In the AI Era, What Hardware Will the Gateway to the Human Mind Be Built Into?

Seizing the gateway to the human mind in the AI era is, at bottom, a secret war over token throughput power.

Author: Koutian Wu; GitHub: ktwu01

My friend boj shared before the Limitless recording pendant he ordered a year or two ago. Back then it only cost him a few dozen dollars, and for various logistics and customs reasons he completely forgot about it, until suddenly, two years later, it cleared customs and arrived.

That evening he wore the pendant to a hotpot dinner, and came back saying the effect was absurdly good. In a noisy, steaming hotpot restaurant, the pendant correctly recognized the majority of what they said, and once he turned on think mode, the AI analyzed the various details in the recording with perfect orderliness.

The Limitless pendant came with a small card bearing a sentence, roughly saying their vision is to liberate the human mind from biological limitations.

That sentence really struck me.

Isn’t that exactly the vision every AI developer and product manager today dreams of?

But after the excitement of these past few days passed, I started mulling another question.

To liberate us from biological limitations, what hardware do we actually need, and what is the true AI gateway?

In the traditional mobile-internet era, we were no stranger to the word “gateway.”

The browser is a gateway, the app store is a gateway, WeChat is a gateway. Whoever holds the gateway holds the distribution rights over traffic, which means control over users’ attention and time.

But besides these visible software gateways, there’s a more underlying, almost-forgotten “physical gateway”: submarine cables.

Although we feel the internet is “wireless,” in reality more than 99% of international data traffic travels through optical cables buried deep beneath the ocean (submarine cables). Giants like Google have long stopped being just software; they’ve personally become main builders of the internet’s “physical infrastructure.” They’ve invested in dozens of dedicated submarine cables worldwide, directly connecting Asia, North America, and Europe.

This grasp over physical transmission rights translates directly into a terrifying cost advantage. You can stream YouTube flawlessly anywhere on Earth, and for Google, YouTube’s bandwidth cost is nearly zero. That’s the kind of cost chasm that platforms like Bilibili or other domestic video sites can probably never cross: every year they pay operators hundreds of millions in “bandwidth tolls.”

This grip on “gateways” works at the system level too. Google holds mobile via Android and desktops via Chrome. So when Apple (Apple) swung the stick of iOS privacy policy and knocked Meta (Facebook) and Snap around, Google was barely affected. That’s also why Zuckerberg is now betting the whole company on Meta’s hardware: he’s tired of living under someone else’s roof, and he must own his own traffic gateway.

But in the AI era, the definition of a gateway has changed.

Today’s AI agents no longer interact with us through polished UIs; they complete tasks for us directly by throughputting tokens.

You input sound, image, text; it outputs thought, instructions, actions.

From this angle, in the future AI era, whoever can capture the input tokens of our daily life most directly and with the least friction, and output the delivery tokens we want fastest and most accurately, is the true system-level gateway to the human mind.

This is no longer a fight over traffic but seamless custody of attention and thinking.

To seize this gateway to the mind, everyone is now frantically building hardware.

Various strange carriers have popped up in the market, including voice recorders, smart glasses, smart wristbands, and smart rings.

I’ve talked with quite a few friends doing hardware and doing AI, and opinions diverge sharply on these carriers.

Let’s take these four carriers one by one and examine where their game points actually are.

The first is the voice recorder or pendant, like the Limitless boj got.

This thing’s advantage is obvious: almost zero friction. One charge records for over a dozen hours; hung around the neck or pinned to your clothes, you can completely ignore its existence. It quietly records every word you hear and say, becoming your second brain.

But its fatal flaw is equally deadly: it has no eyes, and no good proactive output channel. It can only passively record sound. If you want to interact with it frequently in public, talking loudly makes you look like a madman.

The second is smart glasses.

Among my friends who’ve tried smart glasses, the feedback is indeed stunning. Not only do they have a voice recorder’s function, but the coolest thing is they have a pair of eyes exactly like ours.

Just think: glasses can see everything you see and hear everything you hear. First-person-perspective visual input is the richest, highest-dimensional token input source in the world. No need to pull out your phone to photograph or type a description; they directly share the context on your retina with you.

But the pain point of glasses is also that they’re too heavy, battery life is hard to push through, and in many social settings, wearing glasses that can record and film at any moment makes the person across from you deeply uncomfortable. More importantly, how do we control them? Do we really have to shout commands out loud every time?

At this point, the third carrier, the smart wristband or watch, and the fourth carrier, the smart ring, reveal their unique value.

Wristbands and rings probably shouldn’t be seen as standalone AI hardware; they’re more like the control center of this sensory network.

A ring worn on the finger can serve as a physical switch for the whole smart home or other devices, and can even achieve blind operation through tiny finger movements.

A wristband goes further: it stays on our wrist year-round, not only monitoring vitals, but also enabling silent input through subtle gestures or bone-conduction typing when the glasses need to interact quietly.

Continuing this inventory, you hit an interesting conclusion.

No single piece of hardware can, on its own, become the perfect AI gateway.

The real AI-era gateway will absolutely not be yet another plastic-heavy standalone device you have to fumble with in your hands, let alone the glowing glass brick that currently traps us in our chairs.

The screen is the era’s biggest divider.

In our last blog post about Steve Jobs, we discussed how Jobs killed the physical keyboard with multi-touch; if he were here today, he’d certainly find a way to kill the screen.

The future AI gateway is actually a distributed sensory network.

It’s a pair of lightweight smart glasses sharing a first-person view for seeing, a body-hugging recording pendant for hearing, a ring or wristband on the hand for silent control, paired with an invisible earbud for feedback.

This network wraps around us like a layer of invisible air.

No need to look down at a screen, no need to tap open apps blinking with red dots, no need to anxiously hop between different SaaS panels.

You just live, observe, and speak as usual; AI silently captures those tokens behind your back and returns your laundry washed and folded.

That’s the real zero friction. That’s the real liberation of the human mind from biological limitations.

Here’s a heads-up for friends building AI products and hardware startups.

Stop thinking about how to make a better phone app, or how to grab territory inside the smartphone.

We need to adapt to a new ecosystem with no screen and no explicit UI interaction.

If your service can’t be silently called through an API by this distributed sensory network and return accurate results within three seconds, then in the future’s intelligent ecosystem, your product simply doesn’t exist.

Truly valuable AI products should be invisible.

It’s like a redesigned bicycle for the mind.

Forty years ago, Jobs said the computer is a bicycle for the mind, taking us forward at unimaginable efficiency.

But over the past decades, this bicycle got loaded down with too many fussy dashboards and buttons to learn, so ordinary people just feel tired riding it.

Now AI finally gives us a chance to strip away all the buttons and keep only the pure riding experience.

You just pedal and it carries you flying, without you even sensing the bicycle exists.

That’s the change we’re witnessing.

It’s not cool tech showmanship; it’s an everyday magic that quietly rescues the human mind from biological forgetfulness and sluggishness.

Honestly, I can hardly wait.