Lenovo Group hosted its "Tech World 2025" innovation technology conference today at the Shanghai World Expo Center, unveiling its "Super Intelligent Agent" matrix aimed at personal, corporate, and urban applications. This development marks a significant leap in edge AI inference capabilities, bringing them closer to cloud-level performance and ushering artificial intelligence into a new phase characterized by "strong inference and edge computing."
During the conference, Lenovo’s Chairman and CEO, Yang Yuanqing, outlined the three core functions of the Super Intelligent Agent: perception and interaction, cognition and decision-making, and autonomy and evolution. Yang noted that these technologies have transitioned from conceptual phases to large-scale applications, anticipating that "within the next 12 months, edge AI capabilities will experience at least a threefold enhancement."
Lenovo officially launched three main product lines: the “Tianxi Personal Super Intelligent Agent,” “Leshang Enterprise Super Intelligent Agent,” and the “City Super Intelligent Agent,” alongside a new generation of inference acceleration engines. This engine, co-developed with Tsinghua University and Wuren Chip, employs high-performance parallel decoding, heterogeneous computing architectures, and operator fusion optimization to tackle industry challenges related to model lightweighting and inference efficiency.
On the personal front, Lenovo’s Tianxi Super Intelligent Agent offers cross-device data access and features multi-modal perception, intent understanding, and independent task breakdown capabilities. The company also introduced four new smart devices integrating Tianxi technology: the AI Metaverse version of the Moto Razr foldable phone, the Yoga tablet, the world’s first rollable-screen AI PC ThinkBook (expandable to 16.7 inches), and a 3D gaming device. Current Lenovo users can upgrade to the Tianxi Personal Super Intelligent Agent if their devices meet specific hardware requirements.
In terms of enterprise applications, Lenovo has developed a hybrid computing infrastructure based on an "end-edge-cloud-network" architecture, launching the “Wanquan Heterogeneous Intelligent Computing Platform” as a computational foundation for enterprise intelligent agents. This platform utilizes groundbreaking GPU kernel virtualization technology, co-developed with Shanghai Jiao Tong University, to significantly reduce AI training costs and enhance inference efficiency.
The Leshang Enterprise Super Intelligent Agent fully integrates a company’s business data and knowledge assets, covering the entire business chain from supply chain management to marketing. It can function as a product manager, sales representative, or purchasing agent, facilitating deep integration between silicon-based systems and physical operations.
On the urban governance front, Lenovo introduced the “1×N City Super Intelligent Agent” framework, which coordinates a central city hub with multiple domain-specific intelligent agents. This technology has already been implemented in cities such as Wuyishan and Yichang, demonstrating its value in intelligent travel planning, cultural resource optimization, and fine-tuned urban operations.
In a significant announcement, Lenovo also confirmed a technological partnership with FIFA for the 2026 World Cup, pledging to provide intelligent infrastructure support. This collaboration will involve developing AI analysis engines to assist teams and coaches in real-time tactical adjustments, creating 3D player avatars to aid referee decisions, and enhancing fans’ viewing experiences through immersive technology.
