
You need three things: a place to run code, an AI assistant that can see your project, and a folder that keeps each tool tidy.
Advertisement
1. Install Python
Download Python from python.org (tick “Add Python to PATH” on Windows). Check it works:
python --version
2. Pick an assistant
There are two styles of AI coding assistant:
- Editor assistants built into a code editor (for example VS Code with an AI extension, or AI-first editors). You see the files and the changes side by side.
- Agentic assistants such as Claude Code that run in a terminal or desktop app, read your whole folder, run the code themselves and fix their own errors.
For beginners, an agentic assistant is often easiest: you describe the tool, it writes the files, runs them and shows you the result.
3. One folder per tool
csv-cleaner/
README.md what the tool does, in your words
sample/ 2-3 small, anonymised example files
expected/ what the output should look like for those samples
cleaner.py written by the AI
The sample and expected folders are your secret weapon: the AI can test its own work against them, and so can you.
⚠️ Use anonymised samples. Replace real names, phone numbers, PANs and bank details before you put a file where an AI assistant can read it.
4. The rhythm
- Describe one small step.
- Let it write and run the code.
- Compare the output with what you expected.
- Describe what is wrong, specifically.
Small steps win. “Read the CSV and print the first 5 rows” then “remove blank rows” beats “clean my data” in one go.
✨ Ask AI about this article
Stuck on a step? Ask a question and the AI answers using this article.
Free · AI can be wrong