Anthropic launches Claude Tag, an always-on AI teammate that lives inside Slack

Anthropic's Claude Tag is pitched as a new paradigm: rather than a chat window you visit, Claude becomes an inline participant in the channels teams already use, starting with Slack. Claude Code lead Boris Cherny described it as 'proactive, multiplayer, with its own identity and memory,' saying it has 'totally changed how we use Claude' internally over the past few months. Users grant Claude access to selected channels and connect it to tools, integrations, compute environments, data, and codebases, letting it accumulate institutional knowledge and act on tasks across the organization.

Mechanically, the launch rides on Anthropic's existing agentic stack (Claude Code, Claude Cowork) plus the under-the-hood engineering to make cross-tool memory, security, and integrations 'just work'—a point Andrej Karpathy highlighted in a widely shared post calling it significantly more 'inline' with org-wide human activity. It is available in beta through Claude Enterprise on AWS Marketplace, and to Enterprise and Team customers directly.

Competitively, Claude Tag squares off against Microsoft 365 Copilot, Google's Gemini-in-Workspace, and Slack's own AI, but leans on Anthropic's coding strength to differentiate. Alongside the launch, Anthropic updated its privacy policy to let it request ID verification from flagged accounts—a move tied to its tense standoff with the U.S. government over frontier-model access.

What to watch: an HN thread on Claude Tag drew 234 points and 161 comments, but a concurrent 'elevated error rate across multiple models' status incident (205 points, 253 comments) underscored reliability questions for an 'always-on' agent embedded in workflows. The ID-verification policy also drew privacy criticism. Adoption will hinge on whether enterprises trust a persistent agent with channel-wide memory and codebase access.