For the last few days, we have been building something I wanted to make for a long time: a simple course around the AI tools that keep showing up everywhere.
We started with n8n for workflows and automation. Then Ollama for running AI locally. After that came OpenRouter, which makes working with many different AI models much easier. And yesterday we covered Hugging Face, the giant open ecosystem of models, datasets and AI tools.
You can read them here:
Course 1: n8n 101
Course 2: The Ollama Guide
Course 3: OpenRouter
Course 4: Master Hugging Face
There was one tool I kept for the end.
You probably already know its name. You may even have it installed.
But there is a good chance the GitHub Copilot you remember is not the GitHub Copilot that exists today.
And that difference is bigger than I expected.
Copilot quietly became a much bigger tool
GitHub Copilot started with a very simple idea: you write code, and AI helps complete the next few lines.
That was useful.
But today Copilot can do something very different.
You can give it a task such as:
Find why this login page is failing, fix the problem, run the tests and show me what you changed.
It can inspect the project, open the relevant files, edit several of them, run commands, see what failed and try again.
Or you can give it a GitHub issue and let it work on the task separately, then return with a pull request for you to review.
That change matters because GitHub is already where a huge part of software work lives. Copilot now has 50 million users, while GitHub itself has reached 225 million users. Microsoft says roughly one in three pull requests on GitHub now involves an agent.
So this guide is not really about autocomplete.
It is about learning how to use Copilot as a proper AI work tool.
And once you understand that, it becomes much more interesting.
In the full guide
We will set up Copilot from zero, learn the few parts that actually matter, give it proper project context, use Agent mode on a real task, create reusable instructions, work from the terminal, understand models and AI credits, and build a simple workflow that you can reuse in almost any project.
No complicated setup.
No giant agent architecture.
Just the useful part.
Read more
