Anthropic's Claude Tag turns Slack channels into AI-powered team members

Anthropic’s Claude Tag is a new beta feature that turns Slack from a communication tool into a collaborative AI workspace. Employees can @-mention Claude in a shared channel with a natural-language request, and the agent breaks the task into stages, connects to approved tools and data sources, and replies in the thread. Unlike private chatbots, Claude Tag operates transparently — colleagues see its actions and can continue the thread. Slack GM Rob Seaman calls it “making AI multiplayer,” while Anthropic’s Cat Wu notes that tagging Claude as a co-worker fundamentally changes the interaction dynamic. The feature uses Opus 4.8 and replaces the existing Claude Slack app.

Enterprise controls are central: admins can set token spend limits at the channel or org level, configure which tools Claude accesses per channel, and review logs of every request. Different teams maintain separate Claude identities, preventing context leakage between channels, and Claude cannot access private channels. An optional ambient mode lets Claude proactively flag relevant information and follow up on unresolved threads, raising compliance questions. Australia is a key market — it ranks 4th globally in both Anthropic’s AI Usage Index (3.27x above expected) and enterprise Slack deployments (5.4% of 11,858 companies). With Australia’s Privacy Act reforms ongoing, enterprises in finance, healthcare, and critical infrastructure must assess new exposure points when an AI agent is a standing participant in shared workspaces.