@Claude Is Now on Your Team: How Claude Tag Changes the Way Work Gets Done in Slack

What You'll Learn

• What Claude Tag does and how it differs from Claude Code or Cowork
• The 4-step setup process for Claude Enterprise and Team accounts
• How to create separate Claude identities for different teams (sales vs. engineering)
• Three specific workflows that save 30 or more minutes per week
• How Anthropic uses it internally and what that tells you about deployment

On June 23, Anthropic shipped Claude Tag: a product that puts Claude inside your Slack workspace as a team member. Not a chatbot you ping in isolation. Not a sidebar tool you switch to. An actual @Claude that anyone on your team can tag in any channel, hand a task, and walk away from while Claude works through it.

The business case is immediate. Today, most AI productivity gains are individual. One person learns a prompt, saves 20 minutes, and the benefit stays with them. Claude Tag is different because the model operates at the team level. One tag. Everyone in the channel can see what Claude is doing, pick up the thread, and redirect it. The context builds across conversations, not within a single chat window.

Anthropic says 65% of their own product team's code is now written by their internal version of Claude Tag. That number is worth pausing on. This is not a beta experiment. They use it on production work. And now it's available to Enterprise and Team customers in beta, today.

This is a PromptHacker Premium Article

The full breakdown, setup guide, and team workflow templates are available to paid subscribers.

What Claude Tag Actually Does

Claude Tag starts on Slack because Slack is where most collaborative work already happens. An admin pairs Claude to the workspace, grants it access to specific channels and tools (documents, codebases, databases), sets a monthly spend limit, and Claude is live. From that point on, any team member types @Claude followed by a task and Claude goes to work in a thread.

Three capabilities distinguish it from a chatbot integration:

Multiplayer by design. One @Claude per channel. Anyone can see the work in progress, add context, or redirect it. If a teammate tagged Claude to draft a competitive summary and you have a correction, tag Claude in the same thread and it incorporates it. This is closer to managing a junior analyst than using an AI tool.

Memory that builds over time. Claude follows the channel. It reads what passes through and retains relevant context. Next time someone asks it for a market summary, it already knows your company's positioning, the competitors you track, and the format your team prefers. You stop re-explaining the basics on every task.

Ambient mode. If enabled, Claude monitors channels and proactively flags things it thinks you should know. Open loops that haven't been resolved. Threads that went quiet on a deadline. News from a connected data source that contradicts a decision in progress. You do not have to remember to ask.

The 4-Step Setup (Under 20 Minutes)

Anthropic designed the onboarding for administrators, not developers. You need a Claude Enterprise or Team plan. The full process:

• Pair Claude Tag with your Slack workspace. Go to claude.ai admin settings and connect via OAuth. This authorizes Claude to join channels you specify.
• Give Claude access to your tools. You choose: Google Drive, GitHub, Salesforce, Notion, or any MCP-connected data source. Claude only reads what you point it at.
• Set a monthly spend limit. Set an org-level cap and per-channel sub-limits. Anthropic is offering a launch credit to eligible Enterprise and Team customers so you can run the first tests at no cost.
• Test in a private channel. Anthropic recommends confirming the setup works with a test channel before rolling out to the team. Tag @Claude with a small task, verify it has the right tool access, then expand.

Critically: you create separate Claude identities for different use cases. A Claude set up for your sales team has access to CRM data and deal notes. A Claude for engineering has access to the codebase and linear tickets. These do not share memory or data. The admin log shows every task, every request, every output.

Three Workflows Worth Setting Up This Week

1. The Async Research Channel

Create a #research-requests channel. Connect Claude to your document store and web search. Tag Claude with competitive or industry questions during the week. By Friday it has worked through them all and the answers are in the thread, cited and formatted. This replaces roughly 3 to 4 hours of individual research per week across a team of 5.

@Claude Research request: Summarize what Salesforce, HubSpot, and Pipedrive announced in the last 30 days. Focus on changes to AI features and pricing. Format as a three-column table with a "so what" row at the bottom for each.

2. The Support Triage Channel

Give Claude access to your help docs and ticket history. Tag it in a support channel and have it classify incoming tickets, draft first-response suggestions, and flag anything that matches a known issue. Anthropic reports using Claude Tag for this internally. At 50-plus tickets per week, Claude handles the triage work that currently takes 2 to 3 hours of staff time.

3. The Weekly Metrics Pull

Connect Claude to your analytics data or Google Sheets. Every Monday, Claude pulls the week's numbers, compares them to the prior week, flags the 3 biggest changes (positive or negative), and posts the summary to your ops channel before standup. The team shows up to the meeting with context already loaded.

What the Anthropic Adoption Numbers Tell You

The 65% code figure is the most important signal in this announcement. Anthropic is not telling you that Claude Tag is useful for developers. They are telling you that their own product team runs on it. That is a different claim entirely. It means the product has matured past demo stage and into production workflow.

More relevant to most PromptHacker readers: Anthropic says the pattern is spreading beyond engineering to product metrics, support tickets, and bug triage. Those are not developer tasks. They are operations tasks. Claude Tag is not a coding tool that happens to be in Slack. It is a general-purpose async worker that can be configured for any function with clear inputs and outputs.

The transition from individual AI productivity to team AI productivity is the next 12 months of enterprise AI adoption. Claude Tag is the first product built specifically for that transition.

Action Steps

• Check your Claude plan. Claude Tag is in beta for Enterprise and Team customers only. If you're on Pro or Max, this is a reason to evaluate the Team tier.
• Open admin settings and connect your Slack workspace. The pairing takes under 5 minutes via OAuth at claude.ai.
• Start with one channel. A private test channel. Give Claude one data source (Google Drive or web search). Run a real task. Verify the output before expanding.
• Set your spend limit before you go broad. The admin console lets you cap costs per channel. Set conservative limits first, then raise them as you calibrate usage.
• Build one workflow per function. Sales, ops, and support should each have separate Claude identities with tailored tool access and memory scope. Do not share one Claude across everything.