What Is Claude Tag? A Guide to Anthropic's Always-On AI Teammate for Slack

What Claude Tag Is — An Always-On AI Teammate in Slack

Claude Tag is a feature announced by the AI company Anthropic on June 23, 2026, that lets Claude work as a member of your team inside Slack. Until now, Claude was mainly used by typing a question into a chat screen and receiving an answer. Claude Tag changes that relationship and treats Claude as something more like "a colleague who is always in the channel."

Once it is set up, an administrator grants Claude access to specific channels and connects it to internal tools and data. After that, the members in a channel can ask Claude to handle things as part of the normal flow of conversation. The essence of Claude Tag is that it repositions AI from "a tool you call up and use" to "a member who belongs to the team and does work."

How to Delegate Tasks with an @mention

Using it is no different from asking a colleague for a favor in Slack. You mention @Claude in a channel and write what you want done in plain words.

Claude then breaks the request into several stages and works through them in order, using the connected tools. When the work is finished, it posts the result back to a Slack thread.

Because the progress also stays in the thread, it is easy for other members to check on how things are going or to pick the work up where someone else left off.

"Ambient" Mode Lets It Act on Its Own

Claude Tag has an "ambient" mode in which it acts on its own even when it has not been called. This is the part where its character differs most from a conventional chatbot.

When this mode is enabled, Claude finds and shares relevant information from the channels it is responsible for and the tools it is connected to, and on its own chases down threads or tasks that have gone quiet without a reply. Not just waiting for instructions but reading the situation and acting ahead of time is the behavior you would expect from an "always-on teammate."

The "Multiplayer" Model Shared by Everyone in a Channel

Rather than a separate AI responding to each user, Claude Tag has one Claude per channel that everyone shares.

Because of this, the work someone asks Claude to do is visible to everyone in the channel, and another person can pick up a conversation that someone else had partly progressed. Since Claude accumulates context as it follows the channel's conversation, there is less need to explain the background from scratch every time.

How It Differs from a Conventional Slack Bot, and Supported Plans

In that it is "an AI that runs in Slack," Claude Tag looks the same as an existing chatbot, but its design philosophy is very different. Below we organize the differences in a table, then check which plans can use it and how to migrate from the existing app.

What Is Different from a Typical AI Bot

Until now, AI bots in Slack were mainly "responders" that returned an answer one-on-one when a user called them. Claude Tag is designed as a "teammate" that belongs to a channel, remembers context, and acts on its own.

To put the difference in one line, where a conventional bot was "a window that answers questions," Claude Tag is "a colleague you can hand work to."

Supported Plans and Model (Enterprise, Team, Opus 4.8)

Claude Tag is not a feature anyone can use right away; it begins as a beta on the business plans.

In other words, it is available to Claude Enterprise and Claude Team subscribers, and the free plan and the individual Pro plan are not included. The model it runs on is Anthropic's latest generation, Claude Opus 4.8, and the ability to work through complex procedures across multiple stages is the foundation of this feature.

Migrating from the "Claude in Slack" App

Anthropic has offered a Slack integration app called "Claude in Slack" until now, but Claude Tag is positioned to replace it.

Organizations already using the old app can switch to the new Claude Tag by having an administrator opt in (agree to migrate) within 30 days. For a new deployment as well, the flow starts with an administrator configuring the settings in the workspace.

What Claude Tag Can Do and What to Check Before Adopting It

Claude Tag goes beyond simple question-and-answer into actual work in coordination with tools. At the same time, precisely because it accesses your internal conversations and data, you need to design permissions and the handling of information carefully when you adopt it.

Running Multi-Step Tasks Autonomously

Claude Tag's strength is that it handles multiple steps on its own from a single request. It breaks the request it receives into stages, uses the connected tools in order, and returns the result summarized in a thread. Within Anthropic, this mechanism is already said to be a primary way work gets done. On the product team, 65% of the code is said to be written by an internal version of Claude Tag, which suggests it is built into the core of their work.

Admins Decide the Scope of Channels, Tools, and Memory

Just as important as the convenience is the design that lets administrators control how far information reaches.

What Claude remembers and the information it handles stay within the scope of the channels the administrators define. The idea is to prepare separate Claudes for different uses—for example, you can separate what a Claude in a sales channel and a Claude in a development channel can see. In an environment that handles confidential internal information, this scoping is the prerequisite for using it safely.

Points to Check Before Adoption

When considering adoption, it helps to confirm a few prerequisites first. To begin with, at this point it is a beta feature for Enterprise and Team, so the specifications may change going forward. Next, it is safer to start with a minimal scope of channels and tools you allow access to, and expand it as you operate. If you enable ambient mode, letting your team know in advance that Claude will speak up on its own helps avoid confusion on the ground. For details on pricing and usage limits, since they differ by contracted plan, check the latest official information.

Conclusion — Toward an Era Where AI Teammates Are the Norm

Claude Tag pushes AI from "a tool you use" to "a member who joins the team." Handing off work with an @mention, having it act ahead of time in ambient mode, and sharing context per channel—these three are the decisive differences from a conventional chatbot.

The fact that Anthropic itself produces much of its internal code through this mechanism shows that AI teammates are moving from the experimental stage into practical use. The flow in which AI autonomously operates tools to produce results is spreading beyond software development into other kinds of work as well. A realistic way to start is to try it in a limited set of channels first, then confirm its effect while managing the scope of information.

You can check the details of the feature and the latest availability on Anthropic's official announcement page.

By the way, when reading an official announcement from overseas, converting the page into Markdown first keeps the structure of headings and lists intact and makes it easier to grasp.

Free ToolURL to Markdown ConverterConvert any public web page URL to Markdown. Preserves headings, tables, lists, and links — perfect for LLM and RAG preprocessing, research notes, and archiving web articles.Try it now →