Anthropic Launches Claude Tag, Bringing an AI Teammate Directly Into Slack

Anthropic has launched Claude Tag, a new Slack-based AI teammate that lets workers mention @Claude in a channel and assign it real tasks without leaving Slack.

The feature is available for Claude Team and Enterprise customers and is designed for shared work inside Slack channels. According to Anthropic’s Claude Tag documentation, anyone in a channel can tag Claude into a problem, track its progress in a thread, and see the result where the work started.

Reuters reported that Anthropic launched Claude Tag on June 23, 2026, as a research preview for Claude Enterprise and Team customers, with plans to expand the capability to other platforms.

Claude Tag turns Slack channels into shared AI workspaces

Claude Tag works differently from a normal chatbot session. Instead of one person asking a private assistant for an answer, a team can bring Claude into a Slack thread and let everyone in that channel follow, redirect, or continue the work.

Anthropic says Claude Tag can reproduce bugs, summarize decisions, build reports, turn threads into documents, answer data questions, and open pull requests. For longer tasks, Claude posts a checklist in the thread and updates it as work progresses.

The company’s Claude Tag workflow guide says each Slack thread runs as a working session in a sandbox hosted by Anthropic. Claude reads the thread, uses the channel’s approved tools, works through the task, and then posts the result back into the same conversation.

Feature	What it does
Channel tagging	Users mention @Claude in a Slack channel or thread to start a task.
Shared work	Everyone in the channel can see the request, steer the work, and review the result.
Agent identity	Claude acts through service accounts and access set by an organization owner.
Thread-based sessions	Each task runs in its own Slack thread and sandbox.
Channel memory	Claude can retain useful context from channels, subject to admin controls.

Claude Tag uses channel-based access, not personal access

One of the biggest changes is how access works. In a Slack channel, Claude uses the tools and credentials that an organization owner has approved for that channel. It does not simply inherit the permissions of the person who asked the question.

That matters for enterprise use. A sales channel, engineering channel, and legal channel can have different connected systems, instructions, repositories, or credentials. Anthropic says access follows the channel, not the user.

The Claude Tag security documentation says Claude acts under its own service accounts in channels, while direct messages run on the user’s own Claude account and personal connectors.

Admins control tools, data, channels, and spending

Claude Tag requires setup by an Owner or Primary Owner in the Claude organization. The setup process includes pairing a Slack workspace, connecting approved tools, setting spending limits, and launching the feature for selected channels or workspaces.

Anthropic’s Claude Tag setup guide says installation of the Claude app in Slack is only the prerequisite. The main setup step gives Claude its own identity, access bundle, connected tools, and approved operating scope.

Admins can set a monthly spend cap for Claude Tag work. Anthropic lists preset limits such as $100, $250, $500, and $1,000, along with options for custom limits or unlimited usage where supported.

• Owners can choose where Claude Tag works in Slack.
• Access can be scoped to an organization, workspace, or private channel.
• Claude can use service-account credentials instead of personal logins.
• Admins can set organization-wide and per-channel spending limits.
• Audit records show tasks, network calls, and actions taken by Claude.

Security controls focus on isolation and allowed destinations

Because Claude Tag can act on tasks and use external tools, Anthropic is emphasizing governance and security controls. The company says each Slack thread runs in a sandbox and outbound traffic is blocked by default unless an owner allows the destination.

The security and data handling page says Claude Tag sandboxes hold no credentials. Instead, outbound requests pass through an Agent Proxy, which injects credentials only when a request matches an approved rule.

This design gives enterprise teams more control over where Claude can send requests and which connected systems it can reach. It also means administrators need to configure access carefully before allowing Claude Tag into sensitive channels.

How Claude Tag differs from Claude Code and Cowork

Claude Tag is not just Claude Code inside Slack, although it uses some of the same underlying agentic work patterns. Claude Code focuses on developer tasks in a terminal, IDE, or web coding environment, while Claude Tag focuses on shared team work in Slack channels.

Anthropic’s Claude Tag explainer says Cowork and Claude Code usually use a person’s own connectors or local credentials, while Claude Tag channel sessions use team-level service-account access that an admin configured.

That distinction makes Claude Tag more suitable for tasks that multiple people need to watch or steer, such as launch preparation, bug triage, incident follow-up, sales research, or status reporting.

Product	Main workspace	Typical use
Claude Tag	Slack channels	Shared team tasks visible to a channel
Claude Code	Terminal, IDE, web coding environment	Coding, debugging, pull requests, codebase work
Cowork	Claude chat	Personal work using the user’s own tools and context

The old Claude in Slack experience is changing

Anthropic is also moving its Slack integration toward the new Claude Tag experience. The company’s support documentation says Claude in Slack will switch to Claude Tag on August 3, 2026.

The existing Slack experience already let users message Claude, mention it in threads, and route some coding tasks to Claude Code. Claude Tag expands that idea by giving Claude a persistent team-facing role, channel-level memory, and admin-governed tool access.

Introducing Claude Tag, a new way for teams to work with Claude.

In Slack, Claude joins as a team member with access to the channels and tools you choose. Tag Claude in and delegate tasks to it while you focus on other work. pic.twitter.com/R2C6A5Kcye

— Claude (@claudeai) June 23, 2026

For teams that already use Slack heavily, this could reduce the need to move between a chat window, a ticketing system, a repository, a document editor, and an analytics dashboard for routine tasks.

Anthropic is pushing deeper into enterprise AI collaboration

The launch also shows Anthropic’s continued push into workplace AI. Reuters said Claude Tag deepens Anthropic’s move into the enterprise market, where AI companies are competing to become part of daily business software.

Cat Wu, Anthropic’s head of product for Claude Code, told Reuters that the ability to tag Claude like a coworker gives the feature a different form factor from earlier AI tools.

Anthropic says Claude Tag will work in Slack today and expand to other platforms later. For now, its success will likely depend on whether enterprise teams trust the access model, whether admins can govern it clearly, and whether users treat Claude as a shared teammate rather than a private chatbot.

What teams should check before enabling Claude Tag

Organizations should start with a small pilot channel before deploying Claude Tag broadly. That lets admins test tool access, spending limits, memory behavior, audit logs, and user expectations in a controlled setting.

The setup documentation recommends testing the configuration after pairing Slack and connecting access. A simple channel summary task can confirm that the app is listening and that the workspace pairing works before teams connect more sensitive tools.

The practical rule is simple: Claude Tag can be useful when teams want shared AI help inside Slack, but organizations should treat it like any other service account with access to company systems. Give it only the tools it needs, monitor what it does, and review channel memory regularly.

FAQ