Agentic AI Project in Python: ReAct Agent Course Guide

Agentic AI – Project Based Learning is a free video course in which you build a working agentic AI application in Python, one lab at a time. It is for Python learners and developers who understand the idea of AI agents but want to see how one is actually put together: configuration, logging, memory, tools, prompts, the ReAct loop and a command-line interface.

Instead of jumping straight into a framework, the course builds each part of the agent yourself. That makes it a good way to understand what agent frameworks do for you behind the scenes. The full list of videos is on the Agentic AI project course page.

Who this course is for

This is an Intermediate course. It suits:

• Python developers who want to build their first agent from scratch rather than copying a framework example.
• Students and career switchers preparing for AI engineering roles who need a project they can explain in an interview.
• Engineers who already use agent frameworks and want to understand the ReAct pattern and tool calling at a lower level.

You should be comfortable writing basic Python. The course uses an OpenAI API key and serper.dev for web search, and one lab shows how to set both up.

What you will learn

The 16 videos group into five themes.

1. Agentic AI foundations

01 Lesson - Theory - Agentic AI - Foundation And Principles, 02 Lesson - Theory - Agentic AI Architecture - Overview and 03 Lesson - Theory - ReAct Pattern Architecture Overview explain what makes an application “agentic” and how the ReAct (Reason and Act) pattern works. These three short lessons give you the mental model the rest of the project follows.

2. Project set-up

04 LAB - Prerequisite Agentic AI - ReAct Pattern - Setup OpenAI API And serper.dev, 05 LAB - Setting up Environment for Implementing Agentic AI Project and 06 LAB - Setting up Configurations Management prepare the project. Keeping keys and settings in a proper configuration layer from the start is a habit that pays off in every real project.

3. Core building blocks

07 LAB Implementing Logger For Agentic AI Project and 08 LAB - Implementing Memory System For Agentic AI Project add logging and memory. Logging lets you see what the agent decided at each step, and memory lets it keep track of the conversation.

4. Tools

09 LAB - Implementing Base Tool Contract For Agents, 10 LAB Implementing Calculator Tool For Agents and 11 LAB Implementing Web Search Tool For Agents define how tools work. A base tool contract means every tool looks the same to the agent, so adding a new tool later is simple.

5. Reasoning loop and interface

12 LAB Building System Prompts For Agents, 13 LAB How To Build - Reason & Act (ReAct) Pattern In Agentic Project and 14 LAB - How To - Build CLI Interface For Agentic AI Application bring everything together. The system prompt tells the agent how to behave, the ReAct loop lets it think and call tools, and the CLI lets you talk to it. 15 LAB Demo Agentic AI Application shows the finished result.

How the course is organised

• 16 videos, about 2.2 hours in total, free on YouTube.
• Videos are numbered 00 to 15: a 23-minute course overview (00), three short theory lessons (01 to 03), then twelve labs (04 to 15).
• Most labs are 4 to 9 minutes long; the ReAct loop lab is the longest at about 15 minutes.

One thing to know: the YouTube playlist lists the videos in reverse order, starting with video 15, the demo. Watching the demo first is a nice preview of where you are heading, but then go back and follow the numbers from 00 upwards. Find the playlist on YouTube.

How to study it step by step

1. Watch the demo (15) for a preview, then the course overview (00).
2. Watch the three theory lessons (01 to 03) in one sitting. Sketch the ReAct loop on paper: reason, choose a tool, act, observe, repeat.
3. Set up your keys and environment (04 to 06). Get your OpenAI API and serper.dev access working before you write agent code.
4. Code along with every lab. Pause each video and type the code yourself instead of only watching. Run it after each lab so problems show up early.
5. Build the logger and memory (07, 08) and check the log output as you test.
6. Add the tool contract and tools (09 to 11). Once they work, try adding one small tool of your own using the same contract.
7. Write the system prompt and ReAct loop (12, 13). Rewatch lab 13 if needed; it is the heart of the project.
8. Finish with the CLI (14) and compare your app with the demo in video 15.

This is a classic course, so there is no lab guide, practice simulation or mock exam on Skills Pragati for it. Your finished code is the proof that you learned it.

Where to go next

Once you can build an agent by hand, frameworks make much more sense. Building Production AI Agents with LangGraph takes the same ideas into a production framework, and AI Systems Engineering Beginner to Advanced widens the view to complete AI systems.

Frequently asked questions

Is the Agentic AI project course free?

Yes. All 16 videos are free on YouTube, and you can follow them from the Agentic AI – Project Based Learning course page.

How long does it take?

The videos total about 2.2 hours. If you code along and test as you go, plan for roughly two to three times that, spread over a week or two.

What do I need before I start?

Basic Python and a computer set up for development. The course uses an OpenAI API key and serper.dev for web search; lab 04 walks through setting both up.

Why is the playlist in reverse order?

The playlist starts with video 15, the demo of the finished app. Follow the video numbers from 00 to 15 to build the project in the right order.