Topics
On this page
Generative AI & Agentic AI
Generative AI is a branch of artificial intelligence that creates new content such as text, images and code. Agentic AI builds on it: systems that plan steps, use tools and complete tasks towards a goal. This tutorial covers both, from how large language models work to prompt engineering, RAG, AI search, AI agents, the Model Context Protocol (MCP) and agent frameworks.
It is suitable for students, freshers and working professionals. The concept lessons need no maths or coding background, and the Python examples need only basic Python. Each lesson includes a diagram, an example, a short quiz and frequently asked questions.
1. Introduction to Generative AI
- What is Generative AIWhat it means for a machine to create text, images and code.
- How Generative AI WorksTraining, patterns and prediction, explained without maths.
- Generative AI vs Traditional AICreating new content versus sorting and predicting.
- AI vs Machine Learning vs Deep Learning vs Generative AIHow the four terms fit inside each other.
- Types of Generative AI ModelsTransformers, diffusion models, GANs and VAEs compared.
- Top Applications of Generative AI in IndiaWhere Indian companies already use it, from banking to retail.
2. Large Language Models (LLM)
- What is a Large Language Model (LLM)The model behind ChatGPT, Claude and Gemini, in plain English.
- How LLMs WorkFrom your prompt to the next word, step by step.
- Transformer Architecture ExplainedAttention and why it changed language AI.
- Tokens and Tokenization in LLMWhy models read pieces of words, and why it affects cost.
- Context Window in LLMHow much a model can read at once, and what happens past it.
- Embeddings in LLMTurning meaning into numbers a computer can compare.
- Temperature and Top-p in LLMThe two settings that make answers creative or predictable.
- LLM Hallucination: Causes and FixesWhy models state wrong facts confidently, and how to reduce it.
- Open Source vs Closed Source LLMsLlama and Mistral versus Claude and GPT: cost, control and privacy.
- Popular LLMs (Claude, GPT, Gemini, Llama) ComparedStrengths of the major models side by side.
- Fine-Tuning LLMsTeaching a trained model your own task or style.
- Small Language Models (SLM)Compact models that run cheaply, even on a laptop or phone.
3. Prompt Engineering
- What is Prompt EngineeringWriting instructions that get reliable answers from a model.
- Types of Prompting (Zero-shot, Few-shot)When to give examples and when not to.
- Chain of Thought PromptingAsking the model to reason step by step.
- System Prompt vs User PromptWho sets the rules and who asks the question.
- Structured Output (JSON) from LLMsGetting data your code can use, not just prose.
- What is Context EngineeringChoosing what the model sees, beyond the prompt itself.
4. Retrieval-Augmented Generation (RAG)
- What is RAG (Retrieval-Augmented Generation)Letting a model answer from your own documents.
- How RAG WorksIndexing, retrieval and generation in one flow.
- RAG vs Fine-TuningWhich to choose for your data, budget and use case.
- Chunking Strategies in RAGSplitting documents so the right piece gets found.
- Vector Database in RAGWhere embeddings live and how they are searched.
- Reranking in RAGA second pass that puts the best passage first.
- Types of RAG (Naive, Advanced, Graph RAG, Agentic RAG)The main RAG designs and when each one fits.
- RAG Evaluation MetricsMeasuring whether answers are grounded and correct.
- Build a RAG Chatbot in PythonA working chatbot over your own documents.
5. AI Search
- What is Semantic SearchSearch that understands meaning, not just matching words.
- Keyword Search vs Semantic SearchExact words versus intended meaning.
- Vector Search ExplainedFinding the nearest meaning in a space of numbers.
- Hybrid Search (BM25 + Vector)Combining keyword and meaning for better results.
- How AI Search Engines Work (Perplexity, AI Overviews)Search, read and summarise, all in one answer.
- What is GEO (Generative Engine Optimization)Getting your content cited by AI answers.
- SEO vs GEO vs AEOThree ways to be found, and what each one needs.
6. AI Agents
- What is an AI AgentSoftware that decides its next step and uses tools.
- Types of AI AgentsFrom simple reflex agents to learning agents.
- AI Agent vs Chatbot vs AI AssistantThree words people mix up, told apart.
- AI Agent ArchitectureThe parts every agent has: model, tools, memory, loop.
- Tool Calling (Function Calling) in LLMHow a model asks your code to do something.
- Memory in AI AgentsShort-term and long-term memory, and why agents need both.
- ReAct Agent ExplainedReason, act, observe, repeat.
- Planning and Reasoning in AI AgentsBreaking a goal into steps before acting.
7. Agentic AI
- What is Agentic AIAI systems that plan and act towards a goal on their own.
- Agentic AI vs Generative AICreating content versus getting work done.
- Agentic AI vs AI AgentsThe approach versus the individual worker.
- Agentic Design PatternsReflection, tool use, planning and multi-agent patterns.
- Multi-Agent SystemsSeveral agents with different jobs working together.
- Agentic RAGAn agent that decides what to search and when.
- Human-in-the-Loop in Agentic AIWhere a person approves, corrects or stops the agent.
- AI Agent EvaluationTesting an agent that can take many paths.
- AI Agent Security and Prompt InjectionHow attackers hijack agents, and how to defend them.
- Top Agentic AI Use Cases in IndiaReal workflows agents are taking on in Indian businesses.
8. Model Context Protocol (MCP)
- What is MCP (Model Context Protocol)One standard way to connect models to tools and data.
- MCP Architecture: Host, Client, ServerThe three roles and how they talk.
- MCP vs API vs Function CallingHow MCP relates to what you already know.
- Build an MCP Server in PythonYour first MCP server, step by step.
- A2A Protocol vs MCPAgent-to-agent talk versus agent-to-tool talk.
9. Agent Frameworks & Tools
- Claude Agent SDK TutorialBuild an agent with Anthropic's own SDK.
- LangChain TutorialChains, tools and retrievers in LangChain.
- LangGraph TutorialAgents as graphs with state and control.
- CrewAI TutorialA team of role-based agents working together.
- n8n AI Agent TutorialBuild an AI agent without writing code.
- Claude Code TutorialAn AI coding agent in your terminal.
- LangGraph vs CrewAI vs AutoGenWhich agent framework to pick for your project.
Frequently Asked Questions
Do I need to know coding to follow it?
No. The concept lessons are written in plain English and need no code. Some lessons include short Python examples; basic Python helps there, and each example explains what every part does.
Who is this tutorial for?
Students, freshers and working professionals in India who want to understand how tools like ChatGPT, Claude and Gemini work, and how AI agents are built on top of them. It is also useful revision before interviews for AI roles.
In what order should I read the lessons?
Start with Introduction to Generative AI and read the sections in order. Each lesson builds on the ones before it, and the Previous and Next buttons follow that order.
What is the difference between Generative AI and Agentic AI?
Generative AI creates content such as text, images or code when you ask for it. Agentic AI uses those models to plan and carry out tasks towards a goal, calling tools and checking results along the way.