Interactive · how RAG actually works

How AI answers from your documents

In plain terms: instead of guessing, the AI looks up the most relevant passages in your documents, then answers using only those — and shows its sources. Paste any text below and watch it happen.

Runs in your browser; scoring is a transparent stand-in for a production embedding model so you can see why each passage was picked. In production: Snowflake Cortex Search or a vector store (Pinecone, pgvector) + an LLM.
1 · Your document
passage size
2 · Ask a question
3 · Grounded answer

Cited passages: · indexed · top- retrieved
Why these passages — ranked
The pattern, in production
1 · Split
into passages
2 · Embed
into vectors
3 · Retrieve
most relevant
4 · Ground
answer on them
5 · Cite
show sources