Agentic AI Foundation Course

• Concept of Retrieval Augmented Generation (RAG) with 2 real-life examples
• Data pipeline for RAG
• Demo/ Code walkthrough- Naïve RAG
• Breakout Room Activity- Understanding code/ flow of an E2E RAG pipeline
• Business use cases of RAG Apps across Industries and Horizontals/ Functions, including Vertical AI Agents
• Chunking Strategies
• Creating Vector embeddings and storing them in a Vector database, with examples
• Retrieval Mechanics & Similarity/ Vector search techniques
• Langchain for RAG Development
• Techniques for improving RAG accuracy- Reranking, chunk optimization, others
• Langflow for low-code, no-code visual modeling-based RAG development

Lab: Building RAG Data ingestion, indexing, retrieval, and generation pipeline

Lab: FileSearch Tool for Managed RAG- Google Gemini

Lab: Data ingestion & indexing – Multiple files

• Overview of AI Agents/ Agentic AI & role of LLMs/ SLMs
• Design patterns of AI Agentic workflows- Reflection, Tool use/ function calling, Planning, Multi-agent collaboration
• Business use cases of AI Agents across Industries and Horizontals/ Functions, including Vertical AI Agents
• Technical use cases of AI Agents (in SDLC automation)
• Step-by-step – Designing an AI Agent
• Advanced Context Engineering for Agents
• Agentic Memory – Long-term, short-term, with demos

Lab: Agentic Memory

• Agents controlling the browser, keystrokes, and mouse clicks/ computer control
• Deep Agents- working with complex, long-running tasks
• Building a browser-based agent for Software Testing
• AI Agent/ Agentic AI Pricing models & how they are disruptive to existing SAAS models
• Breakout Room Activity: Building use case and specs for an AI Agent

Demo- AI Agent for Web UI Testing

• Tool use/ function calling with API integrations

Demo – Tool use/ function calling

• Model Context protocol (MCP)
• Connecting Agents with 3rd party apps using MCP Servers
• MCP Host and MCP Client

Demo- Setting up MCP Servers on Cline and GitHub Copilot in VSCode

Demos- Working with MCP Servers from Zapier, Composio, AWS/ Azure/ GCP

• MCP Server Security
• Getting website data using MCP-B and llms.txt
• Real-time web data access with tools like Tavily, Serper, and FireCrawl

Demo- Building your own MCP Server and hosting on the Cloud

Lab: Use GitHub Copilot Agent mode & connect with GitHub MCP Server & Playwright MCP Server for specific tasks

Lab: Build an AI Agent with MCP Server Integration

• TOON – Token object-oriented notation

Lab: Working with TOON