What is Copilot?

GitHub Copilot is an AI coding assistant that writes code for you as you type. Think of it like autocomplete on your phone, but for programming. It was created by GitHub (which is owned by Microsoft) in 2021 and became one of the first AI coding tools to go mainstream.

When Copilot first launched, it worked as an extension inside your code editor. As you typed, it would show you suggestions for what code you might write next. You could accept or dismiss them with a keystroke. It supported VS Code, JetBrains IDEs, and a few others. The suggestions got surprisingly good pretty quickly. It could generate entire functions just from a comment describing what you wanted.

For a while, Copilot was mostly about inline suggestions. But GitHub kept expanding it. They added Copilot Chat, which lets you have a conversation about your code right inside the editor. You can ask questions, request changes, and debug issues without leaving your IDE.

Meanwhile, Cursor came along and showed what an AI coding tool could look like when it was built around agentic workflows from the start. Cursor could autonomously work across multiple files, understand your whole codebase, and take on bigger tasks. It made Copilot’s inline suggestions feel a little limited by comparison.

So Copilot built Agent mode. Instead of just suggesting code, Copilot Agent can now work on tasks across multiple files on its own. It can read your codebase, make changes, run commands, and iterate on problems. Suddenly Copilot was back in the game, and in similar territory to tools like Claude Code and Codex.

The pricing is $10/month for individuals and $19/month for business teams. GitHub also offers an enterprise tier with additional security and compliance features.

I used Copilot in WebStorm for a while. At first I thought it was just a better autocomplete. But when Agent mode came out it was much more capable. I’d been using JetBrains IDEs for 15 years, so I probably would have been perfectly happy sticking with Copilot if terminal agents like Claude Code and Codex hadn’t come along. The terminal workflow just ended up being faster for the kind of work I do.