RAG Technology: Principles, Use Cases & Implementation

Format 1: Half-day session – Duration: 3h30

Objectives:

• • Understand how RAG (Retrieval-Augmented Generation) works

• • Identify relevant use cases (FAQ, chatbot, business assistant)

• • Build a basic RAG system using accessible tools (OpenAI, LangChain, LlamaIndex)

•  

Program:

1. Introduction to RAG (45 min)

• • What is RAG?
  → Principle: document retrieval + response generation

• • Why use RAG? Overcoming LLM limitations (hallucinations, context boundaries)

• • Architecture: key components (indexing, retrieval, prompt injection)

•  

2. Real-Life Use Cases for HR (45 min)

• • Internal support chatbot (HR or legal knowledge base)

• • Smart search over business PDFs

• • Dynamic FAQ over a product corpus

• • Customer feedback or support ticket analysis with RAG

•  

3. Simple RAG Implementation (1h30)

• Example 1: RAG using OpenAI + PDFs (via LangChain or LlamaIndex)
  Steps:

• • Ingest and vectorize a document (e.g. HR base, product doc)

• • Embedding via OpenAI or HuggingFace

• • Vector storage (Chroma, FAISS…)

• • Query & response with an LLM

• 🛠️ Workshop: Building useful prompts for managers

•  

4. Key Considerations (30 min)

• • GDPR and data privacy

• • Response quality: re-ranking, chunking, corpus updates

• • Infrastructure and technical limits

•