Welcome to Agentic AI Practice, a curated playground for building autonomous, goal-driven AI agents using the latest agentic frameworks. This repository brings together powerful tools like LangChain, LangGraph, Langsmith, CrewAI, Phidata, Agno, and Model Context Protocol (MCP) to explore the future of intelligent systems.
Agentic AI refers to artificial intelligence systems designed with:
- Autonomous Decision-Making
- Goal-Oriented Behavior
- Dynamic Interaction with environments and other agents
This repo demonstrates how agentic systems can be orchestrated across multiple frameworks to reason, plan, and act with minimal human intervention.
| Framework / Tool | Purpose & Highlights |
|---|---|
| LangChain | LLM orchestration, memory, and tool use |
| LangGraph | Graph-based agent workflows and stateful execution |
| LangSmith | Tracing, debugging, and observability for agentic flows |
| CrewAI | Multi-agent coordination with role-based tasking |
| Phidata | Declarative agent framework with modular app design |
| Agno | Lightweight agent framework for fast prototyping |
| Model Context Protocol (MCP) | Standardized context exchange between agents and models |
| Pydantic | Data validation and structured modeling |
| Mermaid | Graph visualization of agent workflows and dependencies |
| Playground | Interactive environment for testing and refining agent prompts and workflows |
Agentic-AI-Learning/
├── Agno_and_Phidata_Apps/ # Apps built with Agno and Phidata frameworks
├── Crew_AI_Apps/ # Role-based agents using CrewAI
├── LangChain_LangGraph_LangSmith_Apps/ # LangChain + LangGraph + LangSmith agent workflows
├── MCP_Server/ # Model Context Protocol server implementation
├── .vscode/ # Editor settings and workspace configs-
Agent orchestration across frameworks
-
Role-based agent design (e.g., legal agent, marketing agent)
-
Context sharing via MCP
-
Graph-based reasoning with LangGraph
-
Declarative agent apps with Phidata
-
Visualizing agent flows using Mermaid
Explore foundational tools and frameworks for building agentic AI systems:
-
LangChain Documentation
Build applications powered by language models using chains, agents, and tools. -
LangGraph Documentation
Orchestrate multi-agent workflows with graph-based reasoning and stateful execution. -
LangSmith Documentation
Trace, evaluate, and debug LLM applications with powerful observability tools. -
CrewAI Documentation
Design collaborative multi-agent systems with roles, memory, and tool usage. -
Phidata Documentation
Create structured, declarative agents with built-in memory and reasoning. -
Agno Documentation
Build secure, high-performance agentic apps using AgentOS and runtime orchestration. -
Pydantic Documentation
Validate and serialize data using Python type hints with speed and clarity. -
Model Context Protocol Setup Guide
Learn how to build MCP-compatible servers for agent communication and context sharing.
Clone the repo and explore each folder to see how different frameworks are used to build agentic systems:
git clone https://github.com/Kratugautam99/Agentic-AI-Learning.git
cd Agentic-AI-LearningEach subfolder contains its config files to guide you through setup and execution.
These below are from Langchain Academy Official Courses.
- LangGraph Foundation Certification => https://academy.langchain.com/certificates/z4rtkxd8po
- LangSmith Foundation Certification => https://academy.langchain.com/certificates/sey9wsnx9e
This project is open-source under the MIT License. See LICENSE for details.