![UIDEC allows UI/UX designers to specify design constraints [1], generate and organize design ideas on a canvas [2], refine or regenerate specific UI elements [3], navigate through different generated versions [4], and save and arrange design examples [5], among other features.](attachment:a37bb3bd-0c79-4145-8b1e-dc0059fca1c1:teaser_page-0001.jpg)
UIDEC allows UI/UX designers to specify design constraints [1], generate and organize design ideas on a canvas [2], refine or regenerate specific UI elements [3], navigate through different generated versions [4], and save and arrange design examples [5], among other features.
Project Context
Duration: July 20 – August 9, 2024
Role: UX Research Intern & Product Designer
Team: 2 Designers (Atefeh Shokrizadeh & Shivam Kumar), 2 Developers (Bahati Tadjuidje & Sohan Kamble)
Institution: Human-Centered Design Lab, Polytechnique Montréal
Supervisor: Prof. Jinghui Cheng
Publication: Dancing With Chains: Ideating Under Constraints With UIDEC in UI/UX Design, CHI 2025
Presentation Video: https://www.youtube.com/watch?v=VXHOH4VgHF8
UIDEC is a generative AI tool that helps UI/UX designers explore relevant design directions during early-stage ideation by working with constraints, not against them. Unlike traditional inspiration platforms like Dribbble or Behance that show visually appealing but context-agnostic screens, UIDEC generates tailored design examples based on real project constraints: industry norms, target users, brand identity, technical feasibility, and business goals.
Built from insights gathered through 18 in-depth designer interviews and validated through a 10-participant user study, UIDEC demonstrates that when AI respects the realities of design work (deadlines, brand guidelines, technical limitations), it becomes a genuine creative partner rather than a replacement. The tool allows designers to specify constraints through structured inputs, generate multiple screen variations instantly, edit specific UI components without starting over, and organize inspirations into mood boards, all while maintaining full creative control.
This project resulted in a peer-reviewed publication at CHI 2025, contributing new knowledge about how designers ideate under constraints and how AI tools can meaningfully support this process without overwhelming or replacing human creativity.