How Alibaba’s AI‑Powered Design Platform “Luban” Revolutionizes Mass Creative Production

On April 27, the UCAN 2017 User Experience Design Forum in Hangzhou’s Alibaba Xixi Campus opened with the launch of Alibaba’s intelligent design platform Luban by Yang Guang, chairman of the Alibaba Group UED Committee. The platform uses artificial‑intelligence algorithms and extensive data to train a machine‑learning model that can produce designs comparable to professional designers.

According to Wu Chun‑song, head of Alibaba’s Intelligent Design Lab, the platform was built on Alibaba’s massive commercial ecosystem data, powerful algorithms, and computing resources, with development starting in July 2016. Luban now offers three key features: large‑scale design generation, self‑learning growth, and open integration.

During the 2016 Double‑11 shopping festival, Luban personalized ad creatives for millions of users, producing up to 30,000 images per sub‑event and a total of 1.7 billion assets, a workload that would have required 100 designers working nonstop for 300 years. This automation doubled click‑through rates and dramatically improved design efficiency.

Luban operates as a self‑learning machine that mimics human design learning. Its closed‑loop system consists of three core modules: “style learning,” “actuator,” and “evaluation network.” After ingesting large volumes of design files, the system learns styles; the actuator generates designs based on requirements; the evaluation network assesses quality using trained image and rating data.

The platform will be opened to millions of designers in Alibaba’s ecosystem before the next Double‑11, enabling small and medium businesses to generate high‑quality designs at low cost, while large brands can input their own style to produce massive, consistent visual assets.

Yang Guang predicts that while routine graphic work may be replaced by AI, designers will shift toward content‑creativity and end‑to‑end design experiences. He proposes three strategic directions:

From product design to design product : Designers must anticipate consumer aesthetics and create products that resonate emotionally.

From experience design to design experience : With technologies like AR/VR, designers need to master multi‑dimensional interaction to craft seamless shopping experiences.

From technology upgrade to design upgrade : Leveraging big data and Luban’s capabilities enables personalized product presentation and empowers the entire design ecosystem.

Through continuous data feedback, Luban’s AI continually refines its output, creating a virtuous cycle where new technology fuels new design, and new design drives new commerce.