Welcome to AI agentic patterns

Welcome to AI Agentic Patterns Workshop! I'm Param, and in this course we'll master every major design pattern for building AI agents, from simple prompt chains all the way to reflexive metacognition.

This course covers a comprehensive catalog of agent design patterns. Each pattern is self-contained with real Python code using litellm, so the same code works with OpenAI, Anthropic, Ollama, Gemini, or any other provider. You'll learn patterns for reasoning, planning, tool use, multi-agent systems, RAG, safety, and much more.

The agent pattern journey

Your learning path from foundations to frontier patterns.

Setup

bash

# Clone the patterns repository
git clone https://github.com/learnwithparam/ai-agentic-patterns.git ai-agentic-patterns
cd ai-agentic-patterns

# Create venv and install dependencies
make setup

# Copy env template and add your API key
cp env.example .env
# Edit .env to set LITELLM_MODEL (e.g., ollama/llama3.2, gemini/gemini-1.5-flash, openai/gpt-4)

# Activate and test that everything works
source venv/bin/activate  # On Windows: venv\Scripts\activate
python patterns/01_prompt_chaining.py

Clone the repository and install dependencies to follow along with every pattern.

Validation checklist: Environment setup checklist

Loading practice…

Any provider works. LiteLLM abstracts the provider layer, so you can use OpenAI, Anthropic, Gemini, Ollama (free and local), or any other supported provider by changing a single environment variable. No code changes required.

Great job setting up your environment! You are ready to build AI agents. We will start with the two most fundamental patterns: prompt chaining and routing. These are the foundation that every other pattern builds on.