Exclusive Interview with XGIMI's Shao Zijian: No Camera, No Full-Color for Now—The Design Trade-offs Behind XGIMI's AI Glasses

English text is being prepared.

“We would define ourselves as… a company focused on personal intelligence, something like that.”

On November 1, at the INNO100 first “Hard” gift event where the full lineup of XGIMI’s AI new products was unveiled, Shao Zijian, CPO of XGIMI’s smart glasses business, gave this answer when discussing the positioning of the new business.

This answer is quite interesting. After all, for most consumers, when XGIMI is mentioned, the first thing that comes to mind is still smart projectors. Founded in 2013 and listed on the STAR Market in 2021, XGIMI Technology has now grown into a listed company with annual revenue of nearly 3.5 billion yuan. According to its 2025 financial report, its projector complete-machine business revenue was approximately 3.034 billion yuan, accounting for 87.5% of total revenue.

Shao Zijian, CPO of XGIMI’s smart glasses business

But with the rapid development of AI Agents, XGIMI clearly has new ideas. In October 2025, XGIMI established QINGLAI Innovation to comprehensively advance AI hardware. For a company that started with smart projectors and grew into a listed company, this can, to some extent, also be seen as a “second startup” launched by XGIMI in the AI era.

And this offline event happened to let us see the initial results of this startup. At the site, XGIMI showcased three products: the AI display glasses MemoMind One, the AI memory glasses G1, and the AI memory band L1. They adopt different technological forms, yet all revolve around XGIMI’s self-developed personal long-term memory system. Beyond the products, through our conversation with Shao Zijian, we also further learned about XGIMI’s considerations for entering the AI glasses market, the trade-offs in its product technology route, and this listed company’s judgment on the future development of personal intelligence.

1. With the rise of AI Agents in 2025, XGIMI saw new growth opportunities

As an enterprise that has struggled in the smart projector field for more than ten years, XGIMI began laying out in the AI glasses market in 2025, the first year of the “battle of a hundred glasses.”

When talking about this timing for entering the product category, Shao Zijian said: “In early 2025, after general-purpose Agents began to be productized, we felt that AI was beginning to mature. It may not yet be the best experience, but it already has some capabilities, which made us begin to enter this field and do this.”

At the same time, from the supply chain perspective, the miniaturization of Micro-LED microdisplay devices and the continuous improvement of waveguide display effects have also provided conditions for the productization of lightweight AI glasses.

In Shao Zijian’s view, previous large-model applications mostly handled functions such as search and single-turn Q&A, while general-purpose Agents have gradually gained the ability to understand context, call tools, and complete tasks, giving AI a chance to truly participate in users’ daily lives.

For XGIMI itself, the optical, acoustic, hardware structure, and consumer electronics supply chain capabilities accumulated over the past decade or more also provide a foundation for entering this market.

The in-depth cooperation with ZEISS Optics this time is a typical example. At the event, Alex Zhou, head of XRX products for ZEISS Optics China, introduced the ZEISS Lingsi lenses debuting on MemoMind One. In response to the characteristic that AI display glasses need to simultaneously handle virtual information display and real-world vision correction, ZEISS made corresponding optimizations in lens optical design, coating, and precision assembly.

Alex Zhou, head of XRX products for ZEISS Optics China

According to the introduction, the lens’s clear field of view is on average about 3 times that of traditional aspheric single-vision lenses, the lens’s double-sided transmittance is about 96.5%, and the double-sided reflectance is about 1.4%, in order to reduce the impact of additional reflection and stray light. In addition, the bonding process between the lens and the waveguide can achieve ±10 micron-level dispensing precision.

It is worth mentioning that the cooperation between the two parties is not limited to product technology, but also extends to sales channels and service systems. In the interview, Shao Zijian revealed that some ZEISS offline stores have already begun selling XGIMI’s related products, and the two parties are also jointly building pre-sales and after-sales service systems. “Both parties are highly consistent in their understanding of goals and products, which is why we have in-depth cooperation not only in products, but also in channels,” Shao Zijian said.

Within XGIMI, the AI hardware business has also been given an important strategic position. Shao Zijian said: “This product is actually a very important business in our internal strategy. Because our projectors carry the role of naturally displaying content in space, and next we hope to display personal intelligence more naturally to every user.”

2. Starting at 2,294 yuan, without a camera—what technical trade-offs did XGIMI make for its AI glasses?

In XGIMI’s personal intelligent product layout, the AI display glasses MemoMind One are undoubtedly the product with the highest technical value and one of the most representative products in the industry.

These glasses adopt a binocular monochrome green Micro-LED and diffractive optical waveguide display solution, with a single-eye resolution of 640×480, maximum display brightness of 2000 nits, and a field of view of 25°, and can achieve functions such as message notifications, AR navigation, teleprompting, real-time translation, AI subtitles, and meeting records. The lightest whole device is about 46.6 grams, equipped with a Harman acoustic solution and dual speakers, and has a comprehensive battery life of 16 hours.

In terms of price, the square-frame and round-frame plano-lens versions have a subsidized final price of 2,294 yuan; the version equipped with ZEISS Lingsi 1.67 refractive index single-vision resin lenses is 3,699 yuan, and the 1.74 refractive index version is 4,499 yuan.

In terms of product configuration, XGIMI made several notable trade-offs. For example, at a time when cameras on AI glasses are becoming increasingly common, the MemoMind One does not carry a camera. At the same time, XGIMI chose a relatively small monochrome green Micro-LED display solution, yet retained dual speakers.

In his on-site presentation, Shao Zijian repeatedly emphasized that XGIMI wants first to make a good pair of glasses, and only then consider what value AI can deliver. Regarding the camera, he explained that although adding a camera could bring more functions such as photography and visual understanding, it would also mean extra weight, power consumption, and wearing cost. Therefore, before adding any component, the team evaluates whether it is worth the corresponding cost to users.

As for the choice among display technology paths such as Micro-LED and LCoS, Shao Zijian's answer was also quite interesting. "What we care about more is the experience users want, and what kind of technology or product path is needed to achieve it. So we hope to use the right technology at the right time."

He believes that at this stage XGIMI's primary goal is to ensure the product is light enough. Compared with solutions such as LCoS and DLP, the team believes that Micro-LED currently better meets its requirements for optical engine size.

Of course, an optical engine that is small enough is only part of the everyday wearing experience of AI glasses. For a pair of glasses with display functionality, whether the display grating is obvious, whether there is light leakage on the front and sides, and the transmittance of the lenses themselves all affect the final product experience.

Shao Zijian specifically mentioned that these issues actually cannot be fully left to upstream optical waveguide suppliers to solve, nor are they purely supply chain technology. Because the incident angle of a diffractive optical waveguide is closely related to the outgoing light, XGIMI needs to work with the supply chain to adjust the optical engine incident angle, optical waveguide coating, and overall optical structure of the device based on users' head posture during normal conversation, viewing angle, and usage scenarios.

In addition to the MemoMind One, two other XGIMI AI hardware products were also showcased on site: the AI Memory Glasses G1 and the AI Memory Band L1.

The two products officially went on sale on October 28. Among them, the G1 adopts a lightweight design with no display and no camera, offering two styles: retro round frames and business square frames. The lightest product is about 17.9 grams, and the standard version has a first-release price of 1,299 yuan; the L1 adopts a dual-form design as a band and a pendant, with the standard version priced at 599 yuan and the lifetime membership version at 1,059 yuan. (For more review details: "Starting at 599 yuan, XGIMI remembers to put 'memory' into glasses and bands")

The G1 and L1 mainly focus on functions such as daily conversation recording, meeting minutes, to-do reminders, and AI memory management. Although the three existing AI hardware products differ slightly in form, the one thing that remains unchanged is the personal intelligence product positioning that XGIMI hopes to establish.

In XGIMI's view, existing personal intelligence needs to be built around long-term AI memory at its core, thereby extending to more scenarios. Shao Zijian spent considerable time explaining this issue. "What we call memory is not about storing users' records, such as verbatim transcripts of meeting minutes, exactly as they are, one after another. That is not called memory; that is pure recording."

He further explained: "The memory we want has the ability to automatically integrate and refine, extract key information, and then organize events in a coherent thread. When I ask about something, it can find those events through this thread and then extract information from coarse to fine."

According to XGIMI's vision, a long-term memory system needs to continuously understand and update user information. For example, if a user discusses the progress of a project today and communicates new adjustments the next day, the AI needs to identify the connection between these pieces of information and update the status of the corresponding event. When the user mentions related content again, the system can find the latest information.

Shao Zijian revealed that although there are already many open-source long-term memory projects on GitHub, XGIMI did not directly apply existing solutions. Instead, it redesigned a memory system architecture around the actual usage scenarios of its own products, focusing on solving issues such as information refinement, continuous updating, and accurate recall.

This self-developed approach based on its own needs is also one of the keys to XGIMI's attempt to build product differentiation. After all, compared with simply integrating third-party memory services, a long-term memory system that can continuously iterate and adapt to different AI hardware has a better chance of becoming the core capability of its personal intelligence product ecosystem.

At the same time, XGIMI is also trying to connect the personal memory system with more AI applications. Shao Zijian introduced that XGIMI opened its SDK and underlying BLE communication protocol to global developers on September 21, allowing developers to use the glasses' display, microphone, speakers, and other capabilities to develop applications. XGIMI also plans to build a corresponding app market later.

It is worth mentioning that at the event, XGIMI also demonstrated the integration of the MemoMind One with a third-party Kiwear AI ring. By wearing the ring, users can directly control the glasses, enabling another form of interaction. This also showed us more possibilities for XGIMI AI glasses in third-party hardware accessories and the interaction ecosystem.

In addition to hardware interaction, XGIMI also plans to further open the MCP interface of the long-term memory system. Shao Zijian said: "We do not want long-term memory to be a closed set of context, confined only within our system." According to his vision, in the future users will be able to call relevant memory context through their commonly used AI Agents (such as Workbuddy, clawbot, etc.), allowing different applications to utilize information accumulated by individuals over the long term.

IV. Single-green displays still have a window of opportunity, while full-color adoption still faces barriers

As full-color optical engines continue to shrink, the weight of some products has already approached that of single-green display glasses. This also raises a question worth discussing: how long can the market space for single-green displays last?

In his view, single-green Micro-LED optical engines still have room to continue shrinking, reduce costs, and improve display effects. For AI glasses focused on information prompts, translation, teleprompting, and navigation, single-green displays can still meet many practical needs.

Regarding the trend of full-color products continuously reducing weight, even approaching 50 grams, Shao Zijian believes that judgment needs to return to the target audiences of different products. He said that 50 grams may already be a relatively light weight among current full-color glasses, but according to XGIMI's requirements for all-weather wearing experience, this weight still has room for further optimization. In addition, as display functions increase, full-color products may also impose higher requirements on power consumption and battery capacity, which in turn affects the overall weight, appearance, and battery life.

However, this does not mean XGIMI denies the development prospects of full-color technology. Shao Zijian believes that in the future, single-green and full-color displays may continue to cover different needs. The former can leverage advantages in lightweight design, information prompts, and cost, while the latter can provide richer visual content and interaction forms.

"To judge whether a product is good may not be about what functions it has, but whether it can provide help very smoothly and very coherently in whatever scenario." This echoes XGIMI's current key investment in a personal long-term memory system.

Final thoughts

He said that at this stage, glasses are a category that XGIMI believes is worth investing in and continuously refining, but whether to expand into other wearable devices in the future still depends on user needs. For example, in terms of interaction, he believes smart rings are expected to become important accessories for AI glasses, and the linkage between wristbands and glasses also has potential for further exploration.

As for the goal for the coming year, Shao Zijian said he hopes XGIMI can achieve more obvious growth in user scale to verify the market acceptance of this generation of products.

Submissions/tips: tougao@youxituoluo.com

Article/business cooperation: Lin Nan (WeChat 19250561593)

Join the industry exchange group: Lin Nan (WeChat 19250561593)

Copyright notice: This article is original by VR Tuoluo, and no third party may reproduce it without authorization. To reproduce, please contact WeChat: vrtuoluo233 to apply for authorization, and retain the reproduction source, author, and original link information when reproducing. Do not change the content without authorization; legal action will be pursued for violations. Images and video materials in this article come from the internet or sources that cannot be verified. If there are copyright issues, please contact this website to negotiate handling.

Digital technology industry information and service platform

Download the "Tuoluo Technology" APP to obtain cutting-edge in-depth digital technology industry information