Re‑building Design Productivity in the AI Era: Xiaomi’s AI Design Workflow

As the industry discusses whether AI will replace designers, one critical question is ignored: Can AI reshape the fundamental design‑collaboration workflow?

The traditional pipeline"user feedback → business requirements → product solution → design → development"suffers from constant information loss. Most AI‑powered approaches still follow old patterns: AI outputs HTML prototypes for manual rework, failing to fix information decay.

Xiaomi's internal digital team chose a different route: AI directly generates production‑ready code instead of HTML prototypes.

The team has evolved from individual skills to capability sets, then to multi‑agent orchestration. With an"AI expert panel", one designer can complete the full workflow: requirement clarification, solution breakdown, UI design and review, achieving one‑shot information alignment.

If you care about genuine AI integration into design workflows rather than superficial AI‑aided drawing, this talk unpacks the thinking and implementation behind this path:

1、Bottlenecks in traditional design pipelines
2、Rationale and trade‑offs of direct production‑code generation
3、Three‑phase capability evolution toward multi‑agent collaboration
4、How design knowledge base grows into a context‑aware"design brain"

More than an AI‑tool demo, this practical case explores how to redefine designers' roles, competencies and collaboration modes when AI can understand and make judgements.

It brings insights for design leaders, product & tech leads, and all AI‑design practitioners, previewing future‑oriented working paradigms.

1、Understand real B‑end workflow shifts in AI era: redefine designers' roles & value, not replace them.
2、Learn to build design knowledge bases: turn AI into a business‑savvy collaborator beyond drawing tools.
3、Gain actionable insights from Xiaomi's internal B‑end cases & evolution experience.