AICrunchX

Category: AI For Pets

AI pet hardware is a product designed specifically for pet-owning families. It uses artificial intelligence technology at its core and integrates functions such as intelligent sensing, automatic interaction, remote monitoring, and health management.

  • In-Depth Review: MOMOTOY Dundun Beast AI

    In-Depth Review: MOMOTOY Dundun Beast AI

    In an era of rapid iteration for consumer smart hardware, traditional designer toys have long since transcended their status as mere “static ornaments,” evolving toward intelligence, emotional connection, and contextual versatility. While many previous smart toys offered only superficial interactions—such as basic voice dialogue—and suffered from mechanical stiffness, laggy responses, and a lack of personality, they struggled to truly integrate into users’ lives. MOMOTOY’s newly upgraded Dundun Beast AI (2nd Gen) breaks through industry technical barriers by combining a proprietary high-end hardware architecture, a novel four-layer AI system, and full-scenario adaptability, bringing cutting-edge embodied AI technology to a palm-sized designer toy. This review evaluates the device across five key dimensions—design, hardware specifications, AI interaction capabilities, practical utility, and core differentiating advantages—to provide a comprehensive look at the real-world experience of this phenomenal smart AI device and the new form of intelligent designer toys.

    MOMOTOY dundun
    MOMOTOY dundun

    I. Appearance and Texture: A Cute, Trendy Aesthetic Combining Portability with Refinement

    As a smart toy designed for companionship, the MOMOTOY Dundun Beast AI (2nd Gen) retains the brand’s signature “soft and cute” aesthetic, perfectly balancing visual appeal, tactile comfort, and practicality. The device features a skin-friendly, high-density plush exterior that is soft to the touch, shed-free, and odorless, making it comfortable for both casual handling and close companionship; unlike the cold feel of standard plastic smart devices, it offers a soothing, comforting presence.

    Its lightweight, palm-sized design fits naturally in the hand, ensuring effortless portability. It also features a magnetic attachment system, allowing it to adapt flexibly to various setups—such as sitting on a desk, hanging from a backpack, or resting in a car—enabling all-day companionship whether commuting, playing outdoors, or relaxing at home. In terms of details, the product boasts an upgraded dynamic LED eye interaction system. It supports six customizable, expressive eye styles and over a hundred dynamic emotional expressions, moving beyond the simple, static light feedback of traditional smart dolls; every interaction is met with a unique, lifelike expression that creates a truly immersive atmosphere. The overall design balances the collectible aesthetic of designer toys with the practicality of smart hardware; it serves equally well as an exquisite desktop ornament, a charming accessory to complement one’s style, or a dedicated smart companion. Its appealing look suits users of all ages, making it perfect for personal stress relief or as a thoughtful gift.

    II. Hardware Core: Comprehensive In-House Upgrades Breaking Industry Molds
    Compared to the basic configuration of the first-generation product, the MOMOTOY Dundun Beast AI (2nd Gen) features comprehensive in-house hardware upgrades. Moving away from generic industry molds, it incorporates the market’s smallest high-performance computing core. This leap in hardware performance effectively resolves common pain points found in traditional smart designer toys, such as lag, slow response times, and poor compatibility.

    In terms of core performance, the device boasts an interaction response time as low as 650 milliseconds. Voice commands, emotional feedback, and conversational replies occur with virtually no latency, offering interaction fluidity that far surpasses similar products and eliminating the sluggish user experience often found elsewhere. A key highlight of this upgrade is the connectivity module: an industry-first dual-network, dual-mode design (WiFi + 4G) that breaks free from local network limitations. It supports whole-home WiFi integration while also functioning independently without a phone or local network, making it perfect for scenarios without WiFi, such as outdoor activities, commuting, or business trips. Furthermore, it is fully compatible with major smart ecosystems—including Apple, Huawei, and Xiaomi—allowing for seamless integration with IoT devices to enable voice control and scenario-based automation, effectively transforming the toy into a compact smart home hub.

    Regarding safety and battery life, the product employs a dual-layer security system (combining on-device and cloud-based controls) alongside a three-tier emotional memory model. User conversation data and personalized memory data are encrypted throughout storage, ensuring privacy and security. Battery efficiency is optimized through low-power computing algorithms; the device offers extended standby times during light daily interaction and maintains stable performance during frequent conversations or smart automation tasks, meeting the need for all-day companionship.

    III. AI Smart Experience: Four-Layer Agent Architecture for Warm, Two-Way Interaction
    If hardware upgrades form the foundation, then the evolution of the AI ​​system represents the core competitive advantage of the MOMOTOY Dundun Beast AI (2nd Gen). This product marks the first time the four-layer AI Agent architecture—previously found only in flagship humanoid robots—has been fully integrated into palm-sized hardware. It encompasses four core modules: personality shaping, long-term memory, an independent worldview, and scenario-wide skills. By breaking free from the mechanical, scripted responses typical of traditional smart toys, it delivers a smart companionship experience characterized by genuine personality, memory, and emotional intelligence.

    Regarding emotional interaction, the device employs high-precision emotion recognition algorithms across over 50 dimensions. It accurately captures tone and emotional fluctuations in the user’s voice, identifying states such as happiness, sadness, anxiety, and fatigue. When a user feels down or speaks negatively, “Dundun Shou” proactively offers comfort and positive encouragement, using a gentle, soothing, and endearing tone. During daily interactions, it adapts its vocal style—switching between lively, tender, and playful modes—based on the conversational atmosphere. This bidirectional, multimodal interaction transforms communication from a one-way stream of commands into a genuine, soulful dialogue. Additionally, the product supports voice cloning for a custom voice profile; combined with personalized AI agent settings, users can create a unique companion character and forge a special emotional bond.

    In terms of conversation and capabilities, the device supports natural dialogue in over 50 international languages. Interactions are fluid and conversational—free from robotic stiffness—enabling everything from casual chats and Q&A sessions to lighthearted banter, educational explanations, and storytelling. Powered by over 20 proprietary AI agents, the product unlocks a diverse range of practical skills. Beyond basic human-machine dialogue, it offers customized services such as routine reminders, study companionship, emotional support, and interactive play. Over time, it continuously evolves its personality and response style based on the user’s conversational habits, interests, and emotional states. It grows to understand the user better with every interaction, setting it apart from generic, one-size-fits-all smart products.

    IV. Real-World Scenario Testing: Covering Home, Travel, and Study for Maximum Utility
    To authentically demonstrate the product’s value, we conducted comprehensive real-world tests across high-frequency daily scenarios, validating the practical utility of the MOMOTOY Dundun Shou AI (2nd Gen).

    1. Smart Home Companionship Scenario
      At home, the device stays constantly online thanks to dual-network, dual-mode connectivity, allowing for independent interaction without the need to pair with a smartphone. In leisure moments, it offers engaging conversation, plays soothing music, and reads bedtime stories. When the user is alone, it provides real-time interaction to alleviate boredom, senses the user’s emotions, and offers proactive companionship. It also supports whole-home NFC sensing and IoT voice integration, allowing users to control lights, curtains, and appliances via voice commands; acting as a mini smart hub, it makes home life more convenient and emotionally warm.
    2. Learning and Personal Growth Companion
      Designed with students in mind, the product balances fun with utility. It handles daily tasks such as answering academic questions, sharing general knowledge, correcting pronunciation, and creating study plans. Its gentle, patient interaction style—distinct from cold, mechanical study tools—effectively alleviates study-related anxiety. A long-term memory module tracks the user’s study habits and areas of weakness, allowing the device to tailor its Q&A and companionship modes accordingly; it serves as both a fun playmate and a thoughtful study assistant.
    3. Outdoor Travel and Commuting
      With independent 4G connectivity, the device enables interaction anytime—during commutes, outdoor excursions, or business trips—without needing a smartphone or Wi-Fi connection. Users can relieve stress, listen to jokes, or hear news updates via voice commands when tired from traveling. In unfamiliar environments, its continuous smart companionship effectively eases feelings of loneliness. Its lightweight, magnetic, and portable design allows it to be easily attached to backpacks, car interiors, or luggage, providing all-day companionship that transcends specific settings.

    V. Review Summary: Redefining Consumer-Grade Embodied AI “Trendy Toys”
    The comprehensive testing reveals that the MOMOTOY Dundun Beast AI (2nd Gen) is not merely a superficial combination of a “trendy toy” and voice AI, but a full-scale implementation of consumer-grade embodied AI. Leveraging four core strengths—proprietary high-end hardware, a four-layer AI architecture, full-scenario adaptability, and immersive emotional interaction—it effectively resolves common industry pain points found in traditional smart toys, such as slow response times, lack of memory or emotional awareness, limited usage scenarios, and poor compatibility.
    Compared to the highly homogenized smart toys currently on the market, this product’s standout feature is its “vitality”: it possesses an independent personality and long-term memory, demonstrates emotional empathy, and is capable of continuous evolution. It successfully combines the aesthetic appeal, collectibility, and comforting nature of trendy toys with the computing power, practicality, and stability of professional smart hardware. It is perfectly suited for a wide range of scenarios—whether it’s young people looking for daily stress relief, companionship while living alone, or entertainment during commutes, or children needing a companion for learning and growth or parent-child interaction.
    If the first-generation product was merely a foray into the world of smart designer toys, the MOMOTOY Dundun Beast AI (2nd Gen) stands as a benchmark that brings high-end embodied AI technology into everyday life. It redefines the potential of smart designer toys, transforming technology from something cold and impersonal into a source of warm, heartfelt companionship.

    Review Highlights & Areas for Improvement
    ✅ Key Highlights

    • Proprietary computing core delivering ultra-low latency interaction (650ms) for a seamless experience
    • Dual-mode connectivity (WiFi + 4G) allows for standalone use without a smartphone, ensuring unrestricted operation across all scenarios
    • Four-layer AI Agent architecture combined with 50+ dimensions of emotion recognition creates interactions that are warm, memorable, and personalized
    • Full ecosystem compatibility and whole-home IoT integration offer utility far beyond that of ordinary designer toys
    • Lightweight, magnetic, and portable design that balances aesthetic appeal with practicality
      ⚠️ Minor Areas for Improvement
    • 4G connectivity requires a dedicated data plan; outdoor use necessitates prior activation
    • Personalizing the AI ​​personality requires a brief initial adjustment period, with the experience improving significantly through long-term use
  • Jollybaby Baby Elf AI Early Education Plush Toy Actual Testing Review

    Jollybaby Baby Elf AI Early Education Plush Toy Actual Testing Review

    The current children’s AI hardware track is clearly divided: one type is a hard plastic early childhood education robot with a single form of interaction, and there is a risk of collision when used by young babies; the other type is an ordinary plush doll that only has an audio playback function and lacks two-way intelligent interaction. Jollybaby, which has been deeply involved in the sensory enlightenment of infants and children for many years, officially launched the Baby Elf AI early education plush toy on July 24, targeting infants and young children aged 0-6 years old. It integrates plush comfort dolls, multi-touch tactile sensing, children’s large model voice interaction, and age-specific early education content. It breaks away from the “pure voice dialogue” of traditional AI toys and focuses on audio-visual, touch and full-sensory interaction. This actual test comprehensively dismantled this new product from the aspects of appearance safety, interactive experience, AI capabilities, early education content, parent control, and product shortcomings.

    Jollybaby Baby
    Jollybaby Baby
    1. Appearance and hardware workmanship: maternal and infant-grade plush, taking into account soothing properties and technology interaction
      The Baby Elf is designed based on the original IP “Rainbow Butterfly Valley”. There are 5 different color-matched Elf images, corresponding to 5 independent sets of characters. The whole machine has a rounded spherical cute pet shape, with transparent movable elf wings. It is equipped with a high-definition LED dynamic expression screen on the front, which can switch eyes and expressions in real time.
      Material safety: The outer layer is made of Class A baby plush fabric, which has no fluorescent agent and no odor, and is suitable for young babies to cuddle, hug and comfort. The core electronic module has a built-in removable inner bag, and the plush outer skin can be removed and cleaned separately, solving the pain point of smart plush toys that are difficult to clean.
      Touch-sensitive layout: There are built-in touch sensors in many places on the head and abdomen, which is different from competing products on the market that only rely on button control. Touching the head gently triggers a soothing tone, rubbing the abdomen triggers laughter, achieving natural interaction without buttons, and adapting to the grasping, touching and exploring habits of infants and young children.
      Basic hardware specifications
      Type-C charging port, full battery life: about 5 hours of pure audio playback; about 3 hours of continuous conversation + light interaction mode; the built-in speaker has been acoustically adjusted for children, limiting the maximum volume and protecting hearing; the base is stable and will not fall over easily when held or placed flat.
      Advantages: The plush shape naturally has the properties of a comfort doll, and it is both an early education machine and a sleep companion doll; the fully rounded corner design can also be used by babies as young as 0 years old; the outer shell is removable and washable.
      Disadvantages: The LED screen is concentrated on the front, and it is easy to accumulate dust on the screen if it is exposed to strong light for a long time; the plush hair has slight floating hair after long-term use, and needs to be combed regularly.
    2. Core interactive experience: tactile + voice dual interaction to create anthropomorphic emotional companionship
      The vast majority of AI early education toys only support voice commands, and the biggest difference of Baby Elf is touch sensing + AI dialogue linkage, creating a multi-sensory closed-loop interaction.
      Touch native interaction
      When the baby touches different areas of the doll, the elf will give feedback with expression changes, wing flapping, and lighting effects. When the baby gives the voice command “Dance,” the wings will swing synchronously, and the screen light will follow the rhythm, achieving “voice command → action lighting feedback” linkage. In bedtime mode, storytelling simultaneously turns on slow gradient soft light and slight vibration to create a sleep-inducing atmosphere.
      Five independent IP character switching
      There are 5 sets of elf personalities built-in: Qimeng, Xingyue, Yunbao, Aozai, and Yangyang. Each set of characters has its own tone, personality, and story library. Parents can switch roles with one click through the supporting applet, making it less likely to cause aesthetic fatigue after long-term use. The main line of the character story revolves around emotional guidance, teaching children to accept joy, anger, sorrow, and joy, while also taking into account psychological enlightenment.
      AI voiceprint cloning (core highlights)
      The device supports recording parents’ voice samples, and AI can reproduce parents’ voices to read stories and daily conversations. For parents who often work overtime and travel short distances, their children can hear familiar voices at any time to fill the gap in parent-child companionship. This is also an exclusive feature that distinguishes plush AI toys at the same price range.
      Actual test experience: The interaction logic is softer and does not feel cold like a robot. When young babies are not proficient in language expression, they can simply rely on touch to play; children over 3 years old can complete multiple rounds of conversations, and the product can cover the complete growth cycle of 0-6 years old.
    3. AI voice interaction and early education content: Age-specific enlightenment, adapting to the cognitive rhythm of infants and young children
      In view of the characteristics of children’s milky and intermittent speech, the official speech recognition model has been optimized. Compared with ordinary smart speakers, it is easier to recognize children’s non-standard pronunciation.
      ✅AI conversation ability
      Supports multiple consecutive rounds of natural dialogue, interesting questions and answers, and emotional counseling; when children express negative emotions, the elves will provide positive guidance instead of blunt avoidance. The complete large model dialogue capability is opened when connected to the Internet; the offline mode has built-in local story and children’s song resources, and the basic playback function is retained.
      ✅ Age-specific stepped content library
      Locally pre-installed 6 major sections of content: classic children’s songs, English nursery rhymes, enlightenment of Chinese studies, bedtime picture book stories, nature science popularization, good habit development, a total of 600 + genuine audio resources. Content is stratified by age:
      0–2 years old: white noise, lullabies, simple cognitive sound effects, short nursery rhymes;
      3–6 years old: bilingual enlightenment, popular science stories, Chinese studies, and social and emotional guidance.
      After WiFi is connected to the Internet, the audio library will be continuously updated OTA, so parents do not need to download resources manually.
      Objective shortcomings: the offline mode cannot use advanced AI dialogue, voiceprint cloning, and role switching functions, and only retains audio playback; the accuracy of complex long sentence question and answer is lagging behind high-end desktop robots, and it is more biased towards the enlightenment of young children, and is not suitable for subject question answering for school-age children.
    4. Mini program parent control system to solve parenting pain points
      The supporting mobile phone applet is not just a remote control, but also builds a complete parent management system:
      Usage time control: You can set the maximum daily play time, automatically remind you to take a break when the time limit is reached, and automatically sleep to protect your eyes if no operation is performed;
      Triple content filtering mechanism automatically blocks inappropriate topics;
      Children’s growth log: automatically records interaction frequency, content of interest, and generates a simple growth report;
      Remote volume adjustment, scheduled sleep alarm clock, and online interaction with multiple elves;
      Privacy control: local voice data isolation, supports one-click clearing of conversation records, and complies with children’s smart hardware privacy specifications.
    5. Applicable scenarios and crowd division
      ✅ Recommended people to buy
      The family has infants and young children aged 0-6 years old and wants a product that has both comfort doll and early education functions;
      We need tools to soothe people to sleep, and pay attention to parent-child companionship and emotional enlightenment before going to bed;
      I often cannot accompany my children all the time and want to use the voiceprint function to continue the voice companionship of my parents and children;
      Look for low-risk toys to avoid collision risks with hard robots;
      Suitable as children’s birthday and full moon gifts.
      ❌ Not suitable for the crowd
      The main needs are primary school cultural class tutoring and synchronous textbook learning;
      Want the robot to walk autonomously and perform complex body movements;
      You will be in a network-free environment for a long time and need full AI conversation function.
    6. Summary of comprehensive advantages and disadvantages
      Product advantages
      Innovative form: maternal and infant grade plush doll + AI large model two-in-one, taking into account comfort properties and intelligent early education, covering babies as young as one month old;
      Unique multi-touch interaction does not rely on voice, so even young babies can play independently;
      Exclusive voiceprint cloning + multiple IP characters, outstanding emotional companionship value;
      The content is designed in layers, following the growth rules of infants and young children, and does not blindly pile up complicated resources;
      Removable plush fabric for easy cleaning; complete parental control and child content safety protection.
      Existing shortcomings
      Advanced AI functions and voiceprint cloning rely on WiFi networks, and the experience is greatly reduced after the network is disconnected;
      There is no ability to move independently, and interactive actions are limited to wing swings and light expressions;
      Plush fabrics require regular cleaning and are not suitable for families with extreme mysophobia;
      The subject-based learning resources are weak, and the positioning is biased towards enlightenment companionship rather than learning teaching aids.
    7. Final purchase evaluation conclusion
      In a market where AI early childhood education hardware is generally dominated by plastic robots, the Jollybaby Baby elf has opened up the intelligent track for plush comfort dolls. It does not pursue complex mechanical movements, but captures the core needs of children aged 0–6: security, sensory exploration, emotional companionship, and lightweight enlightenment.
      If you are looking for a children’s playmate that can be hugged to sleep, interactive, gentle and non-stimulating, and takes into account early childhood education, and prioritizes companionship rather than high-intensity learning functions, this AI plush toy is highly competitive; if the primary goal is subject tutoring and programming learning, it is recommended to choose a professional educational robot.
  • Meteer AI dancing robot actual test evaluation

    Meteer AI dancing robot actual test evaluation

    In the highly homogeneous children’s AI early education hardware track, the vast majority of robots are limited to voice question and answer and story playback, lacking dynamic interactions that can continue to attract young children. The Meteer AI dancing robot, launched by Yilong Ailuoxun (Quectel Ailuoxun) and Honor Ecosystem, was officially unveiled on July 29. The product is aimed at the 3-8-year-old children’s market, integrating AI natural dialogue, programmable dance rhythm, and Bluetooth audio into one, trying to solve the pain points of traditional early education toys that are “short of freshness and monotonous interaction.” This actual test will comprehensively analyze the appearance and workmanship, dance performance, AI interaction, early education capabilities, ecological experience and product shortcomings.

    Meteer AI
    Meteer AI
    1. Appearance design and hardware workmanship: trendy and playful shape, suitable for children’s aesthetics
      Meteer adopts a cartoon humanoid design, with an orange peaked cap and headphones appearance, which is different from the square early childhood education machines on the market. The trendy style can be placed on the desktop and carried out. The height of the whole machine is about 26cm, the net weight is 600g, it is made of matte ABS safety plastic, and the corners are all rounded. It has passed the safety standards for children’s toys and is not easy for young children to bump when playing with it.
      The front is equipped with dual dynamic LED expression screens, which act as the robot’s “eyes”. Expressions will automatically switch when dancing, talking, and sleeping, and the images of happiness, confusion, sleepiness, etc. will change in real time, greatly improving the sense of personification. The base is widened at the bottom to effectively avoid tipping during large dance movements.
      The interface is equipped with a Type-C universal charging port. The charging time is about 1.5 hours. The battery life in continuous dance mode is close to 2 hours when fully charged. The battery life can be extended to 3.5 hours by simply playing audio through voice conversations.
      Advantages: Highly recognizable shape; moderate weight, children can move it independently; universal charging cable, no special charger required.
      Disadvantages: The outer shell is easily stained by fingerprints, and the light-colored clothing area is easily stained, making it inconvenient to wash and clean.
    2. Core highlights: 10+ degrees of freedom rhythm system, truly realizing music following dance
      Dancing ability is Meteer’s core differentiating selling point. The hardware is equipped with more than ten movable joints, covering the head, arms, and waist, and has a built-in library of hundreds of premade dance moves, including rhythmic swings, gesture dances, and rhythmic click dance steps.
      Two major gameplay methods tested:
      Voice-triggered dance: activate the wake-up word “Hi Meteer” and directly give the command “Dance a dance” or “Play children’s songs and dance”, and the robot will automatically match the song and start the rhythm;
      Automatic audio recognition of rhythm: Play built-in music via WiFi network, or connect to mobile phone audio source via Bluetooth. The robot can recognize the beat and adjust its movements in real time according to the rhythm of the music, instead of playing fixed movements in a loop.
      Advanced features of multi-machine group dance mode: Multiple Meteers are connected to the same mini program group to achieve synchronized dancing, suitable for birthday parties and kindergarten activity scenes.
      Objective experience: The joints operate smoothly, the noise control is excellent, and there is no obvious gear noise. Limited by cost and size, the product is designed with a fixed base and cannot be moved and walked; dance mainly involves movements of the upper limbs and waist, and there is no walking or jumping movements of the lower limbs.
    3. AI voice interaction actual test: optimized for children, taking into account daily companionship and early education
      Relying on Honor AI ecological collaboration, Meteer supports multiple rounds of continuous natural dialogue, getting rid of the keyword-triggered mechanical interaction of low-end toys.
      ✅Basic ability actual test
      Daily Q&A: weather, time, popular science knowledge;
      Early education content: Chinese and English picture books and stories, children’s songs on demand, simple English spoken dialogue, word enlightenment;
      Creative interaction: brain teasers, fun questions and answers for children.
      The voice wake-up recognition performance is stable within a measured distance of 3 meters, and it can still respond accurately in a noisy living room environment; the recognition accuracy drops slightly when long distances and multiple people are talking at the same time.
      The supporting applet supports custom AI character settings. Parents can set the robot’s personality and speaking style to create an exclusive companion. All chat records are saved in the cloud, and it also provides children’s content filtering to block inappropriate content.
      The network solution supports WiFi networking to use complete large model capabilities; when the network is disconnected, only local children’s songs and pre-made dances can be run, and advanced dialogue functions are limited. This is also a common feature of most current consumer-grade AI toys.
    4. Expanded functions: Bluetooth speaker + mini program customization, sustainable gameplay
      Meteer has a built-in stereo speaker. In addition to the built-in early education music library, Meteer can be used as a Bluetooth speaker to connect to a mobile phone to play music. It can be used as both a toy and a speaker in one device.
      Mobile applets are function control centers:
      Manually select dance scripts and custom choreograph action sequences;
      Light expression customization, volume adjustment, and scheduled sleep;
      OTA online upgrades, continuously updating dance moves, early education audio libraries, and dialogue models.
      A very useful child management function for parents: you can set the daily usage time to prevent children from playing with it for a long time, and remotely sleep the device with one click.
    5. Suitable for crowds and scene positioning
      Target audience: Children aged 3–8 years old in preschool and lower grades of primary school
      ✅ Recommended purchase groups:
      Families who want to combine early education + sports interaction, but whose children tend to get bored listening to stories statically;
      Need atmospheric toys for parties and birthday scenes;
      Honor ecological users pursue device interconnection experience;
      Looking for AI trendy hardware that has the attributes of a desktop ornament.
      ❌ Not suitable for:
      The robot needs to walk autonomously and follow the movement; the pursuit of in-depth subject tutoring (not suitable for post-school cultural courses).
    6. Summary of actual measurement: summary of advantages and shortcomings
      Product advantages
      Differentiated rhythm interaction: a complete dynamic dance system that is rare in similar price ranges, effectively extending the life cycle of toys and relieving children’s heat for three minutes;
      Anthropomorphic LED expression system, stronger emotional feedback during dialogue + dance, and the emotional companionship experience is better than ordinary early education machines;
      The software and hardware ecosystem is complete, supports OTA continuous updates, and functions are not capped out of the factory;
      Trendy appearance, suitable for home desktop display; safe material, suitable for young children;
      Opening up the Honor AI ecosystem, the interaction logic is smooth, and the speech recognition is optimized for children’s speech.
      Existing shortcomings
      The base is fixed and cannot move independently, so there is an upper limit to the dance form;
      Heavy dance mode has a shorter battery life, so long-term party use requires extra charging;
      Advanced AI dialogue relies on wireless networks, and offline functions are greatly reduced;
      The content of early childhood education is biased towards enlightenment and entertainment, and lacks systematic graded courses, making it difficult to meet the needs of in-depth learning.
    7. Final purchase advice
      Meteer AI dancing robot redefines the interaction method of children’s AI toys, jumping out of the involution track of “only talking and telling stories” to create an immersive companionship experience with music + dance + AI dialogue. It is not a professional learning aid, but a smart playmate that focuses on fun enlightenment and parent-child interaction.

    If parents value more fun and want their children to get rid of simple screen entertainment and like dynamic interactive gameplay, this product deserves special consideration; if the primary need is synchronized teaching materials and subject tutoring, they can give priority to professional education early childhood education robots.

  • A Must-Have for Multi-Cat Families – Petkit Visual AI Feeder

    A Must-Have for Multi-Cat Families – Petkit Visual AI Feeder

    Introduction: As a leading domestic brand of smart pet hardware, Petkit launched the 2nd Generation Visual AI Feeder (Gemini Dual-Compartment Model) in 2026. It features upgraded edge AI computing power, 140° panoramic visibility, and HarmonyOS intelligent connectivity, precisely addressing three major pain points: single-person working professionals, multi-pet families competing for food, inability to monitor feeding in real time, and food shortages while traveling. This two-week hands-on test provides a complete breakdown, examining the appearance and hardware, core AI functions, feeding practicality, advantages and disadvantages, and comparisons with competing products, offering pet owners a realistic purchasing guide.

    Petkit Visual AI Feeder
    Petkit Visual AI Feeder

    I. Appearance and Basic Hardware Parameters: The device features a simple, pure white ABS body paired with a black and white ceramic food bowl, suitable for various home décor styles, and its compact size saves space.

    1. Food Storage Structure: Dual independent compartment design, total capacity 5L (2L + 3L), one compartment for staple food and the other for freeze-dried/snacks, preventing cross-contamination of odors; the compartments are light-proof and double-sealed with silicone, and include an independent desiccant compartment, preventing cat food from clumping due to moisture during the rainy season in southern regions.
    2. Camera Module: 1080P HD camera, 140° ultra-wide-angle infrared night vision, providing complete coverage of the food bowl area; supports a privacy timer mode, allowing users to customize camera start/stop times, automatically turning off the lens to protect privacy when at home.
    3. Power Supply Safety: Primarily plugged in via adapter, with a built-in large dry cell battery backup compartment. In case of a power outage, it can maintain timed food dispensing for 7-15 days. Remote function only fails when the internet connection is lost; basic feeding remains unaffected.
    4. Cleaning Design: The food compartment, dispensing impeller, and ceramic food bowl are all removable. A food-grade silicone spiral anti-jamming structure allows for smooth dispensing of extruded food and freeze-dried food up to 12mm thick, eliminating cleaning dead spots and preventing grease buildup and bacteria that can cause black chin in pets. 5. Ecosystem Compatibility: Natively integrated with HarmonyOS, enabling data synchronization across Huawei phones, tablets, and smartwatches. Feeding records can be quickly viewed without downloading a separate app.

    II. Core AI Functionality Testing Experience (The Biggest Highlight of This Upgrade)
    Compared to the first generation, the second generation features a new computing chip, offering 7 times faster image processing. The multi-pet facial recognition algorithm has been comprehensively optimized, supporting the recording of facial information for up to 15 cats and dogs—a key advantage for multi-cat households.

    1. AI Multi-Pet Facial Recognition, Preventing Feeding Imbalances: After recording each pet’s face, the camera automatically captures and recognizes the pet whenever it approaches the food bowl. The app generates an independent feeding profile for each pet, recording feeding time and visit frequency.
    • Practical Scenario 1: A family has two cats, one overweight and one underweight. The overweight cat often steals food. The device records the underweight cat’s food intake and sends a “Cat XX hasn’t eaten today” reminder, promptly identifying picky eating or loss of appetite.
    • Practical Scenario 2: During kitten food transition, the device assesses palatability based on feeding frequency, assisting in adjusting the food dispensing ratio in the two food bowls using a 7-day transition method. Actual limitations: It can only recognize pet approach behavior; the food bowl lacks a weighing sensor, making it impossible to accurately measure the actual weight of each cat’s food intake. It can only roughly estimate food intake through video footage, limiting its effectiveness for post-operative or diabetic pets requiring precise food control.
    1. 24/7 Visual + Two-Way Voice Control: Heal Separation Anxiety
      140° wide-angle coverage of the entire feeding area; high-definition color image during the day; clear infrared night vision at night; allows for virtual pet viewing anytime, anywhere, even while traveling or working overtime; built-in two-way microphone for remotely speaking to soothe stressed pets. AI automatically captures short videos of your pet’s food, saving footage before and after each meal. This allows you to easily check for leftover food, vomiting, and soft stool residue, enabling single pet owners to remotely monitor their pet’s health.
    2. Intelligent Feeding Algorithm + Independent Dispenser Control
      The app supports 1-10 timed feedings per day. Each of the two dispensers can be set to dispense different amounts of food per feeding, and two feeding modes can be freely switched:
    • Mixed Food Mode: New and old food are stored separately and automatically released in proportion, making the transition between food types easy;
    • Staple Food + Freeze-Dried Food Mode: Main meals are served as staple food, and freeze-dried snacks are given separately as supplements, achieving refined feeding; Built-in food surplus monitoring will send a pop-up notification to your phone if the food bin is low or there is leftover food in the bowl, preventing waste and food shortages.

    III. Summary of Daily Use Advantages

    Advantages

    1. Unique Dual-Compartment Design: Most feeders in the same price range are single-compartment. The 2nd generation dual-compartment design perfectly adapts to food changes and freeze-dried food supplements, broadening the usage scenarios for multi-pet families.
    2. Highly Practical Visual + AI Recognition Combination: Pure monitoring feeders on the market lack facial recognition, and pure AI feeders lack real-time video. This model achieves dual monitoring of vision and data.
    3. Comprehensive Ecosystem and Strong Compatibility: HarmonyOS connects to all Huawei devices. The APP has a simple operation logic, allowing even seniors to quickly set up feeding plans.
    4. Multiple Preservation and Power-Off Protection: Sealed, light-proof food compartment + desiccant for long-lasting freshness, dual power supply backup, eliminating the need to entrust food to others during short trips of 3-7 days.
    5. Easy to Clean: Seamless construction, ceramic food bowl doesn’t accumulate oil, long-haired cats don’t easily get fur on their food, and wiping after meals is simple. Objective Shortcomings in Actual Testing
    6. No built-in weighing module: Lacking accurate weight statistics, families with strict dietary control needs need to use a weighing bowl as an auxiliary tool;
    7. Occasional Wi-Fi disconnection in the app: Stable in 2.4G network environment, occasional screen delay occurs when switching to 5G, and a membership is required to watch videos;
    8. Slight accidental triggers in recognition: Cats walking past the food bowl or toys falling into the bowl may accidentally trigger feeding record reminders, requiring manual filtering of invalid captures;
    9. The power cord is relatively short, requiring an additional power strip when placed far from an outlet.

    IV. Suitable Users and Purchase Recommendations

    Most Suitable Users

    1. Families with two or more cats who frequently experience food competition and uneven feeding;
    2. Working professionals, frequent business travelers, or short trips who cannot return home daily to feed their pets;
    3. Those who enjoy remote pet interaction and need to monitor their pet’s status in real time to alleviate separation anxiety;
    4. Those who are switching pet food and supplementing their diet with freeze-dried food, requiring separate storage compartments for food.

    Not Recommended for:

    1. Single pet requiring precise weight control (e.g., cats with kidney disease, obesity, or post-surgery);
    2. Limited budget, requiring only basic timed feeding, without visual or AI recognition features;
    3. Large pets whose single-serving volume far exceeds the device’s maximum capacity.

    V. Comparison Summary at the Same Price Point
    Comparing the Hohmann Visual Feeder and a Standard Single-Compartment AI Feeder:

    • Hohmann Advantages: More stable food dispensing, lower app network failure rate, but lacks dual-compartment storage and facial recognition;
    • Standard Single-Compartment Feeder: Lower price, but cannot differentiate between multiple pets or store dry and wet food separately;
    • Petkit Visual AI Feeder 2nd Generation: Balanced overall functionality, combining dual compartments, 140° visual control, multi-pet AI recognition, and the HarmonyOS ecosystem—four core functions in one device—at a regular promotional price of 479 yuan, offering leading value among mid-range smart feeders. VI. Final Overall Score (out of 10)
    • Appearance and Build Quality: 9 points
    • AI Recognition Function: 8.2 points (Deductions: No weighing function, minor misidentification)
    • Visual Interactive Experience: 9 points
    • Food Storage, Preservation, and Safety: 9.5 points
    • Ease of Cleaning: 9 points
    • Overall Value for Money: 8.8 points

    Conclusion The Petkit Visual AI Feeder 2nd Generation doesn’t feature groundbreaking new technology, but it precisely addresses common pain points for pet owners: dual compartments for portioning food, AI facial recognition for managing multiple pets, a wide-angle camera for remote companionship, and comprehensive preservation and power outage protection significantly reduce pet anxiety. If you have multiple cats, frequently travel for work, and value visual interaction and precise feeding, this is a worthwhile all-around smart feeder in its price range. If your core need is extremely precise food control, it’s recommended to use it with a weighing bowl or choose a dedicated model with a weighing function.