What is an AI coding assistant?

Technical definition

An AI coding assistant is software that applies a code-trained language model to development tasks within a working environment. The category divides by how much of the work the tool owns:

• Completion tools suggest the next lines as you type inside an editor. They assume you are writing the code.
• Chat assistants answer questions and produce snippets in a conversation. You move the output into your project.
• Full-stack agents take a described outcome and build, run, and revise a complete project across many files.

In plain terms

An AI coding assistant is a tool you tell what you want, in words, and it produces the code.

The important question when choosing one is how much you want to do yourself. A completion tool is useless if you cannot write code. A full-stack agent is unnecessary if you only want help finishing a function.

Why it matters

The three categories serve different people, and the mismatch is the most common reason someone concludes AI coding does not work for them.

A non-technical founder handed a completion tool will get nowhere, because the tool assumes a developer is driving. Give the same person a full-stack agent and they ship a product that week. A senior engineer has the opposite experience.

So the choice is about your starting point, not about which tool is strongest. Ask what you can do without the tool, and pick the one that covers the rest.