Agentic AI Course · Setup Guide

The software tools you'll need to
build agentic AI.

A complete, beginner-friendly checklist of everything you need installed and set up to follow the Agentic AI course. Almost all of it is free. Work through it once, and you're ready to build real agents.

🟢 Good news: you can start with almost nothing paid
You need three things at minimum: a place to write code, access to an AI model, and a couple of free libraries. Everything below marks what's free, what's paid, and what we recommend for beginners. You don't need all of it on day one — the course tells you exactly when each tool appears.

1Your coding environment

Where you'll actually write and run your agent code. Pick the easiest option — you can change later.
Google ColabFreeRecommended for beginners
Runs Python in your browser — nothing to install. This is what the course uses in its "do-it-for-real" boxes. Just open colab.research.google.com and start typing. The fastest way to begin.
Python 3.10+Free
The programming language of AI. If you'd rather work on your own machine than in the browser, install it from python.org. (Colab already has Python built in, so you can skip this at first.)
VS CodeFree
A free code editor from Microsoft for when you move off Colab to build bigger projects on your computer. Add the Python extension and you're set. Optional early on.
Git & GitHubFree
To save your work and build a portfolio you can show in interviews. Create a free account at github.com. The course encourages putting your first agent project here.

2Access to an AI model (the "brain")

Your agent needs a large language model to think. You need at least one of these — start with the free chat, add an API key when you begin coding agents.
ClaudeFree tierRecommended
Anthropic's AI, at claude.ai. Use the free chat for the early lessons and the DRONA tutor. For building agents in code, get an API key from console.anthropic.com (pay-as-you-go, very low cost for learning).
ChatGPT / OpenAI APIFree tier
OpenAI's AI at chatgpt.com, with API keys from platform.openai.com. A great alternative or second option. Most course code works with either Claude or OpenAI.
Free / open models (optional)Free
Tools like Ollama (ollama.com) let you run smaller models on your own computer at no cost — useful for practice without API charges. Optional and slightly more advanced.
💳 About API costs
API access is pay-as-you-go and usually costs only a few rupees while learning — you pay for what you use. Set a low spending limit in your provider's dashboard so there are no surprises. The free chat versions cost nothing at all for the early lessons.

3Agent libraries & frameworks

The building blocks that turn a model into an agent. You install these with one command (pip install ...). The course teaches from scratch first, then these frameworks.
The provider SDKFreeStart here
The official library for your model: anthropic for Claude or openai for OpenAI. This alone is enough to build your first agents from scratch, exactly as the course does in its early phases.
LangChain & LangGraphFree (open source)
The most popular agent frameworks. LangChain gives ready-made tools, memory, and integrations; LangGraph builds controllable agents as state graphs. Covered in the course's frameworks phase. Install with pip install langchain langgraph.
CrewAIFree (open source)
The easiest way to build multi-agent "crews" — teams of role-based agents that collaborate. Great for the multi-agent phase. pip install crewai.
Claude Agent SDKFree (open source)
Anthropic's official, production-grade toolkit for building agents on Claude — tool use, memory, subagents, and MCP support. For when you're ready to build serious agents.

4Memory, embeddings & RAG tools

For agents that remember and can answer from your own documents. Used in the memory and RAG phases of the course.
sentence-transformersFree
Turns text into "embeddings" (meaning as numbers) right on your machine, no API needed. The course uses this for semantic memory and RAG. pip install sentence-transformers.
ChromaFree (open source)Beginner-friendly
A simple vector database that stores embeddings so your agent can search by meaning. Runs locally with no setup fuss. pip install chromadb. Perfect for learning RAG.
Pinecone / Weaviate (optional)Free tier + paid
Cloud vector databases for larger, production RAG systems. Both have free tiers to experiment with. You won't need these to learn — Chroma is enough — but it's good to know they exist for real deployments.

5Tools your agents will use

Agents act by calling tools. These are common ones you'll wire up in the tools & function-calling phase. Most have free tiers.

✓The minimum to start today

If you only set up these four things, you can begin the course right now — all free:
1. EnvironmentOpen Google Colab in your browser — nothing to install.
2. AI modelA free Claude (claude.ai) or ChatGPT account for the early lessons.
3. First libraryWhen you start coding agents: pip install anthropic (or openai) + an API key.
4. Save your workA free GitHub account to build your portfolio.
🤖 Not sure how to set something up?
Ask the free DRONA AI tutor — paste it into Claude or ChatGPT and say "walk me through installing Python and getting an API key step by step." It adapts to your exact level.
Ready to build real AI agents?
Everything above is used, in order, inside the course.
Explore the Agentic AI Course →