Author: Gavin

  • In-Depth Review: Mooni M1 AI Portable Learning Companion

    In-Depth Review: Mooni M1 AI Portable Learning Companion

    The market for smart hardware for children has flourished this summer, with a constant stream of smartwatches and large-screen learning tablets; however, most of these struggle to avoid the issue of excessive screen time. Developed by Teeni.AI in collaboration with Alibaba Cloud’s Tongyi Qianwen, the Mooni M1 AI Portable Learning Companion debuts in a lightweight, wearable “AI Pin” format. It emphasizes minimal screen time, all-day AI companionship, and learning during spare moments, targeting children aged 3 to 12 (spanning both preschool and school-age groups). Moving beyond the simple playback functions of traditional storytelling devices or audio players, it features conversational capabilities, emotional intelligence, and calling functions. Is it merely a novel toy, or a practical tool for child development? This article provides a comprehensive breakdown of this distinctive AI hardware based on real-world testing across various scenarios.

    Mooni M1 AI Product Gift Box
    Mooni M1 AI Product Gift Box


    I. Design and Hardware: Rounded, Lightweight, and Built for Portability

    Actual photo of the Mooni M1
    The Mooni M1 features a cute, rounded body adorned with cartoon accessories—such as Brown Bear or a rabbit—and comes in a variety of colors, including brown, pink, yellow, and green. Weighing just 130g—significantly less than standard children’s smartwatches—it comes with an adjustable lanyard and supports multiple carrying styles: worn around the neck, attached to a backpack, or clipped onto clothing. This ensures a hassle-free experience for children whether they are out playing or commuting to and from school. Material: The casing is wrapped in skin-friendly silicone with no sharp edges, meeting safety standards for children’s products; it features IP67 dust and water resistance, ensuring peace of mind during rainy weather or accidental splashes.
    Screen: Equipped with a compact 1.2-inch color display used solely for showing the AI ​​avatar and call interface; the screen size is intentionally limited to reduce the urge for children to stare at it.
    Interaction Design: The device features distinct buttons for voice commands and power/screen locking, with differentiated sizes and tactile profiles to minimize accidental presses; a top-mounted four-microphone array enables 360° sound pickup, clearly capturing children’s voices even in noisy outdoor environments.
    Connectivity: Built-in 4G Cat.1 module with included IoT data service; it operates independently of home Wi-Fi, ensuring continuous connectivity while outdoors or on the go.
    Charging: Magnetic contact charging and a high-capacity battery provide 10–12 hours of daily use—covering continuous conversation and audio playback—sufficient for a full day out.
    Overall Design Philosophy: It downplays the “digital gadget” aspect while emphasizing the feel of a “portable companion,” making it ideal for families who prefer not to give their children smartphones or smartwatches yet still desire AI companionship and simple communication capabilities.
    II. Core Highlight: Alibaba Cloud’s Custom Large Language Model for Children—Beyond Q&A, It Understands Emotions
    As a children’s AI device co-branded with Tongyi Qianwen, the Mooni M1’s greatest strength lies in its deep optimization for children’s language patterns—setting it apart from similar products that merely scale down general-purpose large models. The device features six virtual AI agents, each with a distinct personality, allowing children to freely switch between companions:
    ✅ Teeni (Curious Companion): Encyclopedia-style Q&A and general knowledge, answering imaginative questions;
    ✅ Polly (English Companion): Immersive bilingual conversation, vocabulary practice, and spoken language interaction;
    ✅ Mia (Storytelling Companion): Collaborative storytelling, allowing the child to co-write fairy tales;
    ✅ Hug (Emotional Companion): Recognizes emotions, offering comfort during low moods and helping process negative feelings;
    ✅ Bunny (Exploration Companion): Travel-related knowledge and tips for city outings;
    ✅ Curious Cat Box (Wonder Box): Educational quizzes and brain-teasing games. Hands-on Interaction Experience
    Many traditional AI devices for children suffer from stiff dialogue, an inability to handle multi-turn conversations, and a rigid “one question, one answer” format. The Mooni M1 supports continuous dialogue with long-context awareness, accommodating the non-linear, imaginative thinking patterns of children.
    Child: “What is the largest animal in the ocean?”
    After receiving the answer, the child follows up: “Is it afraid of sharks?” The AI ​​maintains the conversational thread and answers coherently without requiring the question to be repeated.
    Furthermore, the voice synthesis moves away from robotic, electronic tones, employing a child-friendly, human-like voice that includes natural pauses and emotional inflection.
    Regarding safety, the device features a multi-layered content firewall and a built-in hardware-level physical mute switch that cuts power to the microphone with a single press. All voice data is transmitted via end-to-end encryption, and content involving violence, sensationalism, or inappropriateness is automatically filtered out, giving parents peace of mind regarding the AI’s output.
    III. Hands-on Testing of Three Core Scenarios: Integrated Study Companion, Communication, and Audio Immersion

    1. AI Study Companion for “Snackable” Learning Moments
      It does not focus on intensive drills or rote practice; instead, it is designed for learning during spare moments—such as commutes, travel, or outdoor activities. It comes with built-in synchronized textbook audio, graded English picture books, traditional Chinese culture content, and “BoBo Radio” (featuring science and general knowledge programs).
      Free from screen constraints, children can listen while walking and learn while playing. If a question arises, they can wake the AI ​​with a single button press to ask in real-time, turning outdoor exploration into an interactive science lesson.
    2. Simple Communication + Basic Location Tracking for Safety
      Many parents of young children worry that smartwatches are overly complex, leading to addiction to social features or apps. The Mooni M1 makes strategic choices regarding functionality:
      It supports two-way voice calls, with a whitelist configured via the parent app to ensure calls are restricted to family members;
      It reports real-time location, allowing parents to check the child’s whereabouts remotely via their phones;
      A note on limitations: The positioning accuracy is suitable for general reference within neighborhoods, parks, or streets, but it lacks the advanced safety features found in some smartwatches, such as floor-level positioning or water-hazard alerts.
      With no social networking apps and no ability to freely add strangers, the risk of addiction is significantly reduced, making it an ideal transitional communication device before a child graduates to a full smartwatch.
    3. Emotional Companionship: Bridging the Gap When Parents Are Away
      When parents are busy and cannot be present, children can confide their worries or share interesting stories with the AI. Unlike cold, mechanical Q&A tools, this AI companion can listen and offer empathetic guidance, helping to cultivate a child’s communication skills. Its advantages are particularly evident during outings, camping trips, and long-distance travel.
      IV. Parental Control System
      Managed via the companion Mooni mobile app:
      Remotely set usage hours and schedule device locks;
      Customize content libraries to block topics unsuitable for children;
      Manage contacts and monitor battery levels and online status;
      Remotely push audio content, such as scheduled stories or English lessons.
      The control logic is simple and intuitive—no complex settings—making it easy for grandparents to use as well.
      V. Objective Limitations (Important to Know Before Buying)
      This review aims for neutrality; please be aware of a few notable shortcomings:
      Limited positioning capabilities: Provides only basic location data without geofencing, location history, or danger alerts; safety features fall far short of children’s smartwatches in the 1,000-yuan price range;
      Magnetic charging is less convenient than universal Type-C; if you forget the proprietary cable while out, you cannot easily recharge;
      No camera or AI image recognition; lacks hardware features like scanning text or photographing homework problems;
      Continuous high-frequency conversation drains the battery quickly; connecting to a power source is recommended for long periods of nighttime listening;
      Does not support large-scale local offline storage; heavy usage relies on a 4G network.
      VI. Target Audience & Buying Summary
      ✅ Highly Recommended For
      Preschoolers and early elementary students (ages 3–8) whose parents do not yet wish to buy a smartwatch;
      Children who frequently travel, go camping, or commute to after-school care and need lightweight communication and AI companionship;
      Parents wanting to limit screen time and seeking screen-free or small-screen learning hardware;
      Families looking for English immersion (“ear training”) and general knowledge enrichment, valuing natural AI conversation.
      ❌ Not Recommended For
      Students over 10 who specifically need to photograph homework or have AI grade their incorrect answers;
      Those whose primary needs are high-precision tracking and comprehensive outdoor safety features;
      Those wanting a single device that handles classroom communication, payments, and extensive social networking functions.
      Conclusion
      Current AI hardware for children tends toward two extremes: one category piles on screens and entertainment features, easily leading to addiction; the other sticks to traditional player designs, offering stiff, limited interaction. The Mooni M1 AI portable learning companion has carved out a unique niche: it serves as a “low-screen” growth companion, featuring a lightweight, wearable design and AI-driven, emotionally engaging conversations. While it is not an all-in-one learning device—and cannot replace dedicated learning tablets or high-end smartwatches for children—it fills a gap in the market for digital products aimed at younger children by offering early-stage companionship, audio-based learning during spare moments, and simple communication capabilities.
      If you prioritize interactive companionship and reducing your child’s screen time—seeking a gentle, lightweight AI-powered early learning experience—the Mooni M1 is worth considering. However, if your primary needs are comprehensive safety tracking and homework assistance, it is best used in conjunction with a children’s smartwatch or a dedicated learning device.
  • In-Depth Review: Redmi Kids Watch Pro

    In-Depth Review: Redmi Kids Watch Pro

    With a wave of new smart hardware for children hitting the market this summer, parents’ core priorities when choosing a kids’ watch remain consistent: accurate positioning, clear communication, robust safety features, and controllable usage that prevents addiction. Recently launched, the Redmi Kids Watch Pro—Redmi’s flagship wearable for children—highlights two key features: “dual-chip offline positioning” and a “flip-style dual-camera system.” Priced at ¥1,199 at launch (and ¥888.25 after government subsidies), it competes directly in the premium (1,000-yuan range) kids’ watch market. Based on real-world testing across various scenarios, this article objectively breaks down the product’s highlights, user experience, and target audience.

    Redmi Kids Watch Pro
    Redmi Kids Watch Pro

    Flip-style side camera structure
    I. Hands-on with the Design: A Flip Structure Balancing Utility and Aesthetics
    The Redmi Kids Watch Pro features a distinctive hexagonal “mecha-style” design and comes in two colors—Vibrant Red and Energetic Purple—making it suitable for both boys and girls. The device utilizes rounded corners and edges to minimize the risk of bumps or scratches.
    It sports a 1.68-inch, 360×390 HD touchscreen. This offers a larger viewing area than most watches in the same price range (which typically feature 1.4- or 1.6-inch screens) and remains clearly visible even in bright outdoor sunlight.
    The standout feature is the flip-style camera module, equipped with 5-megapixel cameras on both the front and back. This highly flexible design allows for angle adjustments during video calls. Children can take selfies, capture scenery, or use AI object recognition without awkwardly twisting their wrists—solving the common issue of limited shooting angles found in standard fixed-camera watches.
    The strap is made of skin-friendly liquid silicone with numerous ventilation holes, ensuring comfort and preventing sweat buildup during prolonged wear. It supports 20-meter swim-grade water resistance, meaning there is no need to remove the watch for daily activities like hand-washing, getting caught in the rain, or playing in the pool.
    Hardware specs include a 1GB RAM + 8GB storage configuration, ensuring smooth system operation and app installation without significant lag.
    The overall weight is well-balanced, making it comfortable for elementary school students to wear daily. A minor downside is the rugged “mecha” aesthetic; the watch may look somewhat bulky on very young children, so it is best suited for school-aged children between 6 and 14. II. Core Highlight: All-Scenario Positioning System – Addressing Parents’ Biggest Concerns
    Positioning capability is the watch’s most competitive selling point and a key advantage distinguishing it from rivals in the same class. Many children’s watches on the market suffer from common flaws: location drift when entering underground garages, subways, or building interiors, and total loss of contact if the device goes offline or shuts down.

    1. Dual-Chip Architecture | Emergency Positioning Even When Powered Off
      The watch employs an independent dual-chip design, with the positioning module featuring its own low-power chip. Leveraging the Xiaomi “Find My” distributed device network, it supports positioning for up to five days without network assistance—even when the watch is offline or has automatically shut down due to low battery.
      Note on real-world testing: This feature requires the parent’s smartphone to run HyperOS 3.0 or later. It relies on nearby Xiaomi/Redmi devices to broadcast and receive location data. While it serves as an emergency fail-safe rather than a standard real-time tracking method, it significantly increases the chances of locating a child if they go missing.
    2. AI Floor-Level Positioning 3.0 – No More Vague Location Points
      Equipped with an upgraded AI positioning algorithm and “Smart Floor-Lock” technology, the watch can identify the approximate floor where the child is located within residential complexes, malls, or office buildings. Parents no longer see just a vague, flat map coordinate; the location data provides much greater utility when children take elevators alone or move between floors at school.
    3. Safety Alerts for Specific Scenarios
      Water Hazard Alerts: When enabled, the watch displays a local pop-up alert and sends a notification to parents if the child approaches dangerous bodies of water such as rivers, ponds, or reservoirs—an incredibly useful feature for summer outdoor activities and park visits.
      Subway Route Tracking: Continuous positioning is maintained while traveling on underground rail systems, fully recording transit stations and routes—solving the common issue where watches lose signal immediately upon entering tunnels.
      Standard features include geofencing, school-commute monitoring, one-touch SOS, and historical route playback, creating a multi-layered safety net.

    III. AI Features, Communication, and Learning Experience

    1. On-Device AI Assistant for Children
      Features a built-in voice assistant powered by a large language model (LLM) tailored for children. It supports voice Q&A, classical poetry recitation, Chinese-English translation, vocabulary lookups, and AI-powered object recognition via the camera. Unlike general-purpose LLMs, the content is filtered through safety controls designed for young users to prevent exposure to inappropriate information. Children can ask questions whenever they arise, turning the watch into a handy, on-the-go learning assistant.
    2. HD Video Calling
      Equipped with a flip-style dual 5MP camera setup, the video call clarity surpasses that of many entry-level watches with single cameras. It features a built-in pressure-relief speaker, ensuring sufficient volume for calls even in noisy outdoor environments. It supports standard voice calls and “Micro-chat” messaging; it also supports logging into the kids’ versions of WeChat and QQ, adding friends via a simple “tap-to-connect” action, and even adding friends across different brands of kids’ watches, offering greater social flexibility.
    3. Learning and Health Monitoring
      The watch integrates a dictionary, spoken language assessment tools, and synchronized audio learning resources. Built-in sensors monitor heart rate, blood oxygen levels, step counts, and sleep status around the clock, automatically generating weekly health reports sent to the parent’s phone. It includes 18 sports modes to track data from outdoor activities like running and cycling, making it easy for parents to stay informed about their child’s physical activity.

    IV. Parental Control System: Balancing Freedom and Boundaries
    Comprehensive management features are essential for families with school-age children:
    ✅ Class Mode: Scheduled deactivation of calls and entertainment functions to prevent classroom distractions;
    ✅ App Whitelist: Restricts app installations and automatically blocks games and short-video apps;
    ✅ Secure Pocket Money Payments: Supports Alipay/WeChat payments for kids with customizable daily spending limits set by parents to prevent excessive or unauthorized spending;
    ✅ Granular Permission Management: Manual control over allowed contacts, friend requests, and location sharing.
    The control system is logically designed and easy to use, allowing even parents who aren’t tech-savvy to get up to speed quickly. V. Real-world Limitations and Things to Note
    Reviews shouldn’t just focus on the positives; here are a few points to consider before purchasing:
    Offline positioning has usage conditions: It relies on the network of other Xiaomi ecosystem devices. In remote suburbs or open areas lacking Xiaomi devices, the effectiveness of emergency positioning may decrease, so do not rely solely on this feature;
    740mAh battery: Offers 1–2 days of battery life with normal daily use, but power drains faster with frequent video calls or continuous positioning; it is best to get into the habit of charging it every night;
    System entertainment apps are moderate in number, leaning more towards utility; if you are looking for a vast library of children’s games or cartoons, the content ecosystem is somewhat limited compared to certain competitors;
    The “mecha-style” casing is drop-resistant but prone to fine scratches; using a protective case is recommended.

    VI. Who Is This For? Buying Summary
    ✅ Recommended for:
    Primary school students aged 6–14 who commute to/from school or take public transport alone;
    Parents who use Xiaomi/Redmi phones and want to utilize the offline emergency positioning feature (which works even when the watch is powered off);
    Parents who prioritize outdoor safety and frequently take their children to parks, waterfronts, or shopping malls;
    Those with a budget of 800–1100 RMB who want features like a flip-up dual-camera setup and precise floor-level positioning, and who seek a well-rounded overall experience.
    ❌ Not suitable for:
    Toddlers under 3 years old; users seeking ultra-long battery life (e.g., a full week); users looking for a massive library of children’s entertainment content.

    Conclusion
    In the competitive market for children’s smartwatches priced around 1,000 RMB, the Redmi Kids Watch Pro avoids simply piling on gimmicky features; instead, it focuses squarely on safety and positioning—the aspects parents care about most. The flip-up dual-camera setup enhances daily communication, and robust parental controls make it well-suited for school environments. With a purchase price of around 888 RMB (after government subsidies), it offers highly competitive value for money.
    For modern parents, a children’s smartwatch is essentially a “safety net tool” rather than an entertainment device. If your top priority is keeping track of your child’s location at all times and having multiple layers of emergency protection, the Redmi Kids Watch Pro—launched this summer—is an AI-powered device well worth considering.

  • AGIBOT, valued at $20 billion, is rushing towards an IPO.

    AGIBOT, valued at $20 billion, is rushing towards an IPO.

    AGIBOT is reportedly aiming for an IPO with a valuation of $20 billion, having brought in CITIC Securities as its sponsoring broker. AGIBOT projects revenue of 4 billion RMB in 2026.

    AGIBOT
    AGIBOT

    According to Unitree’s projected revenue of 1.052 billion-1.128 billion RMB for the first half of 2026 and over 2 billion RMB for the full year, if AGIBOT achieves 4 billion RMB in revenue, it will significantly outpace Unitree in sales, becoming the undisputed leader in the embodied technology market.

    A key reason for AGIBOT’s lead over Unitree is its shipments to industrial applications. Beyond traditional performance and research scenarios, AGIBOT is beginning to ship to industrial applications. The industrial market is larger than performance and research, and is a key target for embodied technology companies.

    Industrial and commercial applications are arguably the two markets with the greatest future shipment potential.

    The IPO of AGIBOT will have a significant impact on the current hot fundraising efforts of embodied intelligence companies. Firstly, this determines the valuation logic for embody startups in the secondary market.

    Many embody startups are preparing for a Hong Kong IPO using the 18C route. For these startups, which are still in the loss-making stage and haven’t yet achieved commercialization, the question arises: how will the secondary market value embody intelligence? If Zhiyuan, with the highest revenue and most comprehensive business layout, successfully IPOs, then the secondary market will have a valuation anchor.

    Zhiyuan’s 4 billion RMB revenue corresponds to a valuation of 20 billion USD, or over 150 billion RMB, based on a price-to-sales ratio of approximately 40. The price-to-sales ratio is the ratio of market capitalization to revenue, a crucial valuation metric in the secondary market, applicable to unprofitable or high-growth companies in the internet and high-tech sectors.

    If the leading company, Zhiyuan, has a price-to-sales ratio of 40, it will become the anchor for the valuation logic of other embody companies. In other words, if other embody companies IPO, the maximum valuation the secondary market will give them is 40 times their revenue.

    This puts considerable pressure on many leading embody companies. Companies like Galaxy, Xinghaitu, Qianxun, and Ziyangv, with valuations exceeding 20 billion yuan, need revenues of over 500 million yuan to justify such valuations.

    Secondly, this determines whether the booming primary market financing can be sustained.

    Many spun asset companies this year have achieved valuations exceeding 10 or 20 billion yuan by playing the pre-IPO card, with investors rushing in with IPO in mind. Many spun asset companies are overly optimistic about IPOs, failing to grasp the hard logic between revenue and valuation. Moreover, with the exception of a few companies, most have revenues below 100 million yuan.

    Therefore, for spun asset companies with low revenues, the IPO race is destined to be difficult. If the IPO fails, or even if it succeeds, the valuations in the primary and secondary markets are inverted, resulting in losses for investment institutions. This would mark a turning point for the booming spun asset financing market.

    In fact, a similar story played out three years ago in the autonomous driving sector, with 2023 being the turning point for autonomous driving financing. In the two years prior to 2023, funding for autonomous driving was as hot as it is now, best described as “flourishing” and “overflowing with enthusiasm.”

    First, startups developing L4 autonomous driving systems, passenger vehicle intelligent driving systems, sensors, LiDAR, domain control middleware, and basic operating systems could easily raise funds. Many investment institutions invested in every sub-segment of the industry chain. Second, much like with Jisheng Intelligence today, leading companies were raising three or four rounds of funding a year, each round larger than the last, with valuations soaring.

    This euphoric period ended in 2023, as many autonomous driving companies went public. For investment institutions, an IPO is the moment the mystery of profit or loss is revealed.

    The 2023 autonomous driving IPOs saw a widespread phenomenon of valuation inversion between the primary and secondary markets. Some leading companies even saw their post-IPO market capitalization fall to half of their pre-IPO valuation, meaning that investors who invested directly lost half their initial investment.

    The valuation inversion between primary and secondary markets has made even institutions with weak industry research capabilities aware of the risks, leading to widespread caution. Starting in 2023, investment institutions shifted from a “broad-based” approach to a “focused” strategy, investing only in companies with solid and stable mass production and delivery records. Therefore, the number of customers, the number of vehicle projects completed, and the revenue generated became the most important performance indicators.

    After 2023, companies in the autonomous driving sector that have consistently received funding have almost exclusively been those with strong mass production and delivery records.

    Therefore, whether the booming autonomous driving financing market can continue will become clear as these companies rush to IPO.

  • Lu Siyuan, head of XPeng ‘s Infra division, joins OpenAI.

    Lu Siyuan, head of XPeng ‘s Infra division, joins OpenAI.

    Sources familiar with the matter revealed that Lu Siyuan, head of AI Infra at XPeng Motors, will be leaving the company and may join OpenAI’s Infra team to participate in embodied intelligence research. At XPeng Motors, Lu Siyuan oversaw the group’s unified AI infrastructure, directly managing a technical team of approximately 200 people, directly supporting the shared underlying computing power and software stack for the three embodied intelligence businesses: autonomous driving, IRON humanoid robots, and flying cars.

    GPT5.6
    GPT5.6

    XPeng was one of the first automakers in China to unify humanoid robots, L4 autonomous driving Robotaxi, and flying cars under its embodied intelligence (physical AI) strategy. He Xiaopeng believes that automobiles, humanoid robots, and flying cars will share a unified AI technology foundation, with 60%-70% of perception, large-scale models, computing power, and motion control technologies reusable across domains.

    Meanwhile, XPeng Motors has changed its name to XPeng Group, with a long-term strategic goal of transforming from an intelligent electric vehicle company into a global embodied intelligence company focused on the physical world.

    In 2023, Pengxing Intelligent was incorporated into XPeng Motors, with He Xiaopeng directly leading the robotics business. 2026 was defined by XPeng as the first year of mass production for robots, clearly defining a roadmap for full-chain mass production and scenario-based deployment, officially completing the leap from R&D to commercialization.

    The head of XPeng Group’s robotics business is Zhao Tongyang, the founder of Duogou Robotics, and an expert in robot hardware and legged motion control. On June 10, 2026, He Xiaopeng released an internal letter announcing that he would personally assume the roles of head of the Robotics Center and head of the Product Department, elevating the robotics business to the highest strategic priority of the group. Gu Jie is the head of the Embodied Systems Engineering Department, and Liu Xianming is the head of the General Foundation Department.

    Shi Xiaoxin, the number one embodied product manager at XPeng Group, left the company in June 2026.

    Lu Siyuan holds a Ph.D. in Physics from the University of Southern California (USC), possessing a strong foundation in high-performance computing and deep learning theory, specializing in large-scale data-intensive computing and model optimization.