Anthropic is bringing a more persistent version of Claude into Slack. Claude Tag is being introduced in research preview as an always-on Claude that can sit inside a channel, follow the flow of work, and respond when people tag @Claude.
The feature is available through Slack for Claude Enterprise and Claude Team customers. It is designed to act less like a one-off chatbot and more like an AI teammate that can understand what a group has been discussing over time.
What Claude Tag Adds To Slack
Claude already has a presence in Slack through existing integrations. Users can direct-message @Claude, mention it in channels for help, or use Claude Code in Slack to send coding requests from a channel mention into a full coding session on the web.
Claude Tag builds on that foundation by adding persistent context and memory. Instead of treating each request as isolated, Claude can follow along with a channel and build a picture of the work happening there. If it receives permission, it can also gather facts from other places in the organization.
That changes the role of the assistant. A team no longer has to re-explain every project, decision, or thread each time it wants help. The idea is that Claude can become more useful because it has more of the surrounding workplace context.
A Shared Claude For Each Channel
One of the most important design choices is that a Slack channel can have a single Claude identity that everyone in that channel can use. That means the assistant’s work is visible to the group, and the next person can continue from the same context instead of starting over in a private conversation.
In practical terms, this makes Claude Tag a shared workspace object rather than an individual user’s personal assistant. If Claude has been working on something, others in the channel can see that activity and pick up where the previous interaction ended.
Anthropic is also putting access control at the center of the feature. System administrators decide which tools, information, and channels Claude can use. Each Claude identity remains limited to the channels the administrator defines.
That scoping matters because different teams handle different kinds of information. A Claude identity set up for legal work, for example, is not meant to carry memories into an engineering channel. The structure is intended to keep context useful without letting it spread everywhere by default.
How Tasks And Ambient Help Work
Claude Tag can be assigned work directly from Slack. When someone gives it a task, Claude breaks that work into stages and uses the tools it is allowed to access. It then responds in a Slack thread with what it has produced.
This public-thread model is part of the product’s appeal. The work does not disappear into a private chat window. It happens where the team can follow progress, review output, and continue the discussion.
The feature also includes an ambient mode. In that mode, Claude can proactively enter a chat, keep a team updated, surface relevant items from elsewhere in the organization, and follow up on threads or tasks that have gone quiet.
That is a different pattern from waiting for a prompt. Claude Tag is meant to participate in the rhythm of the channel, not only answer direct questions. The value depends on context: the more relevant channel activity it can see, the more informed its help can become.
Why Company Context Is The Bigger Prize
The larger theme is enterprise context. Anthropic is not only adding another Slack command; it is trying to make Claude more aware of how work actually happens inside a company.
That is becoming a major focus for AI tools aimed at businesses. Microsoft has Graph, expressed through Copilot and Work IQ. Snowflake and Databricks are positioning their platforms as back-end support that contains tacit organizational knowledge for agents to use. Glean is building an intelligence layer that understands company context and sits between the model and enterprise data.
Claude Tag fits into that same movement. The assistant is more useful when it can connect requests to the relevant projects, people, channels, and prior decisions. Slack is a natural place for that because much of the day-to-day coordination already happens there.
At the same time, the feature’s usefulness depends on careful boundaries. The source material emphasizes administrator-defined access and channel-level scoping. Those controls are central because company knowledge is valuable precisely because it is specific, internal, and often unevenly distributed.
What To Watch Next
Claude Tag is starting in research preview, so its rollout is limited to the customers and environment described by Anthropic. For Claude Enterprise and Claude Team users in Slack, the preview will show how well an AI teammate can operate when it has memory, shared visibility, and scoped access to workplace information.
The key question is whether teams find this kind of persistent assistant helpful in normal work. If it can reduce repeated explanations, keep threads from going stale, and complete assigned tasks in view of the group, Claude Tag could make Slack feel less like a stream of messages and more like a place where organizational memory is actively used.
For now, the direction is clear: Anthropic wants Claude to move from on-demand help toward ongoing participation in team workflows. Claude Tag is the company’s attempt to make that shift inside the communication tool many teams already use.