Can Claude Build a Website?

Yes, Claude can build a website, in two different senses that are easy to mix up. It can produce a single, self-contained web page as an artifact that opens at a claude.ai link — genuinely useful, but not a site on your own domain. Or it can use Claude Code to write real project files for a multi-page site, which becomes a live website on your domain only after you take those files through a separate deployment step. Anthropic's own documentation draws a clear line between these, and this guide follows it.

What can Claude actually produce: an artifact, or Claude Code files?

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 its documented content types explicitly include "single-page websites" alongside documents, code snippets, and dashboards (Claude Help Center, "What are artifacts and how do I use them?", checked September 2026). Ask for one in a chat, or start from the Design template, and Claude builds it in a panel beside the conversation. The same page says artifacts are available on Free, Pro, Max, Team, and Enterprise plans, and in Claude Code on every plan that includes Claude Code — though templates, connecting an artifact to your other apps, and persistent storage need a paid plan.

Claude Code is a different tool for a different job: an agentic coding assistant that reads a project's files, edits them, and runs commands in a real repository (Anthropic, Claude Code quickstart, checked September 2026). Point it at an empty folder and ask for a multi-page static site, and it writes ordinary HTML, CSS, and asset files you can inspect, version, and deploy yourself — our multi-page Claude Code guide walks through exactly that workflow, from the brief you give it to testing every route locally.

Can a Claude artifact be a full website?

Not a multi-page one, no — and Claude Code itself documents a second, related artifact feature worth knowing about here. Separately from the chat-based artifacts above, Claude Code (the CLI and desktop app) can publish its own session output as what Anthropic calls an artifact too: "An artifact is a live, interactive web page that Claude Code publishes from your session to a private URL on claude.ai," available "on Pro, Max, Team, and Enterprise plans" with a session signed in (Anthropic, "Share session output as artifacts", checked September 2026). For a website, the Claude Code docs state the key limit plainly: "An artifact is a capture of work: one self-contained page with no backend, so it can't serve multiple routes. For a hosted internal tool with a backend, deploy it on your own infrastructure instead" (same source). Anthropic's page-constraints table spells out why a multi-page site doesn't work as an artifact: "Relative links do not resolve, because nothing is deployed alongside the page." If you need /about, /services, and /contact as real, separately linkable pages, an artifact is the wrong shape — that's a Claude Code project-files job instead.

There's a technical ceiling too: per the same Claude Code artifacts page, an artifact page can load typefaces only from Google Fonts and scripts only from five public CDN hosts (cdnjs, unpkg, the Tailwind and jQuery CDNs, and selected jsDelivr paths), with every other external image, script, stylesheet, and font blocked, and the whole rendered page capped at 16 MiB (checked September 2026).

How do I share an artifact once it's built?

This differs by when the artifact was made and which plan you're on — our Claude artifacts guide covers the full breakdown, but the short version: an artifact made in a chat before September 16, 2026 is a legacy artifact and shares by Publish (Free, Pro, Max) or Share & copy link (Team, Enterprise); newer artifacts use Share, with access levels, email invitations (in beta on paid plans), and on Team and Enterprise, sharing inside your organization. The account rule matters if you want a public website: "Everyone needs a Claude account. People without one can't open a shared artifact, even with the link. The only exception is a legacy artifact published from a chat" (Claude Help Center, "Share artifacts", checked September 2026). Anthropic's Claude Code artifacts page describes its sharing differently: a public link is one "that anyone on the internet can open, with no claude.ai sign-in required," and "On Pro and Max plans, a public link is the only way to share an artifact. On Team and Enterprise plans, public sharing is off until an Owner enables it for the organization" (checked September 2026). Because the two pages don't match, open your link in a signed-out browser window before you send it anywhere. For most chat-made artifacts, visitors need their own Claude account — the opposite of how a public marketing site is supposed to work.

Where does an artifact actually live, and why isn't that your website?

Every artifact link points back to Anthropic's own infrastructure, not a domain you control. Anthropic's Claude Code documentation is explicit about the address: an artifact opens at a URL like https://claude.ai/code/artifact/5fbea6f3-..., and "the viewer on claude.ai loads each artifact from a sandboxed *.claudeusercontent.com origin" (Anthropic, "Share session output as artifacts", checked September 2026). Nothing in either artifacts page describes attaching a custom domain to one. That's the practical reason an artifact link, however polished the page looks, isn't the same thing as a business website at its own domain — it's a page hosted on Anthropic's servers, reachable only through a claude.ai (or claudeusercontent.com) address.

How do I get a real, multi-page site on my own domain?

This is where Claude Code's project-files route, not an artifact, is the right tool — and where a genuinely separate step comes in. Claude Code can write a full folder of static pages: an index.html, a services/index.html, a contact/index.html, and a shared stylesheet, with real relative links between them, exactly as described in our multi-page Claude Code guide. But writing those files is not the same as publishing them. Once you've reviewed the output locally, publishing an AI-generated static website covers the actual deployment step: uploading the finished folder to a host, verifying every route on the live public URL in a signed-out browser window, and only then connecting a custom domain. Anthropic's own Claude Code documentation doesn't describe a hosting or domain-connection feature at all — that's expected, since Claude Code's job ends at writing and testing the code, not choosing where it lives publicly.

A quick reference

Is Figment a fit for this job?

Only if the actual starting point is a Figma design, not a Claude artifact or a Claude Code project. Figment publishes a Figma design as a live website through its Figma plugin; its AI editor, public now, edits a separate static copy of an existing site, and it has no connection to Claude's artifacts or Claude Code. If you've used a Claude artifact to draft page copy or a layout idea, the practical next step is bringing that content into a Figma file and publishing it from there, rather than treating the artifact link itself as the finished, public site.

Bottom line: Claude can genuinely build a website in the narrow sense of a single page, shareable as an artifact — but that page lives on Anthropic's own domain, often requires a Claude account to open, and can't hold more than one route. For an actual multi-page site on your own domain, the real path is Claude Code writing project files, followed by a separate, deliberate deployment step. Confusing the two is the mistake to avoid: a working artifact link is not evidence that you have a hosted, public website yet.