Anthropic Launches Claude Tag: @Claude Becomes a Slack Team Member

On June 23, Anthropic officially released Claude Tag — a new integration that lets Claude reside directly in Slack as a "team member." Anthropic chose Slack first because, by its own account, 65% of its product team's code is written with Claude Tag, making Slack their primary collaboration hub.

Source: https://www.anthropic.com/news/introducing-claude-tag

Core Features

• @Claude trigger: Anyone in an authorized channel can @Claude to delegate a task. Claude breaks it into a phased plan, executes it, and reports back in the thread.
• Multiplayer: Claude is a *single* instance per channel — everyone sees the same working context and memory. If colleague A delegates a task, colleague B can follow up on the same conversation.
• Cross-session learning: Claude continuously follows channel conversations and builds project context automatically, so users no longer re-explain background each time.
• Ambient (proactive) behavior: Claude surfaces "potentially relevant" information from authorized channels/tools and proactively follows up on open threads.
• Asynchronous tasks: Schedule a task for a future time; the agent plans and runs it on its own.
• Private DM mode: Users can chat with Claude privately, where it works with the user's own authorized toolset.

Model and Availability

• Model: Claude Opus 4.8.
• Rollout: Slack first; beta open now to Claude Enterprise and Team customers.
• Permissions: Admins can configure tools, data access, and monthly token budgets at per-channel granularity, with full audit logs. The existing Claude in Slack app is auto-replaced within 30 days, with enterprise onboarding credits.
• Direction: Anthropic plans to expand to "more work scenarios" — "anywhere teams work."

Analysis: From IDE Paradigm to IM Paradigm

Claude Tag superficially means "putting Claude in Slack," but substantively it moves AI coding assistants from the IDE paradigm to the IM paradigm:

• IDE paradigm = 1 developer × 1 agent × 1 context window
• IM paradigm = N developers × 1 agent × shared memory × async collaboration

When every engineer, PM, SRE, and support person can @Claude against a shared memory, Claude stops being "my coding assistant" and becomes "the team's engineering collaborator" — isomorphic to the role Slack channels already play in enterprises.

The 65% Figure

Anthropic disclosed that 65% of its product team's code is written by Claude Tag — meaning its product code is largely AI-written under human supervision. Industry context:

• Early 2024: GitHub Copilot targeted "AI writing 30% of code"
• 2025: Cursor / Devin positioned as "AI engineers"
• June 2026: Anthropic's own product team at 65%

Publishing this number implies Anthropic believes it is repeatable and wants enterprise customers to treat it as a KPI benchmark.

Ambient Proactivity as the Next Step

Ambient mode examples: pulling related issues when a bug is discussed, fetching traces during incident response, pinging owners about stale support tickets. This "proactively discover → proactively push to the right person" capability is a 2026 Q2 frontier shared with xAI's Grok Build /goal and Google Labs' proactivity research — together signaling the formation of proactive agents as a product category.

Enterprise-Grade Design

• Per-channel tool/data authorization — sales and engineering channels run fully isolated Claude instances with non-interlinked memory.
• Token budgets capped at organization and channel levels.
• Complete audit logs: who asked what, which tools were called, tokens consumed.
• One-click migration from the old Claude in Slack app.

This amounts to an enterprise-ready form factor for AI collaboration assistants, addressing concerns about AI bots misfiring in IM.

Relationship to Claude Code / Cursor / Codex

Claude Tag doesn't replace Claude Code; it fills the gap in multi-user collaboration with IM as the entry point:

• Claude Code: strong reasoning + long context in the IDE
• Cursor: multi-model + high-frequency operations in the IDE
• Codex CLI: command line + cloud sandbox
• Claude Tag: IM collaborator + cross-session memory + proactive push

Anthropic now holds both the IDE (Claude Code) and IM (Claude Tag) positions — a dual layout hard for competitors to replicate short-term.

Why It Matters

• For developers: No more re-explaining project context — context accumulated in team channels carries into your next conversation.
• For enterprise IT: Channel-level permission isolation, org-level token budgets, and audit logs set a de facto standard that Teams, Lark, DingTalk, and WeCom equivalents will be measured against.
• For Anthropic: "65% of product code written by Claude Tag" is the hardest-hitting data point in its valuation narrative — its own products are built with its own AI.
• For the AI collaboration space: IM platforms shift from "AI answering questions" to "AI as active collaborator."
• For competitors: OpenAI's enterprise IM presence remains at the Q&A layer; Cursor has no IM form factor; xAI Grok Build is CLI-based — Claude Tag is the first product to land the "AI team member" concept.

Risks and Open Questions

• How is 65% measured? Anthropic hasn't disclosed the denominator (LOC, PRs, commits) or numerator (auto-accepted AI output vs. AI-assisted human work). Without methodology, 65% is both a claim and a promise.
• Cross-channel memory boundaries: Learning from other channels is a double-edged sword. Claude reportedly won't read private channels, but how are "seemingly public yet sensitive" channels (HR, legal, payroll) handled? An operational gray zone.
• Ambient interruption threshold: Push frequency, relevance, and timing decide UX — too much is noise, too little is useless. Anthropic hasn't published the user opt-out rate for ambient mode.
• Slack lock-in: Slack-first rollout may slow penetration in markets dominated by Teams, Lark, DingTalk, or WeCom.
• Competitive response: Cursor + Slack or Codex + Teams equivalents could appear within 30–90 days. Claude Tag's lead window depends on Anthropic refining "team memory + proactive push" into a cross-platform product within six months.

TL;DR

Claude Tag isn't just "AI in Slack" — it turns the "team-member-grade AI collaborator" from concept into product, and uses the 65% figure to push "AI-written product code" from marketing slogan toward verifiable KPI.

Sources: https://www.anthropic.com/news/introducing-claude-tag · https://claude.com/blog/agent-identity-access-model · https://www.claude.com/product/tag