Defining Explainable AI for Creative Workflows
Perception is an AI-powered color platform that enables designers to search, generate, and refine palettes using semantic mood bridging creative intuition with explainable AI.
Leadership Scope
Led product strategy and UX direction from concept through beta
Defined core principles for AI-assisted creative workflows
Drove alignment on usability, trust, and complexity tradeoffs
Built a modular system for scalable AI interactions
Opportunity
How might we use AI to simplify color workflows while preserving designers’ intent and avoiding opaque, “black box” outputs?
Approach
Defined product direction early by positioning semantic/mood search as the primary workflow to align with how designers think and explore ideas.
Made AI outputs understandable and trustworthy by introducing explainability patterns such as mood mapping and confidence cues so the system felt like a collaborator rather than an unpredictable generator.
Managed complexity through system design by evolving from a single-surface prototype into modular workflows that allowed users to move seamlessly from search to generation to refinement.
Built reusable components for scale by including tags, sliders, and generation modules that could extend across the product as it evolved.
Validated workflows through iteration by refining interactions and outputs based on usability testing and feedback.
Key Decisions
Prioritized semantic search over traditional color tools
Treated explainability as a core requirement, not a feature
Shifted to modular workflows to support scale and usability
Outcome
Beta launch in June 2023
10,000 users within six months
Expanded reach through a partnership with Playbook, a creative platform boasting over 800,000 users.
Scaling Impact
Established reusable patterns for explainable AI and modular interaction design enabling the system to extend beyond its initial feature set.
©2026 • Elliot Yamashiro