LangChain Agentic AI

A hands-on crash course for building AI agents with LangChain — from basics to middleware, tools, structured output, and human-in-the-loop workflows.

---

Table of Contents

• Overview
• Topics Covered
• Project Structure
• Setup
• Environment Variables
• Notebooks
• Tech Stack
• Media
• License

---

Overview

This repository is a structured learning path for LangChain's agentic AI capabilities. Each notebook walks through a core concept with runnable examples — from sending your first message to building agents with human-in-the-loop safeguards.

Whether you're new to LangChain or looking to understand middleware, tools, and structured outputs, this repo gives you copy-paste-ready code with explanations.

---

Topics Covered

Topic	Description
Model Integration	Connect to Groq, Google Gemini, and other LLM providers
Messages	System, Human, and AI message types and conversation flow
Tools	Define and bind custom tools for agent function calling
Structured Output	Get typed, schema-validated responses from LLMs
Middleware	Summarization, token management, and execution control
Human in the Loop	Pause agent execution for human approval, editing, or rejection

---

Project Structure

langchain-agentic-ai/
├── main.py
├── pyproject.toml
├── requirements.txt
├── README.md
└── updatedlangchain/
    ├── langchainintro.ipynb        # LangChain basics
    ├── messages.ipynb              # Message types & roles
    ├── modelIntegration.ipynb      # LLM provider setup
    ├── structuredoutput.ipynb      # Schema-validated outputs
    ├── tools.ipynb                 # Tool binding & calling
    ├── middleware.ipynb             # Summarization middleware
    ├── api-integration-guide.md    # API setup reference
    ├── structured-output-notes.md  # Structured output cheat sheet
    ├── tools-notes.md              # Tools cheat sheet
    └── human-in-the-loop-notes.md  # HITL explanation

---

Setup

# Clone the repo
git clone https://github.com/<your-username>/langchain-agentic-ai.git
cd langchain-agentic-ai

# Install dependencies
uv sync

# Add new dependencies
uv add <package-name>

---

Environment Variables

Create a .env file in the project root:

GROQ_API_KEY=your_groq_api_key_here
GOOGLE_API_KEY=your_google_api_key_here

Note: Use underscores (_) in env variable names — Python's os.getenv() can't read hyphens.

---

Notebooks

1. LangChain Intro

Getting started with LangChain — model invocation, prompts, and chains.

2. Messages

Understanding SystemMessage, HumanMessage, and AIMessage — how conversation context flows through the agent.

3. Model Integration

Setting up LLM providers:

• Groq — groq:llama-3.3-70b-versatile, groq:qwen/qwen3-32b
• Google Gemini — google_genai:gemini-2.0-flash

Always use the provider prefix (e.g., groq:, google_genai:). Auto-detection fails for non-OpenAI models.

4. Tools

Define custom functions with @tool decorator and bind them to agents for function calling.

5. Structured Output

Get typed Pydantic models back from LLMs instead of raw text.

6. Middleware — Summarization

Automatically compress long conversation histories to reduce token usage:

• Message-based trigger — summarize after N messages
• Token-based trigger — summarize after N tokens
• Fraction-based trigger — summarize after X% of context window

7. Middleware — Human in the Loop

Pause agent execution before sensitive operations:

• Approve — let the agent proceed
• Reject — block the action
• Edit — modify arguments before execution

---

Tech Stack

Component	Technology
Framework	LangChain 1.3+
LLM Providers	Groq, Google Gemini
Agent Runtime	LangGraph
Language	Python 3.13+
Package Manager	uv

---

Media

Screenshots

Coming soon

Demo Video

Coming soon

---

License

This project is licensed under the MIT License.

---

Built with LangChain