Back to work

Moha Intel

AI research workspace · Multi-source synthesis

Overview01/

An AI research workspace for investment analysts working across documents, conversations, and live data.

Role

Lead product designer owning product definition, information architecture, and interaction patterns through iterative testing cycles.

Team

Lead designer, researcher, frontend developer, backend developer

The challenge

Existing AI chat tools treated research like a conversation, but research is continuous . Sessions reset every time, notes disappeared, and sources were completely untraceable, so analysts had no way to build on their own work.

The approach

Restructured the product around four persistent research surfaces with embedded trust signals, source attribution, save states, and version history, validated through iterative testing.

Process02/
01/

The problem

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 were untraceable. The fundamental metaphor was wrong: research is not conversation. Research is continuous.

D
Drive
No file context
E
Excel
Lost sources
W
WhatsApp
Chat buried
P
PowerPoint
Manual copy-paste
AI
AI Chat
No memory
N
Notes
No traceability
Diagnosis
Severity...
6Tools
0Connections
40%Time lost
Issues found0/6
02/

Discovery through testing failure

The first version failed immediately. Users couldn't complete basic tasks. Terminology obscured intent. Nothing felt persistent. Core insight: researchers needed memory. They needed to build on their own work, not start fresh every session.

Version 1.0
Task completion failedcritical
Terminology confused usershigh
No persistencecritical
No session memoryhigh

Researchersneededmemory.

Image 1 of 2
03/

From chat to research workspace

Restructured the entire product around four persistent surfaces: Quick Notes: Fast capture, always saved. Workstation: Primary surface where everything stays visible. Channels: Organized threads by topic. Summaries: Condensed insights with traceable sources.

Chat tool
01
Fast capture, always saved
02
Primary surface, everything stays visible
03
Organized threads by topic
04
Condensed insights with traceable sources
Image 1 of 4
04/

Validation

Previously impossible tasks became intuitive. Users trusted the system. They saved freely, built on previous research, and cited sources confidently.

Round 2 testing

Validation results

Save without thinkingWork disappeared between sessions
Find any sourceSources were untraceable
Build on yesterdayEach session started from zero
Trust the outputCouldn't distinguish human from AI
0/4 passed
Image 1 of 3
05/

Designing for trust

Trust required visible proof at every interaction: Source attribution: Clickable sources so users could verify AI output. Save state visibility: Real-time indicators confirming persistence. Version history: Users see how their thinking developed over time. Human/AI separation: Clear visual distinction between user input and AI generation.

Trust architecture

You

Key risks of investing in Kumamoto?

AI

Population decline acceleratingTSMC inflates land values

Image 1 of 2
Reflections02/
  • Challenges1/3

    The primary challenges were conceptual, not technical. Terminology became critical. Wrong words derailed entire task flows. Multi-topic synthesis required new interaction patterns that had no existing precedent.

  • Insights2/3

    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: the interface must make the system's reasoning visible.

  • What’s next3/3

    Evolving into a collaborative surface: shared channels, insight extraction, live co-editing, and smarter memory.

Swipe to explore

Moha Intel — AI research tool case study