Overview
Enables researchers to work across documents, conversations, data sources in unified environment. Led product design, information architecture, interaction frameworks. Built for MoreHarvest International, Chateau Life, Nexara Capital.
Challenge:
Translating abstract AI behavior into tangible, trustworthy interfaces.
Discipline
Product design · Research systems · AX
Client
MoreHarvest International & Chateau Life & Nexara Capital
Location
Taiwan · Japan · Hong Kong · Singapore
Date
2025 - present
Team
Lead Designer · Researcher · Frontend Developer · Backend Developer
Role and scope
Lead product designer. Product definition, information structure, interaction patterns, and UX language. Led testing cycles, transforming the product from an AI demo into a research tool.
Constraints
Translating AI behavior into tangible actions, building trust through visible sourcing, creating memory where chats resets, distinguishing action types, and maintaining context.
Development process
The system wasn't built all at once. Each phase established foundation before adding complexity. Tokens locked before components. Components validated before templates. Automation enforced rules from day one. No manual governance, no exceptions. Built methodically to prevent the chaos it replaced.
Research teams work in fragments. Tools don't share memory. Every tool switch breaks focus. AI chat made this worse. Sessions reset. Notes disappeared. Sources untraceable. Fundamental metaphor wrong: research is not conversation. Research is a continuous process.
01
The problem
The first version failed immediately. Users couldn't complete tasks. Terminology obscured intent. Nothing felt persistent. Core insight: they needed memory. They needed to build on their own work.
02
Discovery through testing failure
Restructured entirely:
Quick Notes → Fast capture, always saved.
Workstation → Primary surface where everything stays visible.
Channels → Organized threads by topic.
Summaries → Condensed insights with traceable sources.
03
From chat to research workspace
Previously impossible tasks became intuitive. Users trusted the system. They saved freely, built on previous research, and cited sources confidently.
04
Validation
Source attribution → Clickable sources.
Save state visibility → Real-time indicators.
Version history → See how thinking developed.
Human/AI separation → Clear visual distinction.
05
Designing for trust
Challenges
Primary: conceptual, not technical. Terminology became critical. Multi-topic synthesis required new patterns.
Insights
Researchers struggled with trust, not input. The solution wasn't better AI; it was better feedback. AI interface design is fundamentally different from traditional product design.
What's next?
Evolving into a collaborative surface. Shared channels, insight extraction, live co-editing, smarter memory.