01
Crossing the distance between tool and assistant
Over the past few years, AI glasses have left a rather poor impression on consumers. They are like a voice assistant that needs to be constantly woken up — worn on your face, yet only present when you actively call out to it. After the novelty of buying them wears off, the long-term usage rate of most products is astonishingly low. Li Hongwei, CEO of Rokid, once publicly stated that in 2025, over 200 new features were launched across various AI glasses, yet the proportion of users using them long-term was less than 6%. This time, Qwen chose a different path. The N1 Pro comes with an all-day visual perception system, relying on a dedicated low-power chip to continuously "watch" the surrounding environment. When you arrive at the train station carrying a suitcase, it automatically brings up your trip order, tells you where the ticket gate is when going through security, and tells you which carriage and seat number once you're on the platform. Throughout, you don't need to take out your phone or say a single word to it. Inside the frame near the nose pad, the N1 Pro hides an eye-tracking module, only 1.7 millimeters on each side, capable of stably tracking gaze across a lighting range from 0 to 100,000 lux, with accuracy in the central area controlled within 1 degree. What does 1-degree accuracy mean? When you stare at a dish on a menu and ask "how much is this," it truly knows which one you're asking about. The realization of this "proactive service" capability relies on a substantial improvement in on-device AI. The N1 series carries three chips working in coordination: the Snapdragon AR1 handles high-performance image processing and complex computing, a Bestechnic chip handles everyday tasks, and a Himax chip handles low-power visual understanding. The intent of the three-chip architecture is clear — to let the glasses maintain constant awareness of the environment without sacrificing battery life. There are also health monitoring capabilities. The N1 Pro is the industry's first AI glasses with real-time vital sign monitoring, capable of tracking metrics such as heart rate, blood oxygen, and skin temperature, and translating these data into easy-to-understand status interpretations. While cycling, the glasses can remind users to adjust their breathing rhythm based on heart rate changes; in daily wear, they can also record eye usage, helping users adjust their eye-use rhythm and reduce prolonged continuous screen time. When stress indicators run high, the AI healing feature generates customized meditation audio and continuously evaluates relaxation effects, forming a complete closed loop of monitoring, early warning, and intervention. Placing these capabilities on a pair of glasses means it is no longer just an information tool, but begins to possess a kind of "presence" — it is always by your side, always perceiving your state. The real dividing line between a tool and an assistant has never been about understanding a few more spoken sentences. The leap from "you command it" to "it proactively judges what you need" is, in fact, the distance from a tool to an assistant.02
Glasses are just the entry point; the ecosystem is the answer.
If you judge the Qianwen AI glasses purely on hardware specs, it is easy to conclude that they are "pretty good, but not particularly disruptive." But the logic behind Qianwen making hardware has never been about the hardware itself. Song Gang, Vice President of Alibaba Group's Qianwen Business Unit, pointed to this intent at the launch event: "AI wearables are Qianwen's extension into the physical world and an important component of the Personal Agent. No matter where you initiate a request, behind it is the same Qianwen, understanding the same person and continuing the same goal." To translate: Qianwen's starting point in betting on hardware is to help the large language model understand you better—the hardware itself is only a means. The glasses and earphones are its "eyes" and "ears," used to continuously capture contextual information from the physical world, and then to permeate it with the services and task capabilities already available in the Qianwen App. What underpins this logic is "a customized large language model plus more than a decade of real ecosystem data from Alibaba." Lai Jianhua, training lead for South China at Alibaba's Qianwen Business Unit, gave an example: ask the AI to generate a 5-day, 3,000-yuan travel itinerary for Guangzhou, and it will offer local details such as "3-day metro pass at 50 yuan for unlimited rides within 72 hours"; by comparison, a general-purpose model's answer often stops at the level of "visit Canton Tower, eat morning tea." The truly useful local information is exactly what a large language model cannot learn from training on public data. Alibaba holds more than a decade of real consumption and travel data accumulated by Amap, Taobao, Fliggy and Alipay, and the dimensions, frequency and authenticity of this data form the moat that distinguishes Qianwen from pure model vendors. For this data advantage to truly take effect, it needs a vehicle for implementation, and the glasses and earphones are that vehicle. When users go out wearing the glasses, navigating with Amap, paying with Alipay, booking tickets with Fliggy—all of these behaviors continuously provide context for Qianwen's Personal Agent. This logic is even more directly reflected in the Qianwen AI clip-on earphones. These earphones, launched jointly by Qianwen and Bose, support real-time translation in 89 languages, AI note-taking and handling tasks by voice, but what truly sets them apart from similar products is their ability to perform life services like "paying utility bills and parking fees." The source of these capabilities lies in the Alibaba ecosystem behind them; the earphones themselves are merely the shell that carries them.Qwen's competitive strategy in AI hardware is fundamentally different from that of pure hardware manufacturers. What it competes on is not the hardware specs of a single device, but the comprehensive capability of "large models + service ecosystem + hardware engineering."The advantage of this strategy is that the hardware can operate at a slim margin or even at a loss, but once users get used to using Qianwen's hardware to handle daily tasks, they become locked into Qianwen's Agent ecosystem. The hardware is the cost, and the services are the profit. Looking at industry data, this window of opportunity is opening. In 2025, global shipments of AI glasses reached 8.7 million units, nearly a 25-fold increase over three years, and in 2026 they are expected to exceed 10 million units. China's smart glasses market is expected to ship 4.51 million units in 2026, a year-on-year increase of 78%. More importantly, in 2026 AI glasses were included in national subsidies for the first time, joining smartphones, tablets, and smartwatches as one of the four major subsidized product categories. The policy signal combined with market growth gives this track the conditions to shift from 'niche early adoption' to 'mainstream everyday use'.03
From Early Adoption to Daily Use: What's Still Missing
Hardware specifications determine the floor; real-world deployment determines the ceiling. When the market window opens, what truly determines whether an AI glasses product can keep users is what it can actually accomplish in real life. Qwen has packed its answer down solid: ordering food delivery, hailing a ride, producing meeting minutes, generating research reports... None of these scenarios alone counts as a disruptive innovation, but the signal they send together is clear: the value of AI glasses lies not in how precise its sensors are or how high-definition its camera is, but in whether, at the moment you need it, it happens to know what you need and gets the job done for you. This is probably the plainest yet most weighty meaning of a Personal Agent. And this definition is taking concrete shape in everyday scenarios such as office work and cross-language communication: for example, during meetings, the Qwen AI glasses' meeting-minutes capability supports mutual translation among four languages—Chinese, English, Japanese, and Korean—and can recognize specialized terminology from different industries, with output dimensions including interview summaries, core insights, memorable quotes, and to-do items, and can even generate mind maps and chapter quick views. For cross-language conversations, the Qwen AI clip-on earphones' translation capability covers 89 languages and supports four modes: simultaneous interpretation, face-to-face translation, call translation, and media translation. In face-to-face translation, one earphone is given to the other person, and both parties speak their native language, with no need to take turns holding down a button to speak. According to Lai Jianhua, people who buy translation devices in bulk before traveling abroad are becoming early users of this batch of hardware. Looking back from these scenarios, Qwen's logic for making AI hardware is already clear: the hardware itself is only a carrier; the real value lies in releasing the real data and service capabilities that Alibaba's ecosystem has accumulated over more than a decade through a personal, close-at-hand device. Compared withOthers make wearables to sell hardware; it makes hardware to find a foothold in the physical world for large models.This narrative is fully self-consistent within the Alibaba system, because Taobao, Tmall, Amap, and Fliggy have accumulated over a decade of real commercial data, giving the Agent a depth of services that other vendors can hardly replicate. But ecosystem self-consistency and user willingness have never been equal signs; the product itself must be good enough to serve as the bridge. However great the data advantage, users still have to be willing to wear it every day. And the prerequisite for users to keep wearing it is that these capabilities can truly fit into their day.When glasses can navigate, translate, take notes, and pay for you, and even offer reminders before you speak, the Agent is no longer a distant concept but something you wear on your body every day. This may be the core intent of Qwen making AI hardware: not to sell a pair of glasses, but to let the Agent first become part of daily life.
(Lei Feng Network)
(Lei Feng Network (WeChat account: Lei Feng Network))
Related articles:
Alibaba Qwen gets on board the BMW iX3
Paid users exceed 20%, e-commerce merchants "flock" to Qwen Office
Qwen Office launches e-commerce edition, opens invitation-only testing
Qwen confirms it will launch "Qwen Go!", experience applications now open
Qianwen Office will integrate with Salesforce on Alibaba Cloud
A hardware card and a battle over context: Qianwen Office bets on enterprise Agents
This is an original article by Leifeng.com. Unauthorized reproduction is prohibited. For details, seeReprinting Notice。