The winners of the 2025 Wu Wenjun Artificial Intelligence Science and Technology Award, known as “China’s Highest Award for Intelligent Science and Technology”, were officially announced recently. The joint team formed by Unisound and the University of Science and Technology of China won the Second‑Class Prize for Scientific and Technological Progress for its original achievement project “Key Technologies and Industrial Applications of Multimodal Digital Experts”. This marks Unisound’s second win following its first receipt of the award in 2019. It ranks among the few enterprises in the industry that have claimed this top‑tier artificial‑intelligence honor twice, with its robust technical strength once again gaining authoritative recognition.

Named after Academician Wu Wenjun, the founding father of artificial intelligence in China, the Wu Wenjun Artificial Intelligence Science and Technology Award is China’s first and most authoritative science‑and‑technology award in the AI sector. Reputed as “China’s Highest Award for Intelligent Science and Technology”, it represents the nation’s top‑tier AI expertise and official endorsement. Back in 2019, Unisound secured its first award for achievements related to IoT AI chips, breaking bottlenecks for the large‑scale deployment of intelligent‑interaction technologies. Six years later, Unisound stood out amid fierce competition among nearly 1,000 leading research institutions and tech firms to win the prize for a second time. This accomplishment testifies to Unisound’s accumulated expertise and innovation capacity forged through sustained dedication to vertical artificial‑intelligence fields. It also highlights its outstanding contributions to tackling common industrial challenges and advancing high‑level self‑reliance in science and technology in the large‑scale application of specialized, high‑stakes scenarioses such as smart healthcare and insurance. It demonstrates the formidable strength of Chinese AI enterprises as they evolve from “technology followers” toward “standard‑setters”.

Unisound’s second‑award success stems from three core technologies rated as internationally advanced: multimodal vertical large model for healthcare, refined on‑chip edge small‑model technology, and multimodal digital humans with both emotional and cognitive intelligence. Past technical breakthroughs together with this authoritative validation from the Wu Wenjun Award create dual endorsement of “technical innovation and achievement accreditation”. This consolidates Unisound’s dual leading position in both the industrial breadth of AI‑powered smart IoT and the in‑depth application value for vertical industries. It also lends greater weight to Chinese AI innovations within global technological competition.

Currently, the implementation of general large models in specialized domains, especially high‑stakes scenarioses including healthcare, finance and transportation, generally faces three bottlenecks: high hallucination risks, high deployment costs, and poor explainability. These hurdles create an industry‑wide predicament where solutions are “too risky to adopt, impractical to use, and difficult to sustain”. Aligning closely with major national strategic demands, Unisound takes “specialized large models and Agents” as its core breakthrough point. It translates the vision of AGI into tangible progress through deep exploration of vertical scenarioses, positioning specialized Agents as a critical solution to pain points within high‑stakes industries. The award‑winning “Multimodal Digital Expert” is more than a superficial interactive form. At its core, it builds highly reliable specialized Agents capable of independent execution, domain‑specific cognition and decision‑making, alongside controllable risks and costs, tailored for real‑world use in high‑stakes sectors such as healthcare, finance and government administration. Built upon a full‑stack technical system of “cloud‑edge‑chip collaborative computing foundation, multimodal human‑computer interaction base, and vertical‑domain cognitive brain”, these Agents form closed loops spanning perception, comprehension, decision‑making and execution. They effectively drive AI’s transformation from a “conversational tool” into a “trustworthy industry expert”.

In healthcare: Unisound’s specialized medical Agents deeply integrate multimodal medical large models and clinical knowledge systems. They achieve over 90 % accuracy in detecting focal liver lesions. The solution claimed double championships at CCKS2023‑PromptCBLUE, and secured a grand slam across three tracks for “medical Agents, medical large‑language models and medical multimodal large models” in the MedBench 4.0 evalsuation in 2025. Beyond delivering auxiliary diagnostic suggestions, these Agents produce explainable reasoning workflows aligned with clinical pathways. They serve as genuine AI collaborators for physicians, substantially boosting diagnostic‑and‑treatment efficiency and reliability.

For edge‑side scenarioses: Through refined on‑chip small‑model technology, Unisound innovatively enables efficient, stable deployment and operation of specialized Agents on edge chips with low computing power and low power consumption. The technology delivers key value in cost‑sensitive scenarioses requiring high real‑time performance and inference accuracy, including consumer electronics, smart transportation and industrial inspection. It facilitates the “downward migration” of AI capabilities to edge‑end devices.

On the interaction front: The anatomically fine‑tuned “digital humans with emotional and cognitive intelligence” act as natural interaction carriers for specialized Agents. Overcoming technical obstacles for dynamic realism, they combine affective computing and domain‑specific expression. They further enhance service quality and credibility for Agents in livelihood‑oriented scenarioses such as government services, education and customer support.

Unisound has converted its Multimodal Digital Expert technology centred on specialized Agents into secure, scalable industry‑ready solutions. These deliver deep‑rooted empowerment for key sectors covering healthcare, transportation, finance and government administration. The technology not only resolves major challenges for large‑model adoption in high‑stakes scenarioses, but also fuels the digital transformation of the real‑economy and accelerates the generation and implementation of new‑quality productive forces. Winning the Wu Wenjun Award for the second time marks a major milestone in Unisound’s scientific‑innovation journey and serves as a vivid example of the high‑quality development of China’s AI industry. Moving forward, Unisound will stick to its development philosophy of “scenario‑driven innovation, technology‑based foundation‑building and ecosystem‑wide collaboration”. Driven by independent innovation, it will deepen R&D and industrial integration of AGI technologies, and continuously advance the industrial‑scale roll‑out of specialized Agents across more vertical scenarioses. Amid intensifying global technological competition among major nations, Unisound will partner with worldwide collaborators to promote inclusive access to intelligent technologies. It will contribute solid momentum toward building a globally competitive intelligent‑technology ecosystem, fostering deep integration between artificial intelligence and the real economy, and charting new directions for global AI development underpinned by core Chinese‑origin technologies.