青心意创亮相2026外滩大会 首次披露技术布局社交世界模型Dino WM

Key takeaways from the 2026 Bund Summit center on next-generation home companion intelligent products developed by Qingxin Yichuang: the company’s embodied AI offerings have entered the commercial rollout phase and will soon enter mainstream daily use.

1. Unlike common camera-equipped voice chat robots on the market, these products run on the social world model Dino WM, a fundamentally different architecture that embeds social comprehension into the inference layer. It anticipates how interactions will unfold to generate responses, coordinating body language, tone of voice, and interaction distance for full-body, coherent feedback. It maintains a consistent, stable persona, eliminating off-topic replies and contradictory behavior.

2. The products are initially launching for home use cases. For children, they serve as growth companions with dedicated backstories, capable of playing, tutoring, supporting language practice, and helping process low moods. For older adults, they act as calm-paced companions that proactively remind users to take medication and put on appropriate clothing, as well as track weather changes.

3. The first product, Amoo, has earned positive feedback from users across all age groups at trade shows in China and overseas, with many attendees expressing on-site purchase intent, making it an intelligent IP with cross-cultural appeal across Eastern and Western markets. Going forward, this category of products is poised to become daily intelligent companions that users will be reluctant to replace.

Qingxin Yichuang showcased its embodied AI product Amoo at two major industry events, the 2026 Bund Summit and IFA 2026. Leveraging its self-developed social world model Dino WM, the company has carved out a differentiated competitive path, offering clear reference value for brand building, product R&D, and commercialization strategy.

1. For brand differentiation, the company has steered clear of the homogeneous industry race to build robots that “look more human” or “deliver stronger task performance.” Instead, it focuses on creating embodied character IP that forges genuine emotional connections with users, building irreplaceable value through lifelike interactive experiences. Amoo has already secured cross-cultural user recognition, with event attendees voluntarily expressing purchase intent at show floors.

2. Product R&D is tightly aligned with real demands in core scenarios. For home use cases, the product is built with dedicated features for children’s companionship and age-friendly services for older adults. It delivers emotional value to users through a stable core persona paired with personalized, user-specific interaction alignment.

3. Its commercialization path departs from the conventional approach of “covering as many functional scenarios as possible.” Instead, it prioritizes making the character resonate deeply with users, unlocking cross-scenario expansion through emotional affinity for the IP, and challenging the prevailing industry view that social world models lack clear commercialization pathways.

The embodied social character category pioneered by Qingxin Yichuang demonstrates clear consumer growth potential, with its product rollout logic and business model offering new reference points for sellers’ product selection and operation strategies.

1. Tappable incremental market opportunities: There is large unmet consumer demand for home companion embodied AI products. Amoo has resonated with audiences across all age groups at both domestic and international trade shows: children are eager to interact with it, adults treat it as an equal communication partner, and older adults have clear demand for its age-friendly companion features. With high cross-cultural acceptance, the product has breakout consumer potential once it officially launches.

2. Actionable user operation insights: Sellers can move beyond the homogeneous competition of piling on hardware features, and instead focus on emotional value and memorable, irreplaceable character traits. Building user stickiness through consistent, stable interaction experiences helps avoid being trapped in price wars.

3. Pre-positioning for derivative business opportunities: These embodied characters will become interfaces connecting users to the digital world in the future, creating upsell potential for character IP peripherals and scenario-derived services. Sellers can track the product rollout timeline in advance and connect with relevant cooperation resources.

Qingxin Yichuang’s disclosed technical layout for the social world model Dino WM, and its long-term plan to build a “character universe,” send clear signals about product design directions and commercial cooperation opportunities for the manufacturing sector.

1. Capturable product design and production demand: Embodied character products supporting Dino WM’s capabilities require coordinated full-body expression across gait, eye gaze, tone, and interaction distance, with support for costume changes and multi-scenario motion displays. At the hardware level, products must adapt to home scenario requirements, balancing child safety, age-friendly design for older adults, and functional requirements for emotional interaction, environmental perception, and generalized operation.

2. Long-term supply chain cooperation opportunities: Qingxin Yichuang will not limit its lineup to the single character Amoo. Going forward, it will build a diverse character matrix based on the engineering capabilities of Dino OS, requiring supply chain support across industrial design, hardware production, and accessory supply, creating long-term, stable cooperation space.

3. Production capability optimization priorities: Unlike smart hardware delivered as a one-off final product, these devices require continuous optimization aligned with model iterations and user feedback. Factories can build flexible production systems connected to user data feedback pipelines to match the products’ ongoing evolution needs.

Qingxin Yichuang’s first public disclosure of the technical pathway for its social world model Dino WM clearly illustrates the iteration direction of the embodied AI track and existing industry pain points, pointing to service entry points and technical collaboration directions for service providers.

1. Clarified core industry pain points: Most existing world models focus only on the physical relationships between objects, placing social recognition at the input layer. These are essentially camera-equipped chatbots capable only of surface-level verbal dialogue, lacking full-body coordinated emotional interaction capabilities. Widespread issues include inconsistent personas, stiff interactions, and an inability to build long-term emotional bonds with users.

2. Referenceable technical solution design logic: Dino WM integrates social comprehension into the inference layer, modeling motion, state, emotion, and relationships simultaneously under a unified prediction framework. Paired with character setting and reinforcement learning from human feedback (RLHF) mechanisms, it delivers persona-consistent, personalized interactions for different users, with three core capabilities: emotional interaction, environmental interaction, and generalized operation.

3. Exploitable supporting service opportunities: Around the home scenarios where embodied characters will be deployed, service providers can offer scenario-adapted content, user feedback data annotation, and model training support, to match enterprises’ iteration needs of “refining models while building products.”

The new embodied social character track represented by Qingxin Yichuang sends clear industry signals for platforms’ merchant recruitment, product selection, operation adjustment, and ecosystem layout.

1. Capturing new track recruitment dividends: Home companion embodied AI products with emotional interaction capabilities are moving toward commercial rollout. Intelligent IP products like Amoo, which boast high cross-cultural acceptance and proactive user purchase intent, can drive new consumer growth for platforms. Platforms can reach out to brands early to secure launch exclusivity and first-mover cooperation rights, capturing traffic dividends from the new product category.

2. Pre-positioning for future ecosystem layout: In the long run, embodied characters will become personalized recommendation portals connecting users to the digital world, integrating consumption, services, content and other multi-scenario touchpoints. Platforms can build ecosystem interfaces compatible with these intelligent terminals ahead of time, integrating their product and service capabilities into the embodied characters’ recommendation systems to capture the next generation of traffic entry points.

3. Operational risk mitigation: Users’ core expectations for these intelligent products are stable personas and warm, natural interactions. When onboarding relevant products, platforms should rigorously assess technical maturity, and avoid promoting low-quality products that only support simple voice dialogue and have inconsistent personas as flagship embodied AI offerings, which would erode user trust in the new category.

Qingxin Yichuang’s disclosure of the Dino WM social world model technical pathway and commercialization exploration at the 2026 Bund Summit provides a new observation sample and research direction for embodied AI industry research.

1. Emerging industry technology trends to monitor: Current world model R&D generally focuses on physical object-to-object relationships, with industry competition concentrated on two core dimensions: “degree of human likeness” and “task performance.” Qingxin Yichuang has pioneered a third technical pathway, integrating social comprehension into the model’s inference layer to build a prediction framework covering physical environments, social outcomes, interaction rhythm, long-term memory, and relationship states. It realizes an interaction logic of “prediction as input,” balancing persona stability and personalized interaction through character setting and reinforcement learning from human feedback mechanisms.

2. Practical exploration of pressing industry questions to track: In response to widespread industry skepticism that social world models lack clear commercialization paths, Qingxin Yichuang has selected homes as its first landing scenario, building user stickiness centered on emotional connection. It proposes the logic of “where the character goes, scenarios open up,” positioning embodied characters as personalized recommendation engines for next-generation intelligent terminals and interfaces connecting users to the digital world.

3. New industry evolution trends to observe: The company has explored an iteration path of “refining models while building products,” driving model evolution through real-scenario interaction data. As the technical architecture proves viable at scale, the industry will see the emergence of a “character universe” ecosystem built on unified system capabilities, with embodied character IP becoming a new form of intelligent product with cross-cultural acceptance.