This course is based on the attached Unit 5: RAG & AI Search presentation, covering Modules 13–15: RAG Fundamentals, Context Injection & Prompt Templates, and AI Search Optimization.

🎯 Unit Learning Objectives

By the end of this unit, learners will be able to:

Explain Retrieval-Augmented Generation (RAG).
Understand how a RAG system retrieves and uses external information.
Explain embeddings and vector databases.
Differentiate keyword, semantic, and hybrid search.
Understand context injection and its different types.
Create structured prompt templates.
Explain AI Search Optimization.
Understand vector search and AI ranking systems.
Identify practical applications of RAG and AI search.

Module 13: RAG Fundamentals

13.1 What is RAG?

RAG stands for Retrieval-Augmented Generation.

RAG is an AI technique in which an AI system first retrieves relevant information from external sources and then uses that information to generate a more accurate and context-aware response.

RAG has three stages:

R — Retrieval
The AI searches external sources such as databases, documents, websites, PDFs, knowledge bases, and vector databases.

A — Augmented
The retrieved information is added to the AI's context.

G — Generation
The LLM uses the retrieved information together with its existing knowledge to generate the final answer.

Visual: RAG Architecture

Simple Flow:

User Question → Retrieve Information → Add Context → LLM → Answer

13.2 Why is RAG Important?

An LLM working without external retrieval can face problems such as:

Outdated information
Incorrect or fabricated answers
Lack of company-specific knowledge
Poor performance on niche topics
Lack of access to recent information

RAG helps provide more accurate, grounded and domain-specific responses while improving reliability.

Example

Imagine an employee asks:

"What is our company's current leave policy?"

A normal AI model may not know the company's latest policy.

A RAG system can retrieve the company's HR policy document and use the relevant section to answer the question.

13.3 How RAG Works

A RAG workflow generally follows four steps.

Step 1 — User Query

The user submits a question.

Example:

"What are the latest AI trends in healthcare?"

Step 2 — Information Retrieval

The system searches relevant sources such as research papers, databases, documents, or websites.

Step 3 — Context Injection

Relevant information is inserted into the prompt.

Step 4 — Response Generation

The LLM uses the enriched prompt to generate the response.

13.4 Architecture of a RAG System

A RAG system consists of several components:

User Interface

Where the user submits a query.

Retriever

Searches knowledge sources and identifies relevant information.

Vector Database

Stores document embeddings and supports semantic retrieval.

Examples include:

Pinecone
Weaviate
ChromaDB

Embedding Model

Converts text into numerical vectors.

LLM / Generator

Uses the retrieved information to generate the final answer.

Visual

13.5 Embeddings and Vector Databases

An embedding is a numerical representation of text that captures semantic meaning.

This allows AI systems to recognize that different words can have similar meanings.

Examples

Dog ↔ Puppy

Car ↔ Automobile

Even though the words are different, they represent related concepts.

Popular Vector Databases

Vector Database	Description
Pinecone	Managed cloud vector database
Weaviate	Open-source database with ML integrations
ChromaDB	Lightweight option for prototyping

13.6 Types of Retrieval

Keyword Search

Searches for exact words.

Example:
Searching for "AI course" looks for documents containing those words.

Limitation: It may miss related terms such as "artificial intelligence training."

Semantic Search

Understands the meaning and intent of the query.

It can recognize related phrases and concepts.

Hybrid Search

Combines:

Keyword Search + Semantic Search

This can provide better accuracy and relevance.

Visual: Semantic & Hybrid Search

13.7 Applications of RAG

🎓 Education

AI tutors
Personalized learning
Smart study assistants
Textbook Q&A
Exam preparation

🏢 Business

Customer-support chatbots
Internal knowledge assistants
HR policy Q&A
Automated report generation

🏥 Healthcare

Medical information retrieval
Drug interaction lookup
Clinical decision support
Patient history analysis

⚖️ Legal

Contract analysis
Legal document search
Case law research
Compliance checking

13.8 Advantages and Challenges

✅ Advantages

Updated information
Reduced hallucinations
Domain-specific knowledge
Personalized responses
Better factual accuracy
Enterprise scalability

⚠️ Challenges

Complex architecture
Requires high-quality databases
Retrieval quality affects answer quality
Higher computational requirements
Additional retrieval latency
Documents need to remain updated

13.9 Future of RAG

Future RAG systems may include:

Faster retrieval
Multimodal RAG
Real-time enterprise AI
Personalized AI
Agentic RAG
IoT and edge-computing integration

Module 14: Context Injection & Prompt Templates

14.1 What is Context Injection?

Context Injection means adding additional information to a prompt so that AI can generate a more accurate and relevant response.

Without Context

"Explain the policy."

The AI does not know which policy is being discussed.

With Context

"Using the company HR policy document, explain the leave policy for employees."

The AI now has specific information to work with.

Visual: Context Injection

14.2 How Context Injection Works

Step 1 — User Query

The user asks a question.

Step 2 — Retrieve Context

The system collects relevant information.

Step 3 — Insert Context

The system creates:

User Query + Retrieved Information + System Instructions

Step 4 — Generate Response

The LLM processes the enriched prompt and generates the answer.

14.3 Types of Context

1. Static Context

Information that remains fixed.

Examples:

Company policies
System instructions
Tone guidelines
Formatting rules

2. Dynamic Context

Information that changes according to the query.

Examples:

Retrieved documents
Product information
Real-time data
API responses
User preferences

3. Conversational Context

Information from previous messages in the conversation.

Examples:

Earlier questions
Previous responses
Clarifications
Conversation history

14.4 Prompt Templates

A prompt template provides a structured format for creating consistent AI instructions.

A good prompt template can include:

Role

Defines who the AI should act as.

"You are a professional teacher."

Task

Defines what the AI should do.

"Explain photosynthesis clearly."

Context

Provides relevant background information.

Output Format

Defines how the response should be presented.

"Use bullet points and provide a brief summary."

Example Prompt Template

You are an expert AI tutor.
Explain the topic: {topic}
Use simple language suitable for beginners.
Provide 2–3 real-world examples.
End with a concise summary in 3 bullet points.

The variable {topic} can be replaced with different topics, making the template reusable.

14.5 Best Practices

For effective context injection:

Use relevant information.
Use high-quality context.
Keep prompts structured.
Keep prompts concise.
Update context regularly.

Common Challenges

Too much context can confuse AI.
Long prompts can increase token costs.
Poor retrieval can produce poor answers.
Context windows have limits.
Outdated context can produce outdated responses.

Module 15: AI Search Optimization

15.1 What is AI Search Optimization?

AI Search Optimization is the process of improving how AI systems search, retrieve, and rank information.

Its major goals are:

🎯 Improve Accuracy

Return results that actually answer the user's needs.

⚡ Faster Retrieval

Find relevant information quickly.

😊 Better User Experience

Make search more natural and intuitive.