Module 1: Foundations of Agentic AI (Sessions 1–4 • 4 hrs)

Session 1: What is Agentic AI?

• Generative vs. Agentic AI
• Workflow vs. AI agent
• The ReAct framework
• Types of agents
• Agentic architecture
• Live demos of real agents

Session 2: How LLMs Actually Work

• Tokens
• Context windows
• Hallucinations — intuition, no math
• What a prompt really is: composition of a formal prompt

Session 3: Prompt Engineering for Agents

• Zero-shot
• Few-shot
• Chain-of-thought
• ReAct prompting
• Context engineering
• System prompts

Session 4: APIs & Workspace Setup

• What an API key is
• Getting your OpenAI/Gemini key
• Setting up your n8n workspace
• No code fear

✓ Checkpoint: Short quiz on foundations.

Module 2: Building Agents Without Code: n8n (Sessions 5–14 • 10 hrs)

Session 5: n8n Fundamentals

• Structure of an n8n workflow
• n8n vs. Zapier vs. Make
• Your first data pipeline: form to CRM

Session 6: Automation Pipelines

• Automating lead generation
• Google Form to CRM
• Triggers, nodes, and data flow

Session 7: Routing & Branching Logic

• Automated social media pipeline with advanced routing and branching

Session 8: AI Support Triage Agent

• Your first AI agent
• LLM + prompts + tools
• Incoming emails classified, routed, and logged to a database

Session 9: Invoice Data Extraction Agent

• Structured outputs
• Forcing strict JSON schemas so AI returns clean, reliable data

Session 10: Agent Tools & Memory

• A recommendations agent with tools, memory, and real-world architecture
• Why agents forget

Session 11: RAG — The Concept

• Knowledge bases
• Vector embeddings
• Chunking
• Semantic search — explained visually

Session 12: RAG — The Build

• Your first RAG workflow
• From knowledge base to conversational AI

Session 13: Telegram RAG Assistant

• A conversational Telegram bot with advanced retrieval
• A real AI assistant on your phone

Session 14: Mini-Project #1

• Your Personal AI Representative
• A RAG agent over YOUR own resume or business docs
• Shared as a working link

✓ Checkpoint: Short quiz, plus every participant shares their working agent in the cohort group.

Module 3: Just Enough Python (Sessions 15–20 • 6 hrs)

Session 15: Python Basics I

• Variables
• Strings
• Lists
• Dictionaries

Session 16: Python Basics II

• Functions
• Loops
• Working with JSON

Session 17: Your First LLM Call in Code

• Calling the LLM API directly from Python
• The same brain n8n used, now under the hood

Session 18: Function Calling Demystified

• How LLMs really use tools
• Manual JSON tool schemas
• The function-calling mechanism

Session 19: Build an Agent From Scratch I

• The ReAct agent loop with the raw API
• No framework
• Full understanding

Session 20: Build an Agent From Scratch II

• The ReAct prompt
• Building an AI agent even without function calling
• The “aha” moment of the course

✓ Checkpoint: Short quiz on Python and tool calling. Everything runs in Google Colab — no local Python installation.

Module 4: Agent Engineering with LangChain (Sessions 21–24 • 4 hrs)

Session 21: LangChain Essentials

• Prompt templates
• Chat models
• Chains
• Build a text-summarizer chain
• Debugging and tracing

Session 22: LangChain Agents

• Your first LangChain agent with tools and LLMs
• Real-world web search with Tavily
• Structured outputs with Pydantic

Session 23: RAG with LangChain

• Document loaders
• Text splitters
• Embeddings
• A vector database
• A complete RAG pipeline in code
• Mini-Project #2 begins

Session 24: MCP — Model Context Protocol

• Why MCP exists
• MCP architecture and servers
• Using a pre-built MCP server with AI clients
• Mini-Project #2 due & shared

Module 5: Multi-Agent Systems & Production Reality (Sessions 25–28 • 4 hrs)

Session 25: Multi-Agent Systems

• A multi-agent marketing team
• Supervisor & worker architecture
• Sequential agents and context handoff

Session 26: Error Handling & Self-Healing Pipelines

• Centralized error handling in n8n
• Building workflows that recover from failure
• Production-grade reliability

Session 27: Agent Security & Safe AI

• LLM application security
• Common vulnerabilities
• Guardrails and human-in-the-loop checkpoints

Session 28: LLM Apps in Production & Capstone Prep

• Privacy and data retention
• Production architecture insights
• Mapping agent patterns to your capstone

Module 6: Capstone Project (Sessions 29–35 • 7 hrs)

Session 29: Capstone Kickoff

• Choose your track and project
• Scoping & success criteria
• 4–6 hours of build time over two weeks

Sessions 30–33: Guided Build Sessions

• Hands-on building with live instructor support

Session 34: Polish & Rehearse

• Debugging clinic
• Demo preparation

Session 35: Demo Day 🎉

• Capstone presentations
• Working links shared
• Certificates
• Next steps