Lets people ask questions about long documents instantly
Intermediate24h1–3
Next.jsRAGLLM
16/20Shop Chat Auto-Responder from Catalog
Answer customer questions on price, stock and delivery from your own catalog.
Intermediate16h1–3 people16/20 fit
Build Shop Chat Auto-Responder from Catalog for small shops selling on chat apps in ~16h with 1–3 people. Stack: RAG + Vector DB + LLM + Node.js.
Why it wins: Real small-business pain with a clear handoff safety net.
Build flow
01 · Problem
Small shop owners answer the same chat questions all day and miss sales overnight.
02 · Solution
Index the shop's catalog and FAQ, answer with RAG grounded in that data and hand off to the owner when unsure.
03 · Build
- 1Import the catalog CSV
- 2Embed products and FAQs
- 3Answer with retrieval and citations
- 4Add a confidence threshold for handoff
04 · Demo
Ask 'do you have the blue one in size M?' and get a grounded answer with a product link.
MVP checklist
- Catalog-grounded answers
- Owner handoff
- Unanswered questions log
- Daily summary
Tech stack
| Tech | Role | |
|---|---|---|
| RAG | Retrieval over your docs | AI |
| Vector DB | Embedding search | Data |
| LLM | Language model | AI |
| Node.js | Backend runtime | Backend |
Reference repos
Open-source projects you can build on. Links checked 2026-10-31.
Related ideas
More AISnap a product and get a title, description, tags and price range ready to list
Beginner8h1–2Vision LLMNext.jsTailwind17/20Incoming tickets get a category, urgency and suggested reply automatically
Intermediate14h1–3EmbeddingsLLMPostgres16/20