Presented by Yanbin Song

Current AI design tools generate polished UIs instantly — but that polish arrives too early, right when designers should focus on structure and flow. Instead of exploring ideas, we end up cleaning up AI outputs.

This talk shows the opposite approach: intentionally constraining AI to produce simple, text-based wireframes — keeping things ugly so you can think clearly. Once the logic is solid, lightweight style rules apply a consistent look with minimal effort. You'll leave with a practical method to use AI without losing your design judgment.

Key takeaways

Identify the cognitive trap of premature fidelity: Understand how instant, high‑fidelity AI mockups create visual noise and pull attention away from core product logic.
Design low‑fidelity AI workflows end to end: Learn how to structure prompts and skills that force AI into low‑fi ASCII layouts for early validation, then translate those into styled wireframes using existing tools and design systems without getting stuck in manual pixel tweakin