Goal and Value
This is a case study of a tool design and development project that integrates AI design systems and LLMs. This tool addresses three major challenges faced by most AI-powered design agents:
Through hands-on design practice (combining design strategy) ranging from initial sketches to high-fidelity prototypes, I guide other designers step by step on how to create AI conversational user experience(CUI) designs based on AI design system.
Establish best practices for the new AI based design process
AI-Search is an extremely challenging technical journey, compress the features that would normally take at least three months to develop into a single four-week design, allowing the AI and dev teams to polishing the technical details with greater certainty
Critical competitor analyzation
By analyzing critical competitors in the market, I led a cross-functional & non-local team to identify the specific pain points faced by local users, and determined the appropriate implementation strategy for our design solution.
 | |
PainPoint | Design strategies |
|---|---|
➊ Only one-time, disposable input | Let user continuous input and multiple rounds of search/reply search (+natural language) |
➋ Complex filtering interactions and logic | Reduce any filter, guide user to prompts and natural language |
➌ Interruption-based interaction | Create an agent-like experience by designing the search interaction as a multi-threaded process. |
➍ Rich results as expected by user vs. listbox output | Enhance results to |
AI design process
The new design process and system offer better storytelling capabilities, making the design mockups more closely resemble the actual product. By incorporating feedback from local users, makes it easier for cross-time-zone teams to understand the true nature of the product.
Building upon traditional design processes/models (such as the double diamond model), the new one added AI scenario simulation and AI content innovation, and combined it with my new AI console, resulting in more precise AI prerequisites and more powerful design multimodal output templates.
Multimodal Design
Compared to most AI agent designs, which primarily rely on text-based chat, Design system components can 100% handling images, text, voice, and even interactive data visualizations simultaneously.
The design system components achieve 100% code compatibility (design = code).
Design system, design strategy, and new interaction framework
This design also incorporates key elements of the revamped design: a consistent design system, a user-friendly interface, and the unique appeal of native AI.
