Description

Introduction

Agentic AI unlocks the full potential of LLMs by enabling goal-driven, autonomous behavior across tasks and domains. This course focuses on building intelligent workflows using two of the most powerful agentic frameworks—LangChain and AutoGen. Learners will explore how to compose agents, manage tools, implement memory, and coordinate autonomous multi-step behaviors that go beyond single prompts.

Prerequisites

Participants should have:

• Working knowledge of Python programming

• Experience using LLM APIs (e.g., OpenAI, Anthropic, or Azure OpenAI)

• Familiarity with basic LangChain or prompt engineering (helpful but not mandatory)

• Prior exposure to JSON, REST APIs, or vector stores is advantageous

Table of Contents

1. Introduction to Agentic Workflows

• 1.1 Agentic vs Reactive LLM Applications

• 1.2 When to Use Agentic Patterns

• 1.3 Key Concepts: Tools, Memory, Planning, Delegation

2. Core Concepts in LangChain Agents

• 2.1 Chains vs Agents: What’s the Difference?

• 2.2 Tool Invocation and Custom Tooling

• 2.3 Memory: Conversational, Buffer, Summary

• 2.4 Planners and Executors (Multi-step workflows)

3. Building Autonomous Agents with LangChain

• 3.1 ReAct and Conversational Agent Patterns

• 3.2 Toolkit Integration: Search, Code, File I/O, APIs

• 3.3 LangChain Expression Language (LCEL)

• 3.4 Building Multi-Agent Systems in LangChain

4. Constructing Agentic Workflows with AutoGen

• 4.1 AutoGen Fundamentals: Agents and GroupChat

• 4.2 UserProxy vs AssistantAgent Roles

• 4.3 Task-Oriented Dialogue & Recursive Planning

• 4.4 Multi-Agent Collaboration with AutoGen

5. Comparing LangChain and AutoGen

• 5.1 Design Philosophies: Modular vs Dialogue-Based

• 5.2 Use Case Alignment: When to Choose Which

• 5.3 Combining Both in a Unified Workflow

6. Hands-on Projects

• 6.1 LangChain: Report Generator with Search + Summarizer Agents

• 6.2 AutoGen: Multi-Agent Research Assistant

• 6.3 LangChain + AutoGen: Hybrid Workflow for Product Planning

7. Optimization and Guardrails

• 7.1 Cost Control and Prompt Efficiency

• 7.2 Error Recovery and Tool Misuse Handling

• 7.3 Sandboxing and Security Controls

8. Deployment and Integration

• 8.1 API Wrapping and Backend Integration

• 8.2 Dockerization and Scalable Hosting

• 8.3 Logging, Monitoring, and Evaluation

LangChain and AutoGen empower developers to shift from static prompts to intelligent, adaptive AI workflows. By mastering these frameworks, you can create agents that think, plan, and act—alone or in teams—across real-world scenarios. This course equips you with the tools and strategies needed to implement scalable, robust agentic systems at the cutting edge of applied AI.