| name | design-research-with-claude |
| description | Use when leveraging AI (Claude, GPT) for design research, analysis, and synthesis. Covers the workflows that work (transcript synthesis, competitive teardowns, copy review, design audits), the limits, and the prompting patterns.
|
Design Research With Claude
AI is a design-research force multiplier. It doesn't replace research; it accelerates it.
What AI is great at
Synthesizing user interview transcripts into themes. Drafting research questions. Critiquing your design (paste screenshot, ask for issues). Generating personas from real data. Translating jargon-heavy customer language into clean themes. Reviewing copy for clarity / tone consistency.
What AI is bad at
Recruiting users. Conducting interviews (no rapport). Empathy for specific subcultures without context. Original strategic insight (good at variation, rare at leap).
Workflow: transcript synthesis
Paste raw interview transcript. Ask: 'What were the top 3 themes? What surprised you? What direct quotes capture each theme?' Iterate.
Workflow: competitive teardown
Paste competitor's homepage HTML or screenshot. Ask: 'What's the value prop? What's the hero pattern? What trust signals? What would you steal? What would you avoid?'
Workflow: copy review
Paste your copy. Ask: 'Does this read like a marketing pitch or like a real human? Where could a 12-year-old not follow? What words would I cut?'
Limits
AI doesn't know YOUR users without you teaching it. Can produce plausible-but-wrong syntheses. Always validate with real users for strategic decisions.
Common mistakes
Treating AI synthesis as user research (it's analysis OF research, not research itself). Generic prompts → generic output. Skipping human review of AI output before acting.
Where this fits in X3 Compass
Applied across the X3 Compass design system.
