Artifacts

Artifacts is a feature that takes results Claude creates—like code, writing, or tables—and displays them in a separate window next to the chat, so you can keep the conversation going while viewing and editing that output on its own.

Here's an analogy. A regular chat window is like exchanging letters: if you want to look at a picture or document someone sent, you have to keep digging through the stack of letters. Artifacts is like taking that picture or document out and laying it flat on your desk. The conversation flows on its own, while the output stays pinned in the side window, updating on the spot whenever you ask for a revision.

This structure became the foundation that many later work tools built upon. Tasks like creating and comparing multiple screen mockups, or writing code and checking it right away, all happen on top of this separate window.

In articles, it appears in phrases like "artifacts, Claude's output format," typically when explaining the foundation that a newly released feature is built on. It's worth not confusing the two: Artifacts itself isn't a new feature this time around—it's been around since before—while the screen-mockup workflow that's gotten attention is a new feature layered on top of it.

Try entering a prompt on Claude's web or app version asking it to write code or organize a document. For example, if you ask "Make a simple to-do list webpage," you'll see the answer appear in a separate window instead of the chat. If you then follow up with something like "Change the button color to blue," you'll see the conversation keep accumulating as usual while only the output in the separate window gets updated.