How It Works
1
Upload customer feedback from any source (CSV export from surveys, tickets, reviews)
2
Claude analyzes each piece of feedback for sentiment and categorizes by theme
3
Dashboard displays priority-ranked insights with volume and sentiment trends
4
PMs see exactly where to focus without reading every comment
Product Decisions
| Component | Technology | Purpose |
|---|---|---|
| Frontend | Streamlit | Interactive dashboard with filters and visualizations |
| Analysis | Claude API | Sentiment detection and theme categorization |
| Backend | Python (Pandas) | Data processing and aggregation |
| Development | Claude Code | Rapid prototyping and iteration |
Built with Claude Code: Used Claude Code to build the Streamlit app, design the categorization prompt, and iterate on the dashboard layout. Focused my time on UX decisions while Claude Code handled implementation.
What I Learned
- 💡 Instead of defining categories upfront, I let Claude identify themes from the data. It caught recurring issues I wouldn't have thought to look for.
- 💡 I wanted PMs to have something to act on, so I made insights specific. For example, "23 users this week said the export button is hard to find" instead of just "negative sentiment detected."