How Do You Build Something in a Claude Artifact?
Ask Claude for a self-contained thing, such as a document, slide deck, small tool or dashboard, and it opens in a panel beside the chat. Then you change that thing rather than asking for a new version in the conversation. Creating artifacts works on every plan, including Free.
Create an artifact: a [calculator / checklist / one-page plan / dashboard] for [who will use it]. It must let the user: 1. [first action] 2. [second action] 3. [third action] Use these real inputs: [paste your numbers, categories or text]. Keep everything on one page, with no sign-in. Don't add features beyond this list. If something is ambiguous, pick the simplest option and tell me which you chose.
- 1Ask for a finished thing
- 2It opens beside the chat
- 3Edit it or ask for changes
- 4Export or share it
- Docs, Slides, Designall plans
- Dashboards (beta)paid plans
- Motion (beta)Team and Enterprise
Try it now
- 1Check that it's on: go to Settings > Capabilities and turn on Cloud code execution and file creation, which may appear as Create and edit files.
- 2Send the prompt above with one real task from your week. To start from a template instead, open the **Artifacts tab, or select Output** in the message box.
- 3Use the thing in the panel. Click every button or read every section.
- 4You'll know it worked if it does your task with your data. If it shows an error, click **Try fixing with Claude**.
Use it when
Use one when you'll keep the result: a pricing calculator, a one-page plan, a client deck, or a checklist tool your team will open again. It also helps whenever you want to see the thing work rather than read about it.
Skip it when
Skip it for a quick answer you'll read once. Also skip it for anything that needs a real backend, user accounts or other people's private data at scale. Storage is limited to 20 MB of text per artifact, so at that point move to a real codebase with Claude Code.
Why it works
Anthropic's help center says Claude creates an artifact when content is substantial, typically over 15 lines, self-contained, and likely to be reused. That covers documents, code, single-page websites, images, diagrams, dashboards and small interactive tools. Because the result lives outside the chat, you can return to it, edit it and export it. You can also start from a template: Docs, Slides and Design are on all plans, Dashboards (beta) needs a paid plan, and Motion (beta) is for Team and Enterprise.
Example
You freelance and quote jobs by hand. You ask for a quote calculator with your real rates: $85 an hour for design and $110 for development, with a 15% rush fee and a field for hours per phase. It opens beside the chat. You enter last month's job to compare with what you actually quoted, and ask for one change at a time from there.
Going further
Artifacts can do more than display. On paid plans they can store data between sessions in personal or shared storage, limited to 20 MB per artifact and text only, and they can read and write connected apps. They can also call Claude, with each user signing in and spending their own plan's limits. Connector tools that need approval for each action aren't available to artifacts. Check whether an artifact uses shared storage before typing anything sensitive into it, since everyone who opens it can see that data.
Common mistakes
- Asking for a whole app in one prompt, then having nothing clear to test. Start with three actions.
- Leaving out your real numbers, so you get a demo full of made-up data you then have to replace.
- Asking for the same thing again in a new chat when it needed a small edit. That throws away work.
- Typing private data into an artifact without checking for shared storage.
Sources
Checked 6 November 2026. Products change often, so confirm details with the vendor.