Anthropic Launches Claude Tag — An AI Coworker That Lives in Your Slack Channels
Summary: Anthropic's Claude Tag lets teams add @Claude as a Slack member that writes code, tracks down metrics, resolves support tickets, and proactively communicates — all from within shared channels. Available in beta for Enterprise and Team customers.
---
Anthropic just shipped what might be the most practical AI-agent product for engineering teams yet. Claude Tag turns Claude into a persistent Slack team member that hangs out in channels, listens to conversations, and jumps in when tagged.
It's not another chatbot you message privately. This is Claude as a coworker — it shares channel context, works on tasks asynchronously, and can even learn from your team's discussions over time.
How Claude Tag Works
The core loop is simple: drop @Claude into a Slack channel, grant it access to your tools and codebases, and start delegating. Tag it with a request in plain English — "find the root cause of the login timeout bug" or "pull last quarter's revenue by region" — and Claude breaks the task into stages, works through them with the tools it has, and posts results back to a thread.
Anthropic reports that 65% of its own product team's code is now created by an internal version of Claude Tag. That's a striking signal for a tool that's been dogfooded before public release.
What's Different From Claude Code or Slack Apps
Claude Tag isn't Claude Code bolted onto Slack. Several design decisions make it feel more like a teammate:
• Multiplayer by default. One Claude instance lives in a channel and interacts with everyone. Anyone can see what it's working on, pick up where someone left off, or assign follow-ups. This is very different from each person having their own private agent session.
• Ambient context. Claude follows along with channel conversations and builds up tacit knowledge. You don't need to re-explain your project architecture every time you tag it. (It does not access private channels unless explicitly granted.)
• Proactive push. When "ambient" mode is enabled, Claude surfaces relevant information without being asked — flagging stuck threads, following up on unresolved tasks, and sharing updates it thinks the team needs.
• Async execution. You can give Claude a task and walk away. It works through the problem on its own schedule, even scheduling multi-hour or multi-day tasks. Anthropic says teams now run "many Claudes in parallel," delegating work instead of doing it themselves.
Security and Admin Controls
Claude Tag is designed for organizations that need guardrails. Administrators define which tools and data sources Claude can access, in which channels. The access model works like separate Claude identities: a Claude set up for sales work won't pass its memories to one set up for engineering. Token spend limits can be set per organization and per channel, and every action is logged with the requester's identity.
For enterprise buyers, this level of granularity matters. You're giving an AI access to internal codebases, customer data, and proprietary tools — the ability to scope that access tightly is table stakes.
Pricing and Availability
Claude Tag runs on Opus 4.8 and is available today in beta for Claude Enterprise and Team customers. It replaces the existing "Claude in Slack" app; existing users have 30 days to migrate. Anthropic is offering introductory launch credits for eligible organizations.
The beta is Slack-first, but Anthropic says it plans to expand to other platforms so teams can tag @Claude wherever they work.
Should You Try It?
If your team already uses Claude, Claude Tag is worth a beta trial. The multiplayer channel model solves a real problem — private agent sessions don't scale for team workflows. Having one shared AI instance that learns from your team's context and can be delegated to by anyone in the channel is genuinely different from what's been available.
For teams using Copilot or other coding assistants, the Slack integration layer is the differentiator. Claude Tag brings AI-assisted development into your existing communication flow rather than requiring developers to context-switch to a separate tool.
The 65% internal adoption stat at Anthropic suggests this isn't an experimental prototype — it's a tool the company itself relies on daily.
Sources:
• Anthropic — Introducing Claude Tag
• TechCrunch — Anthropic's Claude Tag