Anthropic launched Claude Tag, transforming its AI assistant into a persistent team member that operates directly within Slack workspaces.

The integration allows Claude to join Slack channels with configurable access to tools, data, and codebases. Teams can tag Claude into conversations to delegate tasks asynchronously while focusing on other work.

Claude Tag represents a significant evolution from one-on-one chat interfaces to multiplayer, proactive AI collaboration. The product merges Anthropic's existing Claude Code capabilities with team-wide workflow automation.

Key capabilities and internal adoption

The system can tag relevant coworkers who own related code, wait for blocking dependencies across multiple days, and summarize discussion threads into actionable documentation.

In ambient behavior mode, Claude Tag responds to channels without explicit tagging, follows up across different channels, and monitors thresholds to trigger automated fixes or alert teams when A/B tests reach statistical significance.

Anthropic's internal adoption metrics show Claude Tag now writes 65% of the product team's code. The company has used the system internally throughout 2026, with Claude Tag reportedly building much of itself.

"Our first product that is natively multi-player and proactive," said Anthropic product lead Cat Wu on X.

Technical architecture and availability

Claude Tag operates as a workflow layer around existing Claude models rather than introducing new foundation model capabilities. Administrators configure permissions for channel access, tool integration, and codebase connectivity.

The system supports git webhooks for long-running development workflows and can coordinate "stacked prompts" across extended timeframes. Teams can customize Claude Tag through hundreds of configuration options across six common workflow patterns.

The feature launches in beta for Claude Enterprise and Team plan subscribers. Anthropic positions Claude Code as optimal for solo synchronous work, while Claude Tag handles multiplayer asynchronous collaboration.

Alex Albert from Anthropic described the experience as "less like using a tool and more like managing a team."