pendant the same care and feeding you do other battery-powered devices. The company claims that the device lasts for 100 hours. I haven’t had that experience; it feels like it runs out faster if you don’t turn it off.

   1. The point of the device is that you don’t have to turn it off or otherwise do manual power management. That is just not true. If you just leave it on and don’t pay attention, it will run out of power. If you do pay attention, you’ll learn to turn it off when you don’t want it recording, which inevitably leads to forgetting to turn it back on. I’ve missed a few key conversations.

   2. The power button is tiny and black, the same color as the device. When it’s dark, I find it impossible to find the button by feel. The device is symmetrical in design, so there’s no surefire way to know whether it’s on the right or the left side. The button should be bigger and, like white, so it’s obvious where it is. A failure of user-centered design (something that Jony Ive knows a lot about, I think).

4. The key value of the device is its AI interface, which resides in an app on the phone. The device syncs data to the phone by Bluetooth, so the app has to be open and syncing live; it’s pretty quick, lagging by a few minutes. The app has a search box through which you can query the data it has transcribed and summarized. The summaries in the sample photo are not useful (since I don’t want to show the Important Stuff!), but it gives you an idea of what the summaries look like, which are largely useless, a feature that should be removed: given the existence of “AI”, they should label each conversation to help you get a visual of your day. Behind this dumb UX is the source audio and a transcript, but it’s not obvious how to look at it (maybe I’m not a pro user after only a month, but that’s the point; why do I have to be?)

The purple circle at the top indicates how up to date the data sync is from the Pendant. The box at the bottom lets you query that data.

Given the existence of services like Gracenote and IMDB, the device could recognize consumed media and just label it and isolate it from the good stuff, rather than trying to summarize it like everything else. As well, it could determine where you are (if it had GPS) and sync with your calendar to help label conversations. It does sync with Google Calendar, but of course I use both that and Microsoft calendar which it doesn’t sync with. I guess the thought I have as I use this device: Didn’t the product designers think about these usability issues, particularly since they position it as freeing the human mind from biological limitations? (Maybe, like a lot of startups, they were in a hurry to ship since they pre-sold 10,000 units.)

And then, what is the actual usefulness of asking about what the device has heard and recorded? One of my most successful prompts was “What did I promise Alice during our lunch?” (You know who you are, Alice!) It came back with a list of what commitments I made, and I was able to follow up on them. And Alice thinks I’m a superstar. (Well, maybe; she did go buy her own Limitless Pendant.)

The app provides sample prompts like “What did I learn about most yesterday?” (a grammar checker would be helpful), “generate a todo list for next week” (what if you have weeks of data; does it search everything or only the last week?). When you do write a prompt, whether custom or a sample, the results take whole minutes to come back. It ain’t Google search or ChatGPT.

I’m going into details here because, if an AI device is going to be helpful, the details really, really matter. So far, I am not depending on my Pendant. It has not changed my usual habits. I hoped it might replace my todo list, which isn’t hard since I forget to put half of what I’m supposed to do on my todo list. I’m going to keep using the Limitless Pendant, in the hope that the company might improve its usefulness. Or maybe I’ll figure out how to use it correctly. So it won’t join My Pile Of Unused Devices just yet.

That’s precisely the reason that I took the time to review this product. Is this the kind of thing that Jony and Sam are thinking about? Does Sam worry that he might put his portfolio company, Limitless.ai, out of business? Or will OpenAI buy it for a ridiculous price (like anything close to what it paid for “io”)?

I’ve been looking for AI that will help me in my daily life. Apple actually predicted what I really want with Apple Intelligence, even if they couldn’t deliver it (which lead me to recommend that Tim Cook initiate his succession plan). But maybe Apple should buy Limitless to get it started with delivering personally useful and intelligent features without an iPhone or computer? Net net, I have to give props to the founders and team at Limitless for giving us a peak into what a real AI device looks and feels like!

PS: Joanna Stern at the Wall Street Journal reviewed the Limitless Pendant (gift link) but was more focused on something called Bee Pioneer, which is less expensive and sits on your wrist. She seemed to like the Bee more than the Pendant.