Three Diverging Paths of Smart Glasses
Part 2 of Series Beyond the Lens: China’s Smart Glasses and the AI Future
In the first piece of this series, I argued that AI glasses won’t replace the smartphone—not because they can’t, but because we still fundamentally think of AI as a “tool.” To shift AI from tool to companion, we need new devices, new rituals, and most importantly, new expectations. Smart glasses are not just another gadget. They’re the first serious attempt to give AI a body.
But here’s the thing: not all AI glasses are created equal. Even though everyone calls them the same thing, they’re not racing on the same track. Today, we see at least three major design philosophies hidden under the same label: utility-first, platform-oriented, and agent-native smart glasses. Each is a bet on a different future. Each comes with its own logic, trade-offs, and ceiling.
This essay is about those three divergent paths.
The Pragmatist Path: AI as a Utility Layer
This is the Xiaomi approach. Rokid, too, follows a similar pattern. These companies are making AI glasses that don’t try to reinvent your life. They just add a little intelligence to tasks you already perform.
The key characteristics? Light hardware, modest sensors, voice interaction, and low price points. They translate text, answer simple queries, perhaps even summarize a meeting. But at the end of the day, they’re just a little smarter than a Bluetooth headset.
That’s not a bad thing. It’s a marketable thing. These devices are quick to ship, easy to demonstrate, and appeal to early adopters and hardware enthusiasts. They’re also the most likely to scale in the short term.
But there’s a ceiling. These devices don’t ask much of the user—and they don’t build new behavior either. They’re not replacing the phone. They’re not making you think differently about AI. They’re just gadgets with a thin layer of smarts. That might win sales, but it won’t win the future.
The Spatial Path: AI as Operating System
This is where companies like Meta (and to an extent, Apple with Vision Pro) are aiming. Glasses here are no longer just a tool—they’re a portal to a new spatial computing environment.
These products imagine a world where your physical surroundings are filled with digital layers. Contextual awareness, persistent UI, real-time scene understanding—this is about reconstructing space itself as a computing surface.
But here’s the challenge: that dream is heavy. It requires high-end sensors, expensive chips, cloud tethering, and a new set of developer tools. These glasses often can’t survive on their own; they need a phone, a cloud link, or a full ecosystem.
And that’s the paradox. In trying to be the next OS, they lose their autonomy. They become accessories to something else. The ecosystem might be rich, but the glasses rarely feel like the protagonist.
Still, this path is worth watching. Because if someone does pull it off, we might finally get our first spatial-first AI platform—not a smarter assistant, but a smarter world.
The Agent Path: AI as Personality
This is the most ambitious and least understood direction. What if glasses are not a tool or a window—but a vessel for a real-time AI character? A digital being you talk to, rely on, even identify with?
We’re seeing early steps in this direction: Rabbit r1, Limitless pendant, and some agent-first experiments in China. They’re small, imperfect, and often misunderstood. But they represent something new: a relationship with AI.
These devices aren’t trying to do everything. They’re trying to become someone. Someone you bring with you, talk to, offload memory to, and even trust.
If this paradigm wins, then smart glasses will become containers for AI agents. Not apps. Not features. Personalities. And the business model, the UX, the success metrics—they all flip. It’s not about what the device can do, but about how much trust you place in what’s behind the lens.
Conclusion: Three Designs, Three Futures
These three approaches aren’t just design choices. They’re bets on how AI will integrate with human behavior.
Utility-first: AI as a helpful tool. Easy to adopt, but unlikely to transform.
Platform-first: AI as an immersive OS. High potential, high friction.
Agent-first: AI as relationship. Intimate, volatile, and potentially explosive.
So if you’re building or investing in AI glasses, the question isn’t who has the best hardware. It’s: Which future are you building for?
Not all paths lead to the same destination. Some just look similar—until the road splits.