Anthropic launches Claude Tag for Slack
Anthropic's Claude Tag puts an always-on AI teammate in your Slack channels, in beta for Team and Enterprise. It replaces the old Claude in Slack app.
Anthropic launched Claude Tag on June 23, a beta that puts an always-on Claude inside Slack channels for Claude Team and Enterprise customers. You tag @Claude the way you would a colleague, and it reads the channel’s history plus any connected data sources, keeping that context as the conversation grows. It replaces the older Claude in Slack app.
The pitch is a shared teammate, not a personal assistant. One Claude identity runs per channel, so everyone sees what it does and anyone can pick up where it left off. Hand it a task and it breaks the work into stages, uses whatever tools an admin has granted it, and replies in a thread with what it produced. An optional ambient mode lets it act on its own, surfacing updates from elsewhere in the company and circling back on stalled threads.
Admins set the boundaries. They define the tools, data, and channels each instance can touch, with spending limits and full activity logging, and instances stay siloed so one department’s Claude cannot leak into another’s. The feature runs on Opus 4.8.
Why this matters
Anthropic says 65% of its own product team’s code is now created by an internal version of Claude Tag, which is the number to weigh against your own appetite for an agent that reads everything in a channel. Before you switch it on, decide what data it can see and which channels it lives in, the same way you would scope a new hire’s access. The shared-identity model is the real shift: this is less a chatbot you prompt and more a coworker your whole team hands work to, and is accountable for.