Click the blue text to follow us
Recently, I have been focusing on the direction of “robots” and contemplating its possible future.
At the same time, I am also rethinking the positioning of my account.
Future content updates will not pursue frequency but will prioritize quality and depth of thought.
I will share some personal views, musings, and dialogues from time to time,
sometimes industry insights, and sometimes just small talk about life.
Not seeking quantity, but seeking truth;
Not seeking completeness, but seeking specialization and precision.
I hope every updatecan show you a clearer perspective and a more authentic me.
01
AI/AR Glasses
I am not very optimistic about AI glasses and AR glasses. From a technical perspective, the current maturity of the software and hardware ecosystem is insufficient to support a true “grounded experience“. Whether it is computing power, battery life, or interaction ecology, they are still in a transitional stage. Leading manufacturers are also trying to break through. For example, some manufacturers are trying to reduce visible light leakage and lower power consumption through a “periscope” structure; Apple plans to deeply integrate visionOS with iPhone and Mac. However, these efforts are more compromises and trade-offs in usability rather than true innovations.They are a necessary stage from “usable” to “user-friendly”, but we are still far from the level of “indispensable”.From the product definition perspective, the ideal form of smart glasses is a “light intelligent device” that maintains the appearance of ordinary glasses while possessing display, interaction, and even perception capabilities. However, in reality, whether it is Ray-Ban Meta or Xiaomi Smart Glasses, they mostly remain at the functional level of “can only listen, cannot see”.
Everyone is betting on a proposition:
Can a new interaction method be born through smart glasses?
However, the difficulty of this proposition is severely underestimated. Current smart glasses are still immature, and there are many “hard injuries” in the experience;
Although technical solutions are constantly being introduced, they often seem to be “smart for the sake of being smart”. The direction is moving too fast, neglecting the actual usage scenarios and experiential value that users truly need. I tend to believe thatsmart glasses are currently in a vacuum of technology and experience.
They deserve attention, but they should not be mythologized. True transformation may have to wait until the software and hardware ecosystems are re-integrated and the interaction logic is completely restructured.Before that, what we need is not the heat of following trends, but a bit of patience and cold thinking.
Currently, there is a significant issue with the display screen embedded in the lenses, known as “light leakage”. Essentially, the lens display screen acts as a small “projector” (optical machine), projecting light onto the lens, and then usingoptical waveguide technology to confine the light within the transparent material, guiding it to propagate and eventually reach our eyes. This leads to the development of diffractive waveguides and geometric waveguides, which are “sculpted” on the lens through nano-imprinting or photolithography, creating numerous parallel grooves that are opaque, while the smooth parts between the grooves allow light to pass through. The regular nano-structure alters the diffraction direction of light, and these grooves are known as “gratings”.
Currently, most use the former, while Ray-Ban Display uses geometric waveguides. It can be simply understood that the light from the optical machine is directed to the interior of the lens by the incident grating, where it undergoes regular and directional “total internal reflection” propagation. When it reaches the exit grating, the direction of light propagation is altered, directing it towards the eyes. Using two types of mirrors, the light is accurately guided from the optical machine to the eyeball. The density of light reflection and propagation within the lens is greatly reduced, and the light deviating from the expected path is also significantly decreased, allowing most of the light to successfully reach the reflective mirror at the center of the lens, thus reducing the likelihood of light leakage. The cost and energy consumption of pairing with LCoS optical machines are much lower.However, the supply chain is very inadequate, making it difficult to achieve large-scale production and commercial effects.
In addition, due to the micro-reflective mirror structure inside the lens, in the long run, the “non-leakage” characteristic of geometric waveguides has far-reaching value. In the next two to three years, people will feel a bit uncomfortable when they see someone wearing “smart glasses”. Others can clearly perceive that you are using a device, leading to doubts about whether your attention is still on the present. If the product’s screen leaks light, it will obstruct the user’s eye contact with the outside world,and eye contact is a very important aspect of modern communication and social interaction.
What I understand as a so-called smart wearable device should beimperceptible in its technological presence, both for oneself and for others. The logic of large manufacturers is basically to wait or act as solution providers. Wearable devices are fundamentally limited in form and performance applications, making it difficult to pair with a complete graphical interface operating system. Interaction design should be equipped with a very simplified system, with functions and interfaces resembling those of button feature phones. The closer the product is to a smartphone, the better. It should be more similar to the Apple Watch, paired with iPhone and Mac, directly utilizing the computing power and performance of the latter two. This would create a more streamlined interface and reduce resource consumption;
The core issue of wearable devices is the balance between energy consumption, heat generation, and performance. We should not aim for a more independent form, but rather for a supportive and complementary product.
02
Robotic Vacuum Cleaners
This sector is highly competitive. I am not from a technical background, but I have simply analyzed the general landscape from a market perspective—the structure of this industry is far more complex than imagined. Lawn mowers and robotic vacuum cleaners are actually typical highly monopolized industries.
The leading companies in the sector have high profitability and strong concentration. In terms of overall overseas paths, although the US market is mainstream, the real bulk of shipments is actually in Europe. In Europe, it accounts for about one-third of the global market, with six out of the top ten countries and regions in global penetration rates being European. However, Europe is very fragmented. This is also related to specific regions and factors, leading to extreme market fragmentation.
There are significant differences in consumption structure, channel systems, and cultural perceptions among different countries, which also leads to varying strategies for brands when entering the market.If we look at it by country, Germany has a relatively high penetration rate and maintains a stable growth trend. Italy and Russia are experiencing rapid growth, and the future growth prospects are quite good.
Many people may wonder why I, not from this industry, can grasp this data.
To explain simply: my data judgment usually starts from the activation volume of various apps and terminal shipment signals, making preliminary selections and judgments.
I have previously experienced both manufacturing and mobile internet systems, and I am essentially better at combining “data screening + industry judgment” analysis.
This is also the viewpoint I have always emphasized: companies must pay attention to all-channel talent in overseas talent selection—those who understand the market, can read data, and can understand the logical differences of different channels.
The main brands of robotic vacuum cleaners are primarily in Europe, where the penetration rate has exceeded 25%. In global terms, it is only about 10%. This means that Europe remains the core growth engine of the entire industry. The logic of growth here is based on two points:
1. Stock improvement—emerging Chinese brands are continuously squeezing the market share of established manufacturers;
2. Incremental expansion—localization trends of brands in Western and Northern Europe are gradually maturing.
The overall competitive strategy mainly revolves around defending market share and broadening price ranges.
The overall market is basically adopting corresponding pricing strategies to maintain market share. On one hand, the overall price range is broadened to cover a wider customer base with differentiated positioning. For example, during Black Friday, companies like Roborock and Ecovacs adopt low-price strategies to secure market share aggressively. However, overall, there is still a significant push for high-end layouts. Currently,Roborock is transitioning its products from a price of $1000 to $300 to expand its reach.
I understand this as Roborock ensuring its penetration rate through low prices to achieve growth, first securing the market before pursuing profitability.
People have previously asked me how to compete in the highly competitive robotic vacuum cleaner market in Europe. In fact, I believe this is a question of all-channel thinking. I cannot draw a conclusion, but my understanding is to step out of specific product forms, as everyone is competing primarily to push the price of this product series. Currently, different companies have slightly different strategies; it is not purely relying on pricing strategies. There are always differentiation points and opportunities, such as Ecovacs, which I do not fully understand, but I genuinely believe in its overall strategy and high-end layout, primarily focusing on price range descent and market penetration.The core conclusion is that from the overall spatial dimension, the growth space in the European market is quite broad, and the overall penetration rate is still at a relatively low level, making the prospects promising.
Although the intensity of competition among various brands is still quite high, leading brands mainly adopt strategies to broaden price ranges to seize market share. The overall industry’s profitability is facing short-term pressure. In terms of competitive strategies, due to the slowing pace of product technology iteration, the trend of average product prices is becoming more apparent. However, integrated strategic approaches are also a situation. Ecovacs’ high-end layout and Roborock’s steady layout through channel reforms, including strengthening price control to seize market share, are relatively prudent and align with market strategies. Overall, terminal products need to broaden the entire price range, and companies also need to improve the gross margin of their products.
03
Lawn Mowers
I actually do not want to discuss this topic; it has become quite uninteresting. Europe is the main battlefield, while the US is a challenging incremental zone.The industry is deep and highly competitive.
The market scale is primarily aimed at the European market. The first category of competitors includes traditional garden tool leaders like Husqvarna and Bosch, which dominate the boundary-wired product market but have lagged in the layout of boundary-less products; the second category includes domestic players like Ninebot, Kuka (Mowing Robot), Ecovacs, and Roborock, among which Ninebot is the first to launch a boundary-less product, utilizing RTK and visual solutions to create price segment differentiation and stable RTK signal sources; Kuka competes through four-wheel drive differentiation; Ecovacs has shifted from UWB solutions to RTK visual and laser radar integration solutions; these four companies accounted for nearly 90% of the market share last year; the third category includes new entrants, including companies with a robotics background and startups, but due to the high initial investment required for lawn mowers, cash flow challenges have led some startups to exit the market.
In terms of brand strategy, some companies focus on online channels to seize mindshare, such as Kuka, which primarily relies on online sales, accounting for about 90%, while Ninebot and Roborock have a more balanced online and offline presence, leveraging existing resources from two-wheeled vehicles or robotic vacuum cleaners to expand offline channels. InEurope, progress has been relatively smooth, while in the US market, due to product capability issues and low market demand compatibility, large-scale offline distribution has not yet been launched.There are not many significant opportunities for major iterations and upgrades in technology. The RTK solution is essentially a communication platform, while the laser radar solution is a highly competitive migration from home appliance companies, and the visual solution serves as a replacement for wired entry-level experience products. These solutions represent market misalignment and localization.As for lawn mowers, I no longer pay attention to various PRs as they are not very useful. The role of PR is to expose companies and secure funding; the true product capabilities are known to everyone after a round of actual testing in Europe.
The adaptability of boundary-less products: boundary-less lawn mowers solve wiring issues through visual, RTK, and radar technologies, adapting to the complex landscaping needs of the US. The differences between American and European gardens mainly include three aspects:
1. American gardens are larger and have more complex terrains (with pools, many trees, slopes/pits), requiring high obstacle-crossing capabilities, leading to the emergence of four-wheel drive (4WD) and tracked designs to solve climbing and pit problems;
2. Different grass types require chassis floating adjustments and increased motor power to enhance cutting power;
3. Differences in area lead to battery life issues, which can be addressed with batteries of different capacities.
4. American and European products have local adjustments in battery size, drive structure (Europe often uses rear-wheel drive + omnidirectional guide wheels, while the US often uses four-wheel drive/tracked designs), and chassis height, but these are based on user habits and objective factors (similar to regional adaptations in categories like cars and robotic vacuum cleaners). In the era of boundary-less products, the major technical routes in the US and Europe are not fundamentally different.
The most important point in this industry is actually the channel; traditional players’ channels can easily become fragmented. Aggressive cost pricing is expected to end around 2025, but this year there is a significant risk of inventory overflow. Manufacturers clearing inventory will only be able to re-enter the market in 2026; players need to continuously invest in R&D and participate in price wars. The future competition will hinge on each player’s strategic choices, including marketing and product combination strategies, with different companies having clear advantages and disadvantages, necessitating differentiated approaches to seize market share.
Lawn mowers actually require higher product strategy capabilities than robotic vacuum cleaners and past home appliance talents. The reason is that offline channels dominate, and the all-channel thinking is the core. Now, if someone tells me that they are looking for candidates who can operate e-commerce on Amazon’s large channels, I might not consider giving that person a core position because the market strategy and rhythm are mismatched.The lawn mower industry is entering a new era of brand awareness, where those with high brand awareness can easily scale up, and product combinations are also necessary. This way, whether in modern channels (retailers and e-commerce) or traditional garden tools and hose tools, deep binding can be achieved, as well as the brand display advantages of shared battery electric tools, garden tools, and smart lawn mowers, which can easily create scale effects. After that, companies can do well, but the efficiency of converting technology into products (especially software implementation) may grow faster.
As for multi-brand technology combination strategies, it is an ideal strategy. By combining different technical routes, brands, pricing, and target customer groups can be differentiated, but consumer choice overlaps need to be avoided.
04
Intelligent Tracking Recognition – Desktop Mobile Robot
This product was actually discussed with me by a startup team last year, and I turned them down. Why suddenly bring up this topic? Now Apple is making a comeback, and last weekend, there was an unpleasant reason when someone from a company asked me with a freeloading mentality. I told them I had heard of it before, and they immediately asked me which company was doing it. I said many startup teams in Shenzhen, and they looked at me with disdain and said, “You are talking nonsense; we are the first?”Is freeloading so justified? You are not the first, and the concept is also copied from others. I often advise others to communicate more; if you get trapped in your own cognition, then sorry, you won’t grow big.
Recently, this product has been launched by Apple. Apple is exploring new markets where it may find new sources of revenue, and personal robots are clearly one of the areas the company is researching. Bloomberg’s Mark Gurman reported that Apple is looking for “the next big thing”, and robotics technology is one of the key areas. Apple’s engineering team is dedicated to developing various home robotic devices and the AI software and features that could make home robots useful.
Apple is developinga desktop robotic device and is also looking for suppliers for desktop robotic devices. This product is similar to an iPad with a 9-inch display,mounted on a “thin mechanical arm” that allows the display to tilt, move up and down, and rotate 360 degrees. This desktop device will serve as a “home command center”, a FaceTime video conferencing machine, and a home security monitoring tool. Using a rotating base, it will reposition itself to face anyone speaking in the room, and some people within Apple have started calling it the “Pixar lamp”. It will serve as a more interactive version of the iPad or iPhone, and Apple sees it as an AI-based humanoid companion. Apple is designing a new version of Siri for this device,which will be able to remember information and engage in conversations. It will have a “visual personality”, and Apple will test an animated version of the Finder logo for Siri’s appearance. It is said that Apple aims to launch this device in 2027, hoping to price it around $1,000.
Some insiders at Apple reportedly refer to it as the “Pixar lamp”, suggesting that its design not only pursues functionality but also emphasizes expressiveness and approachability, which is a hallmark of Apple’s classic design language. This design philosophy aims to establish an emotional connection with users rather than merely serving as a cold tool. The prototype of this device is equipped with a roughly 7-inch horizontal display, similar in size to the iPad mini, which serves as the robot’s “face” and main interaction interface. Reports indicate that the current testing prototype runs a customized version of the iPad operating system, but the final product will feature a new operating system called “Charismatic”. This indicates that its development environment may be similar to existing products, but the user experience will be revolutionary. At its core is a “slim mechanical arm” or “movable limb” that gives the display the ability to move. This movement capability is not for mobility but for interaction. It allows the robot to automatically adjust its position to face the speaker, track users in the room, and even mimic head movements like nodding.This fundamentally distinguishes it from all static smart displays on the market.
This product is envisioned as a collection of roles, designed to deeply integrate into family life:
-
Smart home command center, with FaceTime calls being one of its main application scenarios, utilizing the mechanical arm to achieve physical “Center Stage” functionality, tracking and following people moving in the room during video calls, keeping them centered in the frame. Ultimately, it is designed to be a remotely controllable home security tool. Users may be able to control the robot’s screen orientation through a virtual joystick interface on their iPhone, allowing them to showcase different perspectives of the room during calls. Finally, it can actively engage in conversations and provide relevant suggestions, such as recommending nearby restaurants or related recipes when discussing dinner.
Logically, Apple is actively avoiding the home navigation challenges that plague competitors, focusing on expression and interaction.The movement of the mechanical arm is intended to make the device feel more alive, more focused, and enhance the emotional bond between users and AI assistants. Apple’s bet is that a robot that can perfectly track you during video calls and “look at” you while you speak is more valuable and poses lower risks than a robot that can follow you around the room but might get stuck on Lego blocks. It prioritizes perfecting human-computer interaction in a fixed position rather than getting bogged down in the complexities and responsibilities of autonomous mobile platforms. In fact, I discussed this product concept with a group focused on companionship last year,but at that time, the mainstream market was still AI plush toys.
I have expressed that AI companion products vary in form, either focusing on content or serving as smart home hubs.
Companion products that are wheeled are less integrated and necessary than wall-mounted iPads. The essence of high-value products is to make the robot itself the hardware, which can be seen as a high-profit “body”. Its primary strategic goal is to introduce and popularize a brand new AI, with the core being dialogue capability, contextual understanding, and proactive intelligence. Only by endowing a charming and useful physical form—design, screen “face”, and expressive movements—through this new AI can user cognition and interaction patterns be changed. This makes AI no longer just a functional tool but more like a companion, existing in the user’s most intimate space: home. This approach encourages users to engage in deeper interactions with AI, which is incomparable to an invisible voice assistant on a phone.
In fact, the key feature of the desktop mobile tracking device is its ability to scan the faces of users approaching the device through its camera and automatically adjust the layout of the interface, widgets, and display content to match the user’s personal preferences. This goes beyond simple voice recognition, achieving truly visual-based, seamless personalized interaction. Its user interface centers around clock dials and widgets, indicating that it emphasizes clear information retrieval and environmental computing. Current smart home hubs from Amazon and Google,are essentially designed for single users, extending into the home environment through “voice matching” technology, but this approach is not only imperfect but also limited to audio recognition.All notifications, calendars, and main account settings are tied to one person, creating friction in shared spaces, which is a common pain point for users. The operating system of such products and their integrated facial recognition capabilities do not just recognize a voice but recognize a person. This design creates a seamless, “zero-interface” personalized layer. When a child approaches, the device can display her homework reminders and Disney+ personal homepage; when a parent approaches, the device shows his work calendar and news widgets. For a shared device, this represents a fundamentally more intuitive and powerful interaction model. If successful, this operating system could become the standard for all future Apple home products (such as Apple TV and HomePod), creating a deeply integrated and user-friendly home ecosystem,which is difficult for competitors to replicate due to their traditional single-user architecture.
In the past, Google Nest Hub Max was a 10-inch smart display equipped with Google Assistant, a camera for video calls and security, and deeply integrated with Google Photos. It also showcased personalized screen information through facial matching technology and offered good sound quality and display effects for its approximately $229 price. However, its camera is static, and its auto-framing feature is purely digital zoom and panning, with AI being passive rather than proactive. Additionally, it lacks a physical camera cover, which is a key flaw for privacy-conscious users.
Samsung’s Ballie is a small rolling spherical robot that serves as an AI companion and can project content onto walls. Its advantage lies in its novel projection capabilities for entertainment and information display, as well as its smart home control capabilities and compact mobile design. However, it remains largely a conceptual product, and the practicality of a mobile projector compared to fixed TVs or displays is questionable.
Lenovo showcased a concept design for a dynamic AI base: a notebook stand integrated with automatic face tracking and other features. At the IFA 2025 conference, Lenovo presented several concept products, among which an interesting device is the Project Ballet dynamic AI base.
Lenovo claims its dynamic AI base is the industry’s first intelligent multi-directional notebook stand concept product, which, in addition to conventional support and height/angle adjustment functions, can utilize the notebook’s built-in camera, microphone, and speakers to achieve automatic face tracking, voice control, ergonomic posture monitoring, and other rich features.
Additionally, the Lenovo dynamic AI base supports gesture control of the stand’s rotation through an AI ring, and its base integrates multiple types of expansion interfaces.
Ultimately, the core of these products is commercialization. The rumored $1,000 price point is a significant barrier to mass market adoption. This is comparable to the initial pricing of Amazon’s Astro, which has struggled to justify its cost. The entire consumer robotics market is plagued by high costs and performance mismatches.
All consumer robots face the core challenge of achieving product-market fit—providing a value that clearly exceeds their cost and complexity. Moreover, the home environment is chaotic and unpredictable, posing a significant challenge to robotic technology. They must flawlessly execute their functions to earn user trust and justify their high prices.
Furthermore, devices equipped with cameras and microphones represent a potential significant intrusion into the most private spaces. They raise fears of continuous monitoring, eavesdropping, and data collection. These devices collect vast amounts of sensitive data (user habits, presence, conversations, facial recognition data), raising concerns about how manufacturers use and share this data. The Verkada hacking incident is a typical example of the risks associated with cloud-stored camera footage.
Heavily relying on device-side processing for AI and facial recognition (as seen in Face ID and Nest Hub Max’s facial matching features), providing clear physical indicators (like the green dot on macOS) to show camera/microphone activity, and possibly including physical blocking mechanisms, may be necessary. If a truly intelligent, reliable, and personalized “LLM” can be provided within an attractive and expressive hardware package, it could define the consumer robotics category just as the iPhone defined smartphones. It could address a fundamental pain point in smart homes, creating an exceptional user experience and deeply locking home users into the ecosystem.
Otherwise, it may become a niche gadget for affluent Apple enthusiasts, leading to disappointing user experiences that fail to deliver on promises, further damaging Siri’s reputation. Despite Apple’s best efforts, the inherent privacy issues of robotic cameras in homes may pose insurmountable barriers for mainstream audiences, thus limiting the product’s market potential.