Build AI Apps with Python: Why RAG? — Give AI Your Own Data | Episode 12

Claude is smart, but it does not know YOUR data. Ask it about your company handbook and it either guesses or admits it does not know. RAG fixes this — retrieve your documents and inject them into the prompt. In this episode, we ask the same question twice. Without RAG, Claude hallucinates. With RAG, Claude answers accurately from a fictional company handbook. Same model, same question — context changes everything. Phase 3 begins! Student code: https://github.com/GoCelesteAI/build-ai-apps-python/tree/main/episode12 Every keystroke is shown on screen with 3-second pauses so you can follow along at your own pace. What You'll Learn: • What RAG is — Retrieve, Augment, Generate • Why Claude hallucinates on unknown data • Injecting documents into the user message with f-strings • The "answer based only on the document" system prompt • Side-by-side comparison: hallucination vs accurate answer • When to use RAG vs fine-tuning • The foundation for document Q&A applications • Running Python scripts with :!python % Timestamps: 0:00 - Introduction 0:12 - Why RAG? (Preview) 0:46 - Creating why_rag.py 1:02 - Imports and setup 1:45 - The fictional company handbook 5:25 - Save handbook 5:35 - The question: annual leave days 5:50 - Without RAG — no context, Claude guesses 7:28 - With RAG — inject the handbook 10:32 - Save and run 10:50 - Without RAG: Claude admits it doesn't know 11:00 - With RAG: "18 days" — exact answer! 11:30 - Code review 11:55 - Recap: 3 Key Takeaways 12:27 - End Screen Key Takeaways: 1. Claude hallucinates on data it has never seen — it guesses or admits it does not know 2. RAG injects relevant documents into the prompt — same API, accurate answers 3. The system prompt says "answer from the document only" — this prevents mixing in guesses This is Episode 12 of Build AI Apps with Python in Neovim — starting Phase 3 (RAG: Retrieval Augmented Generation).