The problem
A color palette alone did not help me see which colors suited me. Repeating my context in a chat made visual recommendations difficult to compare consistently.
The product
A visual personal-style workspace combining colorimetry guidance, persistent profiles and image-generation APIs. Asynchronous tasks and saved results support comparison over time.
My role & implementation
I framed the opportunity, defined user flows and visual comparison criteria, selected the technology and directed implementation with AI coding agents. Personal use informed subsequent UX priorities.
Key product decisions
Keep context in persistent profiles; make color comparison the primary task; generate the selected color rather than requiring a full batch. Treat image generation as an asynchronous flow with visible progress and explicit recovery.
Iteration from personal use
Slow generation, inconsistent images and repeated timeout alerts exposed friction. The current code reuses saved simulations, shows compact queue progress and pauses pending work for shared errors. Visual consistency remains a quality criterion to evaluate, rather than an assumed result.
What this demonstrates
Multimodal AI product design: connect model capabilities to a usable experience, prioritize reliability alongside output quality and turn observed friction into concrete implementation requirements.
From prototype to real use
Next: test whether people can compare colors more easily, evaluate consistency across repeated generations and measure completion time, failed tasks and cost per useful result. This is an independent prototype, without measured user or commercial outcomes. Before production, I would work with a development team to review the code, architecture, security and maintainability, and prioritize any required changes.
Built with AI coding agents
I defined the problem, experience, technology choices and review criteria. AI coding agents generated the implementation code under my direction.