If you are interested to get this project developed further, cosider starring it on my github: life-autopilot-agentic

WITH the way things stand, AI pins promised the FUTURE, but the technology simply couldn’t keep up with the demands and expectations. We expected these devices to be as responsive as our phones and watches. Yet many of them became little more dead weight, so much so, that you’re better off just hanging it on the wall as a moment in history.

It’s easy to speak to someone on loudspeaker or through an earbud because we get instant response. These AI pins promised the same convenience and more. but to speak to an AI model that takes longer than 3 seconds to response, quickly becomes uncomfortable for others around you, and probably you too. Even the addition of screens and laser projectors, while visually impressive, did little to improve the actual usefulness of the products. The vision was exciting, but with the combination of systems contained in this tiny device with a tiny battery, it was bound to be an immense challenge. for one, overheating was a problem, which personally is a shocker, considering we have smartwatches. Granted this ‘AI pin’ industry is very new, so these kinks are inevitable. But do those kinks mean that there is no place for ai pins in this era?

If They Failed So Badly, Why Did Everyone Want One?

Firstly, based on the response received when these products were announced, is a clear indication that the desire to use these devices are there, not only were investors invested, but also consumers. Secondly, considering that we have strong performing smartwatches loaded with various sensors and devices, it would not be hard to recreate a product of a similar form factor. Smartwatches have screens, they are packed with sensors, they have mics and speakers, they can use wifi and data, and they have all day battery life. so these watches are more than capable of functioning like an ai pin, so what stopping anyone from making a specialized ai pin, you might think. It would be nice to have this reliable ai system that is always aware of what’s happening around you and ready to help you out as we were promised.

The Question That Started This Deep Dive

My goal is to understand what resulted in these products performing so poorly despite generating so much excitement. To do that, we'll look at the setup of each product, compare smartwatches and AI pins, and identify where things went wrong. These deep-dives will reveal if there was a better way to execute these products, or if the large response seen by the announcement of these product was just a misleading sign to the demand of this innovative product. The comparison will be between the infamous Humane AI pin and the Apple smartwatch, because it seems fitting that the founders of Humane were former Apple workers.

Now these ‘deep-dives’ that I will do, will basically be surface level observations, because the most common reason for a products success or failure are things that have a direct impact on experience. So therefore, I will examine first the hardware, things like material and parts. After that I will examine the usability of each part of the product then the usability of the whole product. Then the last part I will look closer at is conveniency, how did the product strive to offer convenience and how was the overall outcome of its attempt to do so. Then in the end I will show and tell you about my own take on how ai pins should be done.

Humane Built the AI Pin Everyone Was Talking About

In comparison to the long history that smartwatches have, being that the modern smartwatch first appeared in 2013 by a kickstarter, ai pins are just getting started. they were only introduced in 2023, just three years ago. The Humane AI pin appeared to be like the apple of an innovative ai forward product. It look luxurious, it had no screen but a laser , it seemed to be very responsive according to the ads, and they were made by ex-Apple veterans. But I think that the fact that it was made by former ‘Apple’ employees, it was almost like a sign that the AI pin would flop, considering that Apple itself basically flopped hard in its AI development compared to its competitors. Humane built their AI pin to be

• a smartwatch-sized device made of aluminum which attached magnetically to clothing,
• with no screen, but with a laser-ink projector as the interface.
• Obviously it had a small battery.
• It had a wide-angle camera to constantly stream to the AI.
• Some mics to chat to it.
• 4G sim connectivity and some modern dual-band wifi and bluetooth.
• Also they packed a smartphone-grade processor into it.

The Red Flags Started Immediately

Now already off the mark there are too many red flags about this. Why use a smartphone-grade processor for such a tiny device we aren’t there yet. there are processors like the Snapdragon W processors, specially made for wearables so the battery does not tank after seconds of use. Also it dangling on a t-shirt would obviously become a bother, and the magnets would not be strong enough to hold on something like a thick jacket. This is where they let their creativity over crowd their intuition of common product experience.

A Companion That Never Stopped Watching

The ai pin had gesture tracking via the camera which again, was constantly streaming, so it was always active, burning power, just waiting for you to do a gesture, or ask a question about something you are showing it or what it saw earlier in the day. It was positioned to be a companion but it was not built to be one. As the camera is live, it is also querying the ai about what was observed, and this is done both on the device as well as across the internet. So the battery is taking a beating and the querying time is an eternity. A more obvious solution would be to pair it with a smartphone, but at that point you might as well use the phone, lol.

The Coolest Feature Was Also a Gimmick

The edition of the laser projector is another case of humane letting creativity flow uncontrollably with no consideration of usability. While the idea looked futuristic in demonstrations, the projector struggled in real-world lighting conditions. As a result, it became more of a gimmick than a practical interface.

The Smartphone Replacement Problem

Now the sim connectivity and bluetooth bundled into this device seemed to serve no purpose at all in this device. Smartphones are great, they are fundamental. They were built to handle the essential aspects from the beginning and they have been perfected to handle them, and they can handle many more functions.

So proposing a full blown replacement in the size of a ‘pin’ is unrealistic. We have smartwatches and even they have not replaced smartphones.

Just from this breakdown, the whole product is riddled with red flags.

Meanwhile, Apple Took a Completely Different Approach

Now switching to smartwatches, all the way back in 2015, Apple released their smartwatch, and they presented it as a luxury accessory, packed with some trackers.

• They were built to be strapped to the wrist and they had bluetooth, a low energy edition of it, with the main purpose of syncing with you phone, wifi was only added some years later, for scenarios where you did not have you phone in range.
• They were built with an led derived screen
• They had a small battery, and today still do.
• They also were built of aluminum.
• They were packed with sensors and tech like accelerometers, gyroscopes, digital compasses and even heart rate sensors.
• Apple made their own wearable chip, but some other manufacturers used mobile chips but later developed wearable chips too.

Why Smartwatches Actually Worked

As the years went by smartwatches just became more mainstream, thus more features were added and other things were improved in later generations. But we only need to look at the impression that the first version made.

On a small battery, the smartwatch was able to last closer to 20 hours that 10. They were really smart because of the gyroscopes and accelerometers built into them. This allowed unique features to be form by transforming the motion data they tracked. It had a smaller screen so you could do fewer things. Which helped distance us from using smartphones. since they were strapped to our wrists there was some way to dissipate heat. The first edition screens had poor efficiency, so it would really drain the battery over long usage. But that still allowed for hours of usage.

There really is little or even nothing to complain about smartwatches. They were so well built that it led to their unprecedented success. Unlike the AI pin which suffered with latency, efficiency, and redundancy.

The AI Pins That Learned the Right Lesson

But as you might know, humane is not the only AI pin product in the market. Most of the alternatives are screenless wearables that act as a side presence and not the latest thing to replace your smartphone. like

• the Friend ai necklace. Built strictly for companionship rather than productivity, offering unprompted thoughts, commentary on your day, and emotional support as a persistent "friend".
• The limitless pendent, Designed entirely around "life-logging" and professional note-taking

The reason they weren’t the main focus is because they were not talked about as much as the humane ai pin was.

The alternative AI pins really do execute the ideal way the AI pin should be. In the background and helpful.

So What Should an AI Pin Actually Be?

Now if you’ve made it this far, it’s time for me to explain what I think AI pins should be like. The interesting part is that my version doesn’t require any revolutionary hardware. So to get the point across, picture yourself as a recently employed worker at a well-paying company. It's the weekend and you're lying on the couch scrolling through social media. Life is going well. You've settled into your new job, your finances are looking healthier, and for the first time in a while it feels like you're building something for yourself.

While scrolling, you come across a local hiking group organizing a trip for next Saturday. The photos look incredible, the route isn't too far away, and on impulse you decide to join. You save the event and move on with your day.

That's it. You don’t open an app, or create a task, or set reminders, or even tell your assistant, "Hey, remind me about this next week."

What If Your AI Understood What You Were Trying to Do?

Now the AI that actually understands what you're trying to do.

Throughout the week, it quietly pieces together the situation. It knows you work Monday to Friday. It knows how long your commute usually takes. It notices that on weekends you tend to wake up later than you do during the week. It sees where the hiking event is and notices that rain is expected around midday.

Then Friday evening arrives.

Most AI assistants would sit there doing absolutely nothing until you ask a question. This AI doesn't. You're making dinner when it casually speaks through your earbuds.

"Hey, that hiking trip you're going on tomorrow looks like it's going to be fun."

You pause. "Oh yeah, I almost forgot about that."

"I figured. By the way, if you leave around seven fifteen tomorrow you'll get there comfortably before everyone else. Also... maybe pack something waterproof. The forecast changed this afternoon and it's looking like rain around lunch."

The Secret Isn't Faster AI

What it just did was a co-ordinated participation in your life. A few seconds later you ask, "How bad is the rain supposed to be?" It responds, “Enough that you'll probably be annoyed if you're standing on a mountain without a jacket."

You laugh.

"Fair point."

"There's a camping store on the way home from work. They're still open if you need one." What's interesting is that the assistant isn't improvising these answers one question at a time. The moment it decided to tell you about the hike, it had already prepared answers to the questions you were most likely to ask next. Can you see how different that feels. The agent doesn’t overload you with every detail upfront. Instead it gives you the most important info first then predicts what you can about. If you ask a follow up question, the answer appears instantly because it already did the thinking, its now just matching and returning the response.

A Real Assistant Knows When to Speak

The next morning, things get even more interesting. You oversleep. Your alarm goes off, but you hit snooze. Then again. And again. The AI waits for a response but notices that you are not giving a response. So a voice comes through your phone. “Wake up. You know if you keep pretending to be asleep, you're going to miss that hike. "you groan and ask what time is it?" it responds saying, ‘later than you wanted it to be." You sit up. "Oh no."

The Agent tells you still have time. Get dressed, skip the extra coffee stop, and you'll make it. As you're driving, traffic starts building because of roadworks. Before you even notice, the AI speaks again. "Bad news. The route you're on just got slower."

You ask how much slower, and it tells you and simultaneously offering a different route. That's the difference. Most AI products today act like search engines with personalities.

You ask, they answer, you command, they obey.

But that's not how a real assistant works. A real assistant pays attention, understands your objectives, it notices problems while they're still small.

Most importantly, a real assistant knows when to speak. The idea isn't an AI pin that lets me talk to ChatGPT without taking my phone out of my pocket. Because you will. The idea is an AI that understands enough about my life to casually lean over and say,

"Hey, if you leave in the next ten minutes, you'll make it."

Or even, "You've been saying you want to start exercising. The weather is perfect, your calendar is clear, and there's a walking trail five minutes away. i rsvp’d you for tomorrow’s run.”

Not because I asked.

Because it thinks it would help.

The Future of AI Companions

Finding ways it integrate AI systems into our life in meaningful ways is challenging. So it’s best to start with what is already available, and make improvements as we go. I have already done some development on this idea which you can try on your own phone by following the step on my GitHub. AI pins are still being experimented with, different form factors, different use cases, because people are trying to differentiate themselves. But the moment when a set of features has been proven to be necessary, then a complete, well-functioning product can be built, which can then be built in different, ways. Much like there are smartphones from different companies built to be unique but still contain all of the essential features.