Taipei, December 11, 2025: Dimerco Express Group ("Dimerco") has announced that it is strengthening its integrated, Asia-focused logistics solutions to help customers capture growth from AI and "Taiwan+, China+" manufacturing while managing risks from congestion, regulatory changes and shifting capacity

At COMPUTEX 2025, ASUS showcased its comprehensive AI infrastructure solutions, targeting the computing needs of data centers, edge computing, and a wide range of intelligent applications. With a robust product portfolio, ASUS brings to life the NVIDIA AI Factory vision, illustrating how AI will be deeply embedded into enterprise IT infrastructure. This evolution continues to drive forward applications and services that shape the future of AI and high-performance computing

Technology news, product launches and show highlights from the event attendees available for download (PDF).

Tokyo-based robotics startup Yukai Engineering will debut Baby FuFu, a portable fan robot designed for babies and toddlers to play with while keeping themselves cool. The robot's unique fan mechanism keeps the blades fully enclosed, making it safe even for babies to handle. It also securely attaches to baby strollers, enabling parents to fan their children hands？free.Designed to look identical to - but twice the size of - Nekojita FuFu, Yukai's popular drink？cooling robot, Baby FuFu is aimed at making personal cooling more accessible and fun amid the surging global popularity of portable cooling devices."Baby FuFu grew out of Nekojita FuFu fans' feedback that their children love not only cooling their food with it but also playing with it, pretending to fan their faces and blow？dry their hair," said Shunsuke Aoki, CEO of Yukai Engineering. "Many users also wanted FuFu made into a fan, so we created a robot that encourages toddlers to cool themselves with it to prevent heat stroke in hot weather, all the while having fun. Portable fans are now an essential summer item in many parts of the world, and we are very excited to introduce this adorable robot at CES 2026."Safety FeaturesBaby FuFu uses an internal fan to pump air from bottom to top. A slit plate in front of the fan prevents direct contact with the blades, even if a finger or other object is inserted through the robot’s mouth, making it safe for users of all ages.Use on Baby StrollersBaby FuFu's hands and feet are shaped to securely latch onto stroller handles. Users can easily adjust the robot’s angle to direct airflow onto a targeted area, such as a child's face.A press of a button allows users to switch among three airflow strength modes.Also on Display: Mirumi Now Live on KickstarterIn addition to Baby FuFu, Yukai will showcase many of its award？winning products during CES 2026, including Mirumi, the charm robot that turns around to steal a look. Mirumi is currently live on Kickstarter and will remain so throughout CES 2026; the campaign has already raised nearly $150,000 in funding within its first 10 days and runs through January 21.About Yukai EngineeringYukai Engineering is a Tokyo-based startup known for creating robots that bring joy to life. The company's innovations have earned numerous accolades, including recognition in TIME magazine's Best Innovations of 2025 for Nekojita FuFu (a drink？cooling robot), CES 2023 Innovation Awards for the BOCCO emo Platform (used by several Japanese utility companies to provide smart？living services), and fufuly (a robotic cushion for guided breathing that also received a 2024 Good Design Award). Many of Yukai’s flagship products successfully launched on Kickstarter, including Nekojita FuFu, which raised more than $120,000 from nearly 1,800 backers worldwide during the April–June 2025 campaign. Qoobo (a therapeutic pillow for companionship) also achieved global succes.

Service robots are moving quickly into healthcare, security, and public environments. Yet their increasing reliance on visual and spatial perception technologies has introduced a major hurdle: uploading raw visual data to the cloud now poses significant privacy and regulatory risks. European and U.S. regulators have recently designated mobile image collection as a high-risk activity, shifting the robotics industry from a "performance-driven" model to one fundamentally shaped by compliance requirements. This shift is redefining what it takes for robotic systems to scale safely and legally.To address these emerging requirements, DeCloak Intelligences has developed a portfolio of privacy-preserving AI systems that rearchitect how visual intelligence is handled at the edge. This includes DeCloakFace, designed for safeguarding personal biometric traits; DeCloakVision, tailored for medical and caregiving environments; and the company's third-generation core platform DeCloakBrain, developed specifically for autonomous robotics. DeCloakBrain functions as a generic AI decision-making engine for robotic autonomy, enabling spatial perception and task execution without relying on sensitive visual data, making it a foundational technology for future large-scale deployments across regulated environments.According to DeCloak President Dr. Yao-Tung Tsou, the company's core innovation lies in "privacy-by-design at the data source," ensuring that all sensor inputs - such as images and numerical data - undergo irreversible de-identification directly on edge devices at the moment of capture. This is achieved through multilayer techniques integrating differential privacy on edge hardware, image anonymization, spatial computing, global visual fusion, and federated learning. With its proprietary irreversible de-identification technology, DeCloakBrain maintains high-accuracy, low-latency decision-making without using original sensor data. The technology is already modularized across software and hardware, enabling rapid, secure deployment and seamless integration with AMRs, medical service robots, humanoids, and quadruped robots, while supporting scalability in both performance and functionality.From a compliance standpoint, DeCloakBrain aligns with stringent privacy regulations in the US and EU. This allows international customers to meet privacy and safety requirements for public spaces, medical applications, and defense-related use cases, mitigating cross-border deployment risks. Dr. Tsou emphasized that DeCloak's intelligent system is built upon years of accumulated datasets, patented de-identification models, VLM/VLA scenario libraries, and long-standing integration experience with global robotics supply chains. This foundation forms a difficult-to-replicate ecosystem for privacy-preserving intelligence, offering verifiable data protection and robust multi-device interoperability.DeCloak enters a pivotal scale-up year as Japan, US, and Europe markets take shapeIn its global expansion strategy, DeCloak identifies Japan as its fastest-moving commercialization market, with public-space monitoring and smart healthcare positioned as the initial priority sectors. The company's integrated systems for these applications have completed deep technical alignment and are now progressing into phased deployment.In the United States, market development is led by partners specializing in military and law-enforcement robotics. Their systems - already used in defense and policing scenarios - are undergoing military - and police-grade certification, targeting applications such as public-area patrol and security inspection. In Europe, where data and imaging regulations are among the world's strictest, DeCloak follows a software-licensing and system-integration model that enables local manufacturers to rapidly incorporate privacy-protection capabilities into existing hardware platforms.Looking ahead, DeCloak plans to scale shipments across Taiwan and Japan while supporting major clients in deploying smart healthcare, public-space security, and related applications. Japan is expected to become the company's first overseas market to reach meaningful revenue scale. In the United States, following nearly a year of testing DeCloakBrain hardware and software modules with military and law-enforcement partners, the company anticipates entering a "mass production and shipment" phase in 2026, covering use cases such as defense systems, public-area surveillance, and inspections in high-risk operational environments.Additionally, DeCloak's US partners will arrange on-site demonstrations for key customers during CES 2026, serving as a critical evaluation step prior to formal procurement. Europe will enter a full-scale development phase in 2026, centered on software licensing and system integration, enabling manufacturers in Poland, the UK, Germany, and other regions to meet stringent regulatory requirements while adopting solutions for smart healthcare, public-space monitoring, and data-privacy protection.President Dr. Yao-Tung Tsou emphasized that as DeCloak's multi-market strategy matures, the company is positioned at a crucial inflection point where technological readiness converges with commercial acceleration. He noted that 2026 will mark DeCloak's transition from "technology completed" to "global scale-up," with the company committed to expanding real-world deployments and advancing privacy-preserving intelligence and autonomous robotics onto the global stage - defining the next generation of AI-driven robotic intelligence standards.DeCloak's next-generation privacy-preserving AI universal decision platform for robotic systems. Credit: DeCloak

As generative AI continues reshaping industries worldwide, enterprises are accelerating adoption across production, inspection, and automation workflows. Yet, as the era of human-machine collaboration and robotics unfolds, traditional AI systems still face two major deployment barriers.The first challenge is the lack of adaptability: when on-site conditions change - such as in visual inspection scenarios where production lines switch tasks, new defect types emerge, or inspection conditions shift - conventional AI models often require days or even weeks of retraining, falling short of real-time operational demands.The second challenge is catastrophic forgetting: when a model learns new information, it often loses accuracy on previously mastered tasks, a long-standing technical hurdle in deep learning.To address these pain points, Memorence AI developed its proprietary instant learning technology, inspired by the mechanisms of the human neocortex. The technology significantly shortens the time needed to adjust AI models while maintaining stability of prior knowledge. This provides a critical foundation for real-world human-machine collaboration. The company has secured invention patents in Taiwan, the United States, Japan, and China, laying the groundwork for its global expansion.Founder and CEO Pai-Heng Hsiao explained that the core principle is designed to support human-inspired capabilities such as perception, association, understanding, reasoning, and memory. For example, when an operator identifies a misjudgment and corrects it, the system can immediately incorporate the new information and apply it to subsequent tasks. This capability delivers rapid feedback and can be extended to support multi-modal knowledge applications.Focusing on Three Application Domains, Gaining Traction Among IPC VendorsMemorence.AI aims to bring AI from the cloud to the front line. Its end-to-end AI solution, Memorence Suite, is designed for large-scale visual inspection tasks and features a no-code architecture, allowing enterprises to deploy AI without relying on in-house engineers. One major industrial customer reduced manual inspection miss rates from 2-4% to below 0.1%, significantly improving quality and operational efficiency.Operagents targets high-variability environments in human–machine collaboration and future robotic systems. Leveraging instant learning, the solution responds quickly to changes in workflow, operating procedures, and recognition conditions. After deployment at a major manufacturer, operators were able to provide corrections that the system rapidly incorporated within a human-in-the-loop workflow, updating guidance instructions in real time, effectively reducing assembly errors and minimizing variation among different personnel.Memoragents provides a multi-modal knowledge query system that enables companies to continuously accumulate and update domain expertise on the shop floor. This allows employees to resolve technical issues quickly while mitigating gaps in institutional knowledge and expertise transfer.Together, the three solutions - from large-scale visual inspection to real-time operational assistance and multi-modal knowledge understanding - form a complete foundation for next-generation human-machine collaboration and robotics applications.Hsiao stated that Memorence AI's instant learning technology and AI solutions form a complementary advantage with the hardware offerings of IPC vendors. Customers can integrate IPC systems with Memorence AI products to enhance the applicability and competitiveness of their overall solutions. In 2025, the company partnered with a leading IPC manufacturer to participate in a major European trade show, leaving a strong impression on local industry players and laying a solid foundation for its market expansion in Europe.Advancing Toward CES2026 and Accelerating Global ExpansionWith support from Taiwan Tech Arena (TTA), Memorence AI is set to participate in CES2026 and has been named a CES Innovation Awards 2026 Honoree. Given the United States' strong foundation in advanced manufacturing, precision manufacturing, and large-scale automation - coupled with the rapid growth of human-machine collaboration, robotics, and the stringent quality and process requirements in aerospace and other high-spec sectors - Memorence AI will showcase the instant learning capabilities of Operagents in operational guidance and quality assurance applications. The company will also demonstrate its multimodal visual-and-text knowledge base solutions, enabling enterprises to continuously update professional knowledge and workflow procedures directly on-site.Hsiao noted that Taiwan's government has significantly increased support for startups in recent years, providing strong momentum for industry-wide development. He expressed deep appreciation for TTA's long-term assistance, particularly the opportunity to join the CES delegation, which has not only boosted Memorence AI's global visibility but also strengthened its confidence in expanding its international presence. Moving forward, Memorence AI will continue to advance AI and instant learning technologies, expand into additional modalities and interactive capabilities, and support global partners as they embrace a new era of human-machine collaboration and robotics-driven applications.Operagents offers operational guidance and adapts via instant learning with human oversight.Credit: TTA