Back to Home
Retrieval-Augmented Generation (RAG)

Vector Search

Vector search finds the document chunks whose meaning is most similar to your question.

This demo uses simple learning vectors. Real RAG systems use high-dimensional embeddings and optimized search algorithms.

Interactive Playground

Knowledge Base

Live Visualization
📄 Question → 🔢 Create Query Vector 0 / 6 compared

Top Matches

Similarity — Every Chunk

2D View (Optional)

★

Statistics

—
Question
6
Chunks Compared
0
Comparisons Made
—
Best Match
0%
Highest Similarity
No
Search Completed

How It Works

❓
Question
→
🔢
Query
Embedding
→
🔍
Compare with
Every Chunk
→
🏆
Rank
Results
→
🎯
Return Best
Matches
Every chunk gets compared. Vector search doesn't stop at the first match, it scores all chunks, then ranks them.
Closer in the 2D view means more similar. Chunks near the star share more meaning with your question.
Try editing a chunk's wording, similarity scores update instantly, showing how meaning (not exact words) drives the ranking.
💡
Key Takeaway

Vector search compares your question with every document chunk and returns the chunks whose meaning is most similar.

Previous: Embeddings in RAG Next: Retrieval