Anthropic’s Claude Tag is learning your company, one Slack message at a time
TL;DR
Anthropic is launching Claude Tag in beta for Slack across Claude Enterprise and Claude Team, letting teams mention @Claude in channels, assign tasks, and get work back inside threads. The real shift is persistent context: one Claude identity can follow a channel, remember prior work, and, if permitted, pull facts from other channels in the company. Admins define which tools, information, and channels Claude can access. Anthropic says memories stay scoped, so a legal Claude does not leak context into engineering.
Nauti's Take
The product sounds useful, but the strategic message is bigger: Anthropic wants Claude to become company memory, not just a response box. The pitch is PR-heavy, especially when it frames the system as a real colleague.
In practice, this only works if admins can tightly scope context, audits are serious, and teams clearly know when Claude is watching, remembering, and acting.
Briefingshow
Claude Tag is less a Slack feature than a move for the enterprise knowledge layer. A tool that continuously follows work channels learns processes, ownership, loose ends, and informal decisions. That context base is what determines whether enterprise agents become useful operators or stay polished chatbots.
Sources