01 Quick Start
Artifacts in 20 minutes
20 min·Reviewed Sep 30, 2026
Prerequisites:Syllabus
Search for a command to run...
This lesson is the fast on-ramp for everyone who is comfortable chatting with Claude but has never, or only rarely, worked with artifacts. In twenty minutes you will learn the handful of ideas the rest of this course leans on: what an artifact is, where it appears on screen, how to ask for one, the one setting that has to be switched on, where your artifacts are collected, how versions let you change your mind safely, what to do when an artifact shows an error, and how sharing works (spoiler: everything starts private). Each idea gets one short section and nothing more. The lesson ends with a warm-up you build yourself: a small interactive tin price calculator for Kettle & Kite, the fictional tea shop that is launching its Autumn Ember Blend on Saturday 17 October 2026. Mara, Jonah and Priti, the shop's team, will need a lot of small tools before launch day, and this calculator is the first. You will not write any code. You describe what you want in plain language, look at what Claude builds, and ask for changes.
Anthropic's help center defines an artifact as "anything Claude makes for you that you'd put in front of someone: a design, a deck, a document, a dashboard, or a small interactive tool," and notes that it opens beside your conversation so you can edit and share it (Anthropic Help Center, What are artifacts). That definition is a useful test. A quick answer to a question stays in the chat. A price list that Priti will print for the counter, or a calculator Jonah will use at the till, is something you would show to other people, so it belongs in an artifact.
A second help center article describes when Claude creates an artifact on its own: when the content is "significant and self-contained, typically over 15 lines," and is something you are likely to edit, iterate on or reuse outside the conversation, that stands on its own without extra context, and that you are likely to refer back to later (Anthropic Help Center, artifact creation criteria). Short replies do not qualify. That is why a two-line answer never turns into an artifact, while a request for a full page usually does.
This course focuses on the last item in Anthropic's list, the small interactive tool: a page with inputs, buttons and results that change as someone uses it. Calculators, checklists, quizzes and trackers are all in this family.
Glossary: artifact. A self-contained piece of work Claude makes (a document, page, dashboard or tool) that opens in its own panel beside the chat, where you can view it, interact with it, revise it and share it.
When Claude makes an artifact, it opens beside your conversation rather than inside the message stream (Anthropic Help Center, What are artifacts). The result is a two-part screen. The chat, where you ask and revise, sits on one side. The artifact panel, where the actual thing renders, sits on the other. For an interactive tool the panel is live: you can type into its boxes and click its buttons just as a colleague would.
Image coming soon: A simplified two-column mockup. On the left is a sidebar with an "Artifacts" entry highlighted and labelled "Gallery of everything you made". In the middle, a chat shows a user request for a tin price calculator and Claude's reply containing a clickable artifact card. On the right, a large panel titled "Tin price calculator" with "Version 1" and "Share" buttons in its header has a placeholder saying the artifact renders here and is interactive.
Simplified mockup (course-made, not the real interface): the chat on the left, the artifact panel on the right, and the sidebar entry that leads to your gallery.
Two habits help here. First, when you want a change, type it in the chat. The panel shows the result, but the conversation is where you steer. Second, the interface changes often, so treat every button name in this course as "look for something like this" rather than an exact label. The concepts stay stable even when a button moves.
The help center lists three ways to start (Anthropic Help Center, What are artifacts):
Templates (Docs, Slides and Design) are available on paid plans (Anthropic Help Center, template types). Interactive tools usually start from a plain request, so you can ignore templates for now.
A good request names three things: what the thing is, who will use it, and what it has to do. Saying "interactive" or "page" makes it clear you want a working tool rather than a paragraph of explanation. For example:
Make me a small interactive page for Kettle & Kite, a tea shop. Jonah will use it at the counter to work out
the price of an order of our 100g tins. It needs a box for how many tins, and it should show the total as he types.
You do not need to describe colours, layout or code. Claude chooses sensible defaults and you correct them afterwards. Revision is cheap, as the sections below show, so a quick first draft beats a perfect first prompt.
If Claude answers in plain text when you expected an artifact, check one setting first. Artifacts need "Cloud code execution and file creation" to be turned on. On Free, Pro and Max plans it is under Settings > Capabilities. On Team and Enterprise plans it is under Organization settings > Capabilities, so an administrator may be the person who controls it (Anthropic Help Center, required setting).
Plan matters less than you might expect. Anthropic's availability table shows that every plan, from Free to Enterprise, can create artifacts in chat. The extras (templates, connections to apps, and data storage) are "available on Pro, Max, Team, and Enterprise plans" (Anthropic Help Center, What are artifacts). Creating and revising artifacts works on every plan; sharing options vary by plan (see the table below).
Glossary: capability. A switch in Claude's settings that turns a feature on or off for your account or organization. If a capability is off, Claude simply can't use that feature, even if you ask it to.
You do not have to scroll back through old chats to find last week's calculator. The help center describes an Artifacts tab in the Claude sidebar, where everything you create is automatically saved and organized (Anthropic Help Center, viewing artifacts). The same tab is also one of the three places to start a new artifact (Anthropic Help Center, What are artifacts).
Think of the tab as a gallery. By launch day Kettle & Kite might have a tin calculator, a tasting-notes card and a pre-order checklist in there. Giving each one a clear name when you ask for it ("call it K&K Tin Calculator") makes the gallery far easier to scan later. For artifacts published from Claude Code (a different product with its own gallery), Anthropic's docs say Claude picks the title, which shows in that gallery and in shared links, unless you ask for a specific one (Anthropic, Claude Code artifacts docs). Naming your artifact when you ask for it is this course's advice for the chat gallery too.
Revising an artifact does not throw away what came before. There are two documented kinds of version history, and it helps to keep them apart:
Image coming soon: A simplified mockup in two parts. Part A shows an artifact panel titled "Tin price calculator" with a version dropdown open (labelled as documented for Cowork), listing Version 1 "first draft", Version 2 "added 250g tin" and Version 3 "autumn colours" (selected). Part B shows the same panel at Version 4 with a red error card reading "Something went wrong while running this artifact", the message "total is not defined", and a "Try fixing with Claude" button with a note that success is not guaranteed.
Simplified mockup (course-made, not the real interface): (A) version history, as documented for Cowork; (B) an error card with the fix button described in the next section.
In practice this means you can experiment with less risk. If Mara asks for a bold ember-orange redesign and hates it, use the version control if your panel shows one, or edit the message before the redesign to branch the chat back to the earlier look.
Interactive tools run real instructions behind the scenes, so sometimes one fails to load or shows an error in place of the page. When that happens, the help center says you will be offered a "Try fixing with Claude" button. It also warns that "success isn't guaranteed" (Anthropic Help Center, What are artifacts).
A sensible routine, in order:
You can also edit an artifact without the chat. The help center says artifacts can be changed through conversation or edited directly in template-based artifacts (Anthropic Help Center, What are artifacts). For the interactive tools in this course you will mostly revise by chatting.
Nobody sees your artifact unless you choose to show them. In Anthropic's words, "Artifacts start private to you. Nothing is shared until you share it." To share, click Share on the artifact, choose who can see it, and copy the link (Anthropic Help Center, Share artifacts).
Who you can share with depends on your plan (Anthropic Help Center, Share artifacts):
| Plan | Who you can share with |
|---|---|
| Free | Can publish legacy artifacts from chats publicly |
| Pro / Max | "Anyone with the link", or specific people invited by email |
| Team / Enterprise | Organization members, specific invitees, or a public link if enabled |
To stop sharing, reopen the Share menu and choose "Only you" (Pro/Max) or "Only people invited" (Team/Enterprise) (Anthropic Help Center, Share artifacts). Shares can be tied to versions. The Cowork documentation says you choose whether to share "the latest or a specific version" (Anthropic Help Center, Use artifacts in Claude Cowork). The screenshot below shows that choice in Anthropic's artifact viewer.
Screenshot, Image: Anthropic, Claude Code artifacts docs. The Share menu with a version picker and audience selector. That page covers artifacts published from Claude Code, but the Share controls it shows illustrate the same private-until-shared idea.
Every tool you build in this course follows the same loop: ask, look, revise, keep the versions, and share when it's ready.
Go deeper (optional): the course Claude Artifacts Essentials covers these basics at a slower pace. You do not need it for anything here.
Time to make your first interactive tool. Kettle & Kite sells the Autumn Ember Blend in 100g tins at 12.50 each, and anyone who buys 10 or more tins gets 10% off the whole order. Jonah wants a page he can use at the counter. Paste this into a new chat:
Build a small interactive tin price calculator for Kettle & Kite, a fictional tea shop, as an artifact.
- Our Autumn Ember Blend comes in 100g tins at 12.50 each.
- Add a box where I type how many tins.
- Show the subtotal, the discount, and the total, updating as I type.
- If the order is 10 tins or more, take 10% off the whole order and show a small "10% off" badge.
- Use warm cream, brown and ember-orange colours, and put "COURSE EXAMPLE, FICTIONAL COMPANY" in a banner at the top.
Call it "K&K Tin Calculator".
Your result will look different from ours, which is expected. It should behave like this:
Image coming soon: A cream-coloured calculator card under a dark brown banner reading "COURSE EXAMPLE, FICTIONAL COMPANY". The card is titled "Kettle & Kite" with the subtitle "Autumn Ember Blend: tin price calculator". It lists price per 100g tin 12.50, a quantity box containing 12, subtotal 150.00, a bulk discount row with an orange "10% off" badge showing -15.00, and a large ember-orange total of 135.00, above a note saying 10 or more tins get 10% off.
Course example (course-made): a working tin price calculator with 12 tins entered. 12 × 12.50 = 150.00, less 10% (15.00), gives 135.00.
Now test it the way a careful colleague would. Type 9, then 10, and check that the badge appears exactly at 10. If something is off, say so in the chat. That habit of testing and then asking for a fix is what the rest of this course builds on.