What Is Claude Design? Anthropic’s Design Tool, Explained for Agencies

Claude Design is Anthropic’s tool for making visual work by talking to Claude. It makes mockups, interactive prototypes, one-pagers and slide decks. It is in beta on Pro, Max, Team and Enterprise plans. Since September 16, 2026, it also works inside ordinary conversations and Claude Code, not only at its own address.

For an agency building sites for nonprofits, associations, colleges or public agencies, the canvas is easy to grasp. This piece walks through what Claude Design makes and where you can use it. It explains how a client’s design system gets in, how work gets out, and what it costs in usage. It ends with the limits Anthropic lists in its own help pages.

The short version: set up the client’s design system before anyone generates a mockup. Anthropic says new projects inherit the organization’s design system automatically. That one step decides whether the output looks like the client or like a template. Plan the exit before you start, because Anthropic documents PDF, PPTX, HTML, Canva and other routes, but not Figma. And budget for it, since Design draws from the same usage pool as chat and Claude Code.

What Claude Design does

The Get started with Claude Design help article describes a conversation paired with a canvas. You describe what you want in the chat. Claude generates a working design on the canvas beside it. You then edit on the canvas directly or keep asking Claude for changes.

That pairing matters for how teams work. A project manager can start a draft in plain language. A designer can then adjust spacing and layout by hand without re-prompting for every change. Anthropic’s June update added what it called "rich layout controls" for dragging, resizing and aligning elements.

Claude Design is not a replacement for a design file format. It produces its own projects, which you then export or hand off. That distinction comes up again in the exports section below, because it shapes where the work can go next.

From Labs preview to beta to every conversation

Claude Design has changed status twice in six months. The dates, from Anthropic’s own pages:

The April launch post said Claude Design was "available in research preview for Claude Pro, Max, Team, and Enterprise subscribers." The June update changed that to beta on paid plans.

The September change is covered in our post on the Claude Cowork merger. According to the release notes, "Claude Design works inside your conversations with all of its features, including on-canvas editing and importing your design system."

What Claude Design makes

Anthropic’s launch post listed "designs, prototypes, slides, one-pagers, and more." The help article today describes designs, interactive prototypes, one-pagers and other visual work. The product page adds mockups, clickable flows, campaign assets and branded presentations.

For our readers, that maps to familiar agency work:

• Homepage and template mockups for a redesign pitch or an early client review.
• Clickable prototypes of a membership join flow, a donation page or an event registration path.
• One-pagers for a program, a sponsorship package or a board summary.
• Decks that need to carry the client’s brand, such as a conference session or an annual report summary.
• Campaign assets in a consistent visual style.

None of these are final brand assets by default. They are drafts that a person with design judgment still needs to review.

Where you can use Claude Design

Anthropic now lists five places to reach it, according to the help article:

• In a conversation: ask for a design, such as a one-pager from a proposal you just discussed.
• In the Artifacts tab: pick a Design template.
• In Claude Code: ask for a design, or use the `/design` command, on desktop or in the terminal.
• In the Claude app for iOS and Android: ask for a design, check back later, and view it in the Artifacts tab.
• At claude.ai/design: the standalone experience keeps working, and existing projects stay there.

Mobile is view-only in practice. Editing on the canvas and changing sharing settings need Claude on web or desktop. Starting from a template does too.

One naming note. A Claude Design "project" is a design workspace. It is not the same thing as Claude Projects in chat, which we explain in what Claude Projects are.

The design system step decides the output

The feature that matters most for an agency is not the canvas. It is the design system.

Claude Design can import one or several design systems. The June update names three sources: "a GitHub repo, design files, or raw uploads." The help article adds a fourth, your local codebase through the /design-sync command in Claude Code.

Once a design system is in place, the help article says: "When you create a project, it automatically inherits your organization’s design system." Brand colors, fonts and components are then already there when someone starts a mockup.

That is the reason to do this step first. A mockup built on the client’s real tokens and components can move toward production. A mockup built without them can look finished and still be off-brand. Our explainer on brand consistency covers why that gap costs clients more than it seems.

Anthropic is candid about the limit here. Its help article warns that "Design system import is only as good as its source. A messy codebase or an incomplete file will show up in the output." If the client’s style guide lives in three half-updated PDFs, clean that up before importing.

/design and /design-sync in Claude Code

Claude Code has three design commands, according to Anthropic’s Claude Code commands reference.

/design drafts "UI mockups, screen flows, landing pages, or posters as artboards on one canvas," published as a Claude Design artifact. You edit the artboards in a desktop browser, and edits save automatically. Each artboard exports as PNG or PDF. The command requires Claude Code v2.1.265 or later and a session where artifacts are available. If an organization turns off the Design template, /design does not draft designs.

/design-sync converts a repository’s React design system and uploads it to Claude Design. The goal, in the reference’s words: "so designs it produces use your real components." Plan time for the first run. The reference warns that "A first-time sync verifies every component and can take a few hours on a large repo." A companion command, /design-login, authorizes design-system access with your claude.ai account.

Both are bundled skills. Our 101 on what AI skills are explains that model.

Our August 19 post on the Claude Code design skill reported that the commands reference then had no /design entry. As of September 29, 2026, the commands reference and the artifacts documentation both document it.

Note the React requirement. If a client’s front end is a WordPress or Drupal theme, /design-sync is not the documented route. Import from design files or uploads instead.

Getting work out of Claude Design

Pick the exit before you start the project. As of September 29, 2026, the help article lists these options:

The partner list reads: Adobe Experience Manager, Adobe for Creativity, Adobe Journey Optimizer, Base44, Canva, Gamma, HubSpot, Hyperframes, Lovable, Miro, Netlify, Replit, v0, Vercel and Wix. This list has changed between Anthropic’s pages before, so confirm it on the day you plan around it.

The Claude Code handoff is the route for build work. The launch post described a handoff bundle you pass to Claude Code with one instruction. Our explainer on what Claude Code is covers the other side of that handoff.

Figma is not on the list. The word does not appear in the Design help article, launch post, June update or product page. We found no documented Figma import or export for Claude Design. If your team reviews designs in Figma, test the PDF or HTML route first.

Usage limits and cost

Claude Design has no separate allowance. The help article says design activity "draws from the same pool as the rest of your work with Claude, including Claude Code, so there’s no separate Claude Design allowance to track."

That has been true since June. Before then, the launch post described Design using subscription limits with an option to continue through extra usage.

When you hit your limit, "Claude Design is unavailable until your limits reset," per the help article. If you have enabled usage credits, you can keep working past your included limits.

For a small agency, the practical effect is competition for one budget. A long design session can use up headroom a developer was counting on in Claude Code the same week. Watch that before a client deadline.

Enterprise plans billed by consumption work differently. Anthropic’s Enterprise consumption guide says Claude Design "bills from your organization’s consumption at standard API rates with org, group, and per-user limits applying." The same guide warns that agentic work and skills can generate token use that end users may not see.

Plans, defaults and admin settings

Anthropic’s pages do not fully agree on who can use Claude Design.

The help article says it is "available in beta on Pro, Max, Team, and Enterprise plans." The product page says it is included in all paid plans. The September 16 release notes say "Artifacts, including Claude Design, Claude Slides, and Claude Docs, are available on every plan, including Free." We could not settle the Free plan question from public pages as of September 29, 2026.

Defaults for paid plans are consistent. Design is on by default for Pro, Max and Team. On Enterprise, the help article says "it’s off by default until an owner turns it on in Organization settings > Artifacts."

The Design admin guide splits control in two. Standalone Claude Design has its own toggle under Organization settings > Claude Design. Design inside conversations and the Artifacts tab follows the Artifacts settings. The guide also says access to standalone Claude Design can be controlled through custom roles.

If you manage a Team plan for a nonprofit client, assume Design is already on for everyone. Decide whether that is what the client wants.

Known limits Anthropic lists

The help article lists seven known limitations as of September 29, 2026:

• Multi-person editing: “Two or more people editing a design project at the same time is still basic and may not work reliably.”
• No version history: Claude Design does not have it yet.
• Large codebases: linking very large repositories can cause lag or browser issues; Anthropic suggests Claude Code and `/design-sync`.
• Comments: inline comments sometimes do not appear on the page, though the comments view still shows them.
• Chat errors: for a “chat upstream error,” start a new chat tab in the same project.
• Mobile: view only; editing needs web or desktop.
• Import quality: a messy source produces messy output.

Two of these shape agency process. Without version history, export a PDF at each review milestone. That gives you a record of what the client approved. And do not plan a live co-editing session with a client yet. Have one person drive the canvas.

Data residency, audit logs and regulated clients

Some of our readers serve clients with strict data rules. Three documented gaps matter.

The Artifacts admin guide states: "Claude Design doesn’t currently support data residency requirements." It also says "Standalone Claude Design doesn’t support audit logs."

In Claude Code, /design needs artifacts to be available. The Claude Code artifacts documentation lists three setups that block artifacts: customer-managed encryption keys, HIPAA and zero data retention. Health-adjacent associations and some public agencies run under one of those setups.

If a client has residency or audit requirements, keep their materials out of Claude Design until Anthropic documents support. Use it for internal concept work on your own account instead.

Claude Design, Slides, or your designer’s tools

Claude Design exports PPTX, so it can make decks. Anthropic also offers a separate Slides template for presentations. Our explainer on Claude Docs and Slides covers that one.

A reasonable split, based on what each is documented to do:

• Reach for Slides when the job is a presentation, and brand fidelity is covered by a standard template.
• Reach for Claude Design when the deck is one piece of a visual system, or when you need mockups and prototypes on the client’s design system.
• Stay in your designer’s tools for final brand assets, print production and any review process built around Figma.

Where Claude Design fits for an agency serving nonprofits

Based on what Anthropic documents today, Claude Design fits early and internal work best. Concept mockups for a pitch. A prototype of a member join flow to test with staff. A one-pager draft that a designer then finishes.

It fits less well where the documented gaps bite. Final brand assets need a designer’s review. Multi-person editing is basic. There is no version history. Clients with residency or audit needs are not covered.

The deciding step is the design system. If you import the client’s system cleanly, Claude Design produces drafts that start close to the brand. Skip it, and you get polished output that someone has to redo.

Frequently Asked Questions