⚠️ Testing prototype — not legal advice. This is a demo of search technology only. Its sample documents are simplified, illustrative summaries that may be incomplete or inaccurate, and results must never be used as a substitute for advice from a qualified lawyer. Do not rely on anything from this page for any real decision.

← Back to chat

RAG Pipeline — Interactive Demo

Semantic search, no LLM (V1). Each step below is a real call to the backend.

1. Chunking

Paste (or edit) some text, then split it into overlapping chunks the way chunking.py does.

2. Embedding

Turn the chunks above into vectors using BAAI/bge-m3. Only a preview of each 1024-dim vector is shown.

3. Storage

What's actually indexed right now, in the persisted vector store.

Loading…

View the corpus used →

4. Query

Ask a question. It gets embedded and compared against the indexed sample corpus — and, if you embedded text above, against your own text too (not indexed, compared live for comparison only).

?