Website Assistant (RAG)
Powered by AI SDK embeddings & semantic search
Loading...
🔍
RAG-Powered Search
Click 'Index Website' to populate the knowledge base first
🧠 How Embedding RAG Works
- 1. Ingest: Crawl website & chunk content into pieces
- 2. Embed: Convert chunks to TF-IDF vectors (free, no API)
- 3. Store: Save vectors to JSON file or PostgreSQL
- 4. Search: Cosine similarity + keyword boost finds matches
- 5. Generate: Groq LLM answers using retrieved context
Ingest → Embed → Store → Query → Similarity Search → Context → LLM