Claude Design, Slides and Docs: how to make designs, decks and documents in any Claude chat

On September 16, 2026, Anthropic made it possible to ask Claude for a design, a deck or a document in any conversation, including in Claude Code and the Artifacts tab, and to edit the result with Claude or directly as you work. This guide explains the three templates, who can use them, how to start and edit, how sharing and export work, and the limits, using Claude’s own Help Center. We haven’t tested every feature, and the templates are in beta.

What an artifact is now

Claude’s help page 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”. It opens beside your conversation, and on paid plans it is saved in the Artifacts tab in the sidebar, so you can find it again from any conversation. Artifacts made in a chat before September 16, 2026 are now “legacy artifacts”: they keep working and can still be published and shared, but you can’t make new ones.

The three templates

Who can use them

Artifacts need Cloud code execution and file creation turned on in Settings under Capabilities. Claude Docs isn’t yet available for organisations that use customer-managed encryption keys, zero data retention or a HIPAA-ready configuration. Claude’s plans are covered in Claude Pro vs Max.

How to start

• In a conversation. Ask, for example, “Make a one-pager from this proposal” or “Turn the plan we just worked through into a product spec I can share with the team.” You can also choose Output in the message box and pick a template, or start a request with /docs.
• From the Artifacts tab. Open it and choose a Design, Slides or Docs template.
• In Claude Code. Use /design to create, edit and sync designs, or ask for a doc. On desktop it opens in the side panel and in the terminal Claude gives you a link to the web.
• On iOS and Android. Ask for a design, deck or doc in any chat and view it in the Artifacts tab. Starting from a template, editing and changing sharing need Claude on web or desktop.
• At claude.ai/design. The standalone Claude Design experience keeps working, with your existing projects.

A good request says what it is for, who will read it and what it should cover. Anthropic’s example: “a two-page launch brief for our sales team covering pricing, timing, and the top three customer questions” beats “a launch brief.”

How editing works

Nothing is locked while Claude works, and you can edit alongside it. There are three ways to change an artifact:

1. Edit directly. Type in a doc, edit any slide, or drag, resize and align elements on a design canvas.
2. Comment. Click an element and leave a comment for a targeted change, or mention @Claude in a doc comment, and Claude replies in the thread, makes the edit and explains why. In a doc or Markdown document you can also highlight text and choose Edit with Claude.
3. Ask in the conversation. Use chat for broad changes such as “Make the color scheme darker and more minimal” or “Show me 2–3 alternative layouts.”

To explore a different direction without losing your current work, Anthropic suggests telling Claude “Save what we have and try a completely different approach”, or editing an earlier message to create a separate version of the chat.

Claude Design and your design system

Designs and decks can use a design system, so new work picks up your colours, fonts and components. In Claude Design a new project inherits your organisation’s system automatically, and you can bring in systems from a GitHub repo, design files, uploads or your local codebase with the /design-sync command in Claude Code. Anthropic says Claude builds with your real components and checks its own output against the system. When a design is ready to become software, you can hand it off to Claude Code, which continues from your existing work rather than a screenshot.

Sharing

Artifacts start private. To share a design, deck or doc, open it and choose Share, then pick who can open it and what they can do. Some rules differ by type:

• Designs and decks: people can view, comment on or edit.
• Docs: viewers can read, and editors can read, edit, comment and export. There’s no comment-only level yet, and only the owner can rename or delete a doc. Deleting is permanent, with no trash.
• Outside your organisation: docs can’t be shared by email invitation on any plan. On Team and Enterprise they can’t be shared outside the organisation by link either, while Pro and Max users can share with anyone who has the link. Opening a shared doc needs a Claude account.

Usage and limits

Claude Design counts toward the same usage limits as the rest of Claude, including Claude Code, with no separate allowance, and Docs likewise counts toward your plan’s usage. Complex projects with large code bases or many iterations use more. If you hit your limit, Design is unavailable until it resets, unless you’ve enabled usage credits. Known limitations include an intermittent issue where inline comments can disappear before Claude reads them, for which Anthropic’s workaround is to paste the feedback into chat.

Artifacts that act like small apps

Artifacts can also connect to your apps, store data between sessions (20 MB per artifact, text only) and call Claude directly so that they work as small apps, with usage counted against each person’s own plan. Anthropic advises only building on published artifacts from people you trust.

How it compares

OpenAI’s answer, announced on September 29, is ChatGPT Space and Pages, and Google has Workspace. For Claude’s agent work, see What is Claude Cowork?, and for the chat side, the Claude profile and Claude models explained. The newest model behind all this is Claude Sonnet 5.5, which Anthropic says is strong at polished documents, slides and spreadsheets.