AI

Claude’s artifacts just went public and multiplayer. The chat window is quietly becoming a software factory.

Anthropic rolled out public sharing and team editing for Artifacts — the mini-apps, dashboards and tools Claude builds in chat — and made them buildable from Slack via Claude Tag. The output of an AI conversation is now a product with a URL.

N Noah · The Sharp Brief · July 14, 2026 · 3 min read

Anthropic shipped a quiet update with loud implications: Claude users can now share artifacts publicly, teams can edit a single artifact together — including ones built with Claude Code — and artifacts can be created straight from Slack through Claude Tag. The rollout surfaced Monday night via feature-tracker TestingCatalog and was confirmed in industry briefs Tuesday morning.

If you haven’t used them, artifacts are the working things Claude builds inside a conversation — calculators, dashboards, trackers, internal tools, small apps. Until now they lived mostly where they were born: in one person’s chat. This update changes their nature. A public link turns an AI output into something you can hand a client. Multiplayer editing turns it into something a team can maintain. Neither requires anyone to open a code editor.

The Slack piece is the strategic tell. Anthropic has been widening Claude’s surfaces all year, and Claude Tag — the shared @Claude teammate it launched into Slack channels on June 23 for Enterprise and Team customers — is the newest front door. Making artifacts buildable from inside Slack means the software gets made where the work conversation is already happening, not in a separate tab. That’s the same ground OpenAI is chasing with ChatGPT Work and Google is chasing with its Gemini Enterprise agent stack: whoever owns the place teams delegate work owns the relationship.

The stakes are not small. Anthropic closed a $65 billion Series H in May at a $965 billion post-money valuation — above OpenAI’s most recent $852 billion mark — with run-rate revenue the company says has crossed $47 billion, and a confidential S-1 already filed. At those numbers, feature races stop being about features. They’re about which platform becomes the default place work product gets made.

Our take: “Share” buttons sound like table stakes, and they are — every lab will have this within months. The real shift is what it does to the definition of software. When a chat output is a working product with a URL, the AI assistant starts competing with no-code tools, internal IT queues, and the bottom tier of custom dev shops all at once. For a one-person business, the play is immediate: prototype the client dashboard or intake tool in a prompt, send the link, and look like you have an engineering team. The thing to be clear-eyed about: features converge, distribution doesn’t. The Slack integration matters more than the share button, because switching costs live where your team already talks.

What to watch

Advertisement

Get the day, decoded — at 7 PM ET

The Sharp Brief: AI, money, business & performance in five sharp minutes. Free.

Free bonus: subscribe today and The 2026 AI Playbook (PDF) lands with your welcome email.