Anthropic launches Claude Tag for Slack AI agent collaboration

Anthropic unveiled Claude Tag, a beta feature for Claude Enterprise and Claude Team subscribers that integrates its AI assistant directly into Slack channels. Employees can now tag @Claude with a task in natural language, and the agent will decompose the request into stages, access approved tools, data, codebases, and conversational context, then respond in the thread. Unlike a private chatbot, Claude operates in shared channels, allowing team members to see its actions and continue conversations. Rob Seaman, Slack's GM, described it as making AI multiplayer. Anthropic's head of product for Claude Code, Cat Wu, noted that treating Claude like a coworker via tagging is a powerful form factor. Administrative controls are a core selling point: IT admins can define separate Claude identities per channel or team (e.g., sales vs. engineering), set token spend limits per organization and channel, and view detailed logs of every task and request. Claude will not report from private channels. Additionally, Claude Tag supports asynchronous work—users can assign tasks and check back later—and an ambient behavior mode proactively flags relevant updates, follows up on unresolved threads, and alerts users. Anthropic revealed that 65% of its own product team's code is already produced by an internal version of Claude Tag, underscoring its real-world effectiveness. The beta is limited to Slack initially, with plans to expand to other platforms soon.