Build It in an Artifact, Then Iterate in Place

What it is

Using Claude Artifacts — the side-panel where code, documents, and small apps render live — to build something and then refine it in place, instead of re-asking for a fresh copy each time.

The artifact is a living document you refine, not a message you re-ask.

Why it works

An Artifact is a persistent, editable object rather than a throwaway chat reply. Iterating on it keeps a single source of truth, so each change builds on the last version instead of a slightly different one Claude regenerated from scratch.

When to use it

Anything you'll revise: a landing page, a chart, a script, a formatted document, a small interactive tool. Best when you expect several rounds of change.

When not to use it

A one-off answer you'll paste elsewhere and never revisit — the Artifact overhead earns nothing there.

Prompt

Build this as an artifact so we can iterate: <what to build>.

Keep it in one self-contained artifact. When I ask for changes, update the same artifact rather than starting a new one.

Example

A pricing-table component starts as an Artifact; over four messages you tweak the tiers, colours, and copy — each change updating the same live preview instead of spawning four near-duplicates.

Advanced version

Ask Claude to keep the Artifact self-contained (no external dependencies) so you can copy it out and run it anywhere — a good default for shareable widgets and demos.

Common mistakes

Re-asking 'now make one with X' and getting a fresh artifact that lost your earlier tweaks.
Cramming several unrelated tools into one artifact so changes collide.
Using an artifact for a value you were only ever going to read once.