Hidden Features

Artifacts

Turn answers into editable, runnable side-panel documents.

Claude Artifacts

Answers you can use, not just read

Artifacts turn a response into a live object in a side panel — a document, a component, a small working app — that you refine in place rather than rebuilding in chat. Instead of scrolling back through a conversation to find the latest version of something, you have one artefact you iterate on directly. For anything you are building rather than just asking about, this changes the workflow: the output becomes a thing you shape, with the chat reserved for the instructions that shape it.

Build, then iterate in the panel

The natural loop is to build something in an Artifact, then iterate on it there — 'make the header sticky', 'tighten this section', 'handle the empty state' — while the artefact updates in place. Because the object persists, you keep a stable thing to point at instead of regenerating from scratch each turn. Reach for Artifacts whenever you are producing something concrete and revisable; keep the plain chat for quick questions where there is nothing to hold on to.

Presentation

Claude Artifacts, in three slides

Claude Artifacts, in three slides — slide 1
1 / 3

A short deck generated from this article — swap in your own NotebookLM slide export and audio overview.

Frequently asked
What are Claude Artifacts?

Artifacts turn a response into a live document, component, or small app in a side panel that you refine in place, rather than rebuilding it in the chat stream.

When should I use an Artifact?

Whenever you are producing something concrete and revisable — a document, UI, or app — and want to iterate on it directly instead of regenerating from scratch each turn.

How do Artifacts differ from normal chat answers?

A normal answer lives in the conversation; an Artifact is a persistent object you shape over many turns, with the chat used for the instructions that change it.