# Your own data (RAG) for your AI agent, part one - step by step with Python and PostgreSQL

From unreliable memory through full-text and embeddings to hybrid search with re-ranking.

## Key points
- The problem: answering from memory is not always correct
- A better option: look up the supporting material first
- RAG in one sentence
- Today's journey: from simple search to re-ranking
- Notebook and database basics
- Extensions we will use
- Full-text: useful for exact words
- Embeddings: meaning instead of exact words
- Semantic search and its weaknesses
- Hybrid search: the best of both worlds
- Re-ranking: a second pass over relevance
- What we solved and what remains
- PostgreSQL for learning, Azure AI Search for practice