How Alibaba’s AI‑Powered UED Platform Transforms Design Outsourcing

From Tmall to Cainiao, from Double 11 to the Creation Festival, Alibaba’s UED outsourcing team of over 700 designers supports more than 50 business units with tens of thousands of design requests each year. Coordinating these external designers and providing fair settlement prices has been a major challenge.

From Distribution to Settlement, Fully Controllable

To address these pain points, Alibaba’s Information Platform team applied SOLID principles and a DDD‑inspired layered architecture, using DIP to separate the system into Interface, Infrastructure, API, and CQRS layers.

The Infrastructure layer integrates ACL, finance, procurement, workflow, and support systems, enabling end‑to‑end online processing from task distribution to verification and settlement. Budget control prevents overspending, and finance can settle automatically without manual hand‑off. Delivered assets flow into Alibaba’s central material library for reuse.

For security, open API interfaces allow external designers to connect quickly while protecting core data.

Because design requirements and personnel change frequently, the platform adopts the CQRS pattern to separate command (write) and query (read) operations, improving scalability and safety.

Task data is fetched using a hybrid long‑short polling method, presenting workflow tasks in a familiar Photoshop‑like UI to help designers onboard swiftly.

Intelligent Image Review for Original Design

Originality is crucial in design. The platform classifies assets (A, B, C) based on originality, usage, and duration, with significant price differences and copyright implications. Previously, manual reviews were inefficient and inconsistent.

Now, AI algorithms automatically analyze PSD source files, comparing content features, size, texture, and color against historical data to determine originality and complexity, achieving over 90% accuracy.

For example, four submitted banner assets were evaluated; only one met the creative criteria for type C, while the others were re‑classified as type B, a judgment that would be difficult manually.

The system also parses layer, size, and position data to assess material complexity, supporting smarter material recommendation and reducing design‑to‑development friction.

Platform Integration Empowers Designers

Alibaba’s Design Committee envisions a three‑pillar strategy: expanding designer types, enriching the material center for reuse, and leveraging external design resources. The UED platform serves as a crucial hub, feeding high‑quality assets into AI models that further enhance design automation.

Future plans include integration with supplier management and the “Luban” intelligent design system, enabling self‑learning distribution and maximizing resource utilization.