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 users on Claude Enterprise and Claude Team. Teams can tag @Claude in channels, assign work, and get results back in the thread. The key shift is persistent channel memory: each channel gets a shared Claude identity that can follow work over time and let teammates continue from prior context. Admins define which channels, tools, and information Claude can access. Anthropic frames this as a way to keep legal, engineering, or other workspaces scoped.
Nauti's Take
Claude Tag is framed as a productivity feature, but the strategic move is much bigger: Anthropic wants to sit where work actually happens. That matters because agents without context quickly become generic assistants.
The real colleague language is PR-polished and deserves skepticism. Companies will need strict governance, clear data boundaries, and solid offboarding rules, or useful memory can turn into a hard-to-audit shadow workflow.
Briefingshow
This is less about another Slack bot and more about turning company context into machine-readable infrastructure. Whoever understands day-to-day chat can infer decisions, ownership, friction, and informal knowledge. That context is becoming the real enterprise AI moat, not just the model behind it.
Sources