Helps people spot important emails without reading everything
Beginner12h1–2
Next.jsOpenAI API
15/20Product Review Pros and Cons Digest
Paste reviews and get honest pros, cons and who the product is not for.
Beginner8h1–2 people15/20 fit
Build Product Review Pros and Cons Digest for online shoppers in ~8h with 1–2 people. Stack: LLM + Embeddings + Next.js.
Why it wins: Quoting real lines with counts makes it more trustworthy than a generic summary.
Build flow
01 · Problem
Shoppers skim hundreds of reviews and still miss recurring defects.
02 · Solution
Cluster review sentences into themes, count how often each appears and have an LLM write a pros and cons table with frequency, quoting real review lines.
03 · Build
- 1Split reviews into sentences
- 2Embed and cluster them
- 3Summarise each cluster with counts
- 4Render the table with quotes
04 · Demo
Paste 200 reviews of a popular earphone and reveal the hidden 'left bud dies' theme.
MVP checklist
- Theme clustering
- Pros and cons with counts
- Quoted evidence
- Not-for-you section
Tech stack
| Tech | Role | |
|---|---|---|
| LLM | Language model | AI |
| Embeddings | Semantic similarity | AI |
| Next.js | Frontend + API routes | Web |
Related ideas
More AIDescribe what went wrong and get a firm, polite complaint with the right details
Beginner8h1–2LLMNext.jsTailwind17/20Drop an episode and get timestamps, chapter titles and show notes
Intermediate16h1–3WhisperEmbeddingsLLM16/20