Upload a CSV, ask a question and get a chart plus the SQL that produced it
Intermediate14h1–2
DuckDBLLMCharts
18/20Open-Ended Survey Theme Finder
Upload survey responses and get themes with counts and quotes.
Intermediate12h1–2 people17/20 fit
Build Open-Ended Survey Theme Finder for product teams, ngos and student councils in ~12h with 1–2 people. Stack: Embeddings + LLM + Python + Charts.
Why it wins: Works for any organisation, so it scores high on impact.
Build flow
01 · Problem
Nobody reads 800 open-ended survey answers, so the best feedback gets ignored.
02 · Solution
Embed responses, cluster them, have an LLM name each theme and pick representative quotes, and show counts by segment.
03 · Build
- 1Load responses from CSV
- 2Embed and cluster
- 3Name clusters with the LLM
- 4Chart themes by segment
04 · Demo
Upload real hackathon feedback and show the top five themes.
MVP checklist
- Theme clustering
- Named themes
- Representative quotes
- Segment filter
Tech stack
| Tech | Role | |
|---|---|---|
| Embeddings | Semantic similarity | AI |
| LLM | Language model | AI |
| Python | Backend / scripting | Backend |
| Charts | Data visualisation | UI |
Related ideas
More AISee where in a call the customer got frustrated and why
Intermediate16h1–3WhisperLLMCharts16/20Speak a two-minute journal entry and see mood and themes over weeks
Beginner10h1–2WhisperLLMCharts16/20