Anthropic has introduced Claude Tag, a new way for teams to work with Claude[1]. It starts on Slack, where Claude can be welcomed into channels as a member of the team. Once you grant access to selected channels, tools, data, and even codebases, anyone can simply tag @Claude in a message and hand off a task[1]. It is available in beta starting today for Claude Enterprise and Team customers[1].
Claude Tag Joins Slack as a Member of the Team
What sets Claude Tag apart is that it treats Claude less like a chat partner and more like a colleague who lives in the channel. When you tag @Claude and describe a request in simple terms, Claude breaks the task down into stages, works through them in turn using the tools it has access to, and replies in a Slack thread with what it has created[1]. If you have used Claude Code or Cowork before, the experience should feel familiar[1].
Anthropic positions Claude Tag as an evolution of Claude Code, making the model more proactive and extending it from solo work to collaboration across a full team[1]. Inside the company, tagging @Claude has already become one of the main ways work gets done, and 65 percent of the product team's code is now created by Anthropic's internal version of Claude Tag[1]. The pattern has spread well beyond engineering to chasing down product metrics and data, working through support tickets, and helping find the root cause of tricky bugs[1].
Slack was chosen as the first home because it is a natural place for collaboration between teams and AI, and because much of Anthropic's own day-to-day work already happens there[1]. The company says it intends to expand availability to the many other places where teams work[1].
Four Advantages Unique to @Claude
Claude Tag offers advantages that differ from a traditional one-on-one chat[1].
First, it is multiplayer. Within a given Slack channel, there is a single Claude that interacts with everyone, so anyone can see what it is working on and pick up a conversation from where the last person left off[1]. This is described as much closer to working alongside a teammate than to a single chat or a single task[1].
Second, it learns over time. As Claude follows along with a channel, it builds up context about the work, so users do not need to explain things from scratch every time[1]. With permission, Claude can also automatically learn from other Slack channels and data sources, though it does not report out from private channels[1].
Third, it takes initiative. When "ambient" behavior is enabled, Claude proactively surfaces information it judges you may need to know[1]. It flags relevant information from the channels it is in and the tools it is connected to, and follows up on threads or tasks that have gone quiet without being resolved[1].
Fourth, it works asynchronously. Set Claude a task and you can focus on other priorities while it works. Claude can also schedule tasks for itself and pursue a project autonomously over hours or days[1]. Anthropic says this asynchronous quality has been especially helpful, and that the team now spends much more of its time delegating tasks to many Claudes in parallel[1]. You can also send Claude a direct message, and it will respond privately using the personal tools and connectors you have set up[1].
Administrators Control Access and Spending in Detail
Claude Tag is designed with teams and organizations in mind, allowing tight control over access to sensitive data and task-specific tools[1]. System administrators specify which tools and information the model can reach, and in which channels[1].
In practice, it is like creating separate Claude identities for different uses. Everything, including its memories, stays scoped to the channels the administrators define[1]. A model set up for sales work, for example, will not pass memories on to one set up for engineering, and engineers will not gain access to sales data or tools[1]. Once permissions are set, anyone can begin tagging right away[1].
Administrators can set token spending limits both for the organization and for individual channels, and can view a log of everything @Claude has done along with who requested each task[1]. Getting up and running takes four steps: pairing Claude Tag with the Slack workspace, granting access to tools, setting a monthly spending limit, and testing it in a private channel[1].
Migrating from the Existing App and Beta Availability
Claude Tag replaces the existing Claude in Slack app. Administrators can migrate by opting in within 30 days[1]. Several outlets report that the older Claude in Slack app will shut down on August 3, 2026[2]. Eligible Enterprise and Team organizations receive an introductory launch credit so the whole company can try it out[1].
Claude Tag works with Opus 4.8[1]. If you are a Claude Enterprise or Team customer, you can use the beta starting today[1].
Summary
Claude Tag is an attempt to lift Claude from a chat partner to a colleague who lives in the channel. Multiplayer collaboration, continuous learning of context, proactive follow-ups, and long-running asynchronous work are its four strengths, and the fine-grained administrator control over access, memory, and spending sits at the core of a design clearly aimed at enterprise use. The figure that 65 percent of Anthropic's own product code now comes from its internal version underscores how seriously the company takes the approach. That said, availability is currently limited to a Slack-based beta for Enterprise and Team customers, and how safely and usefully an autonomously acting AI can be operated will be settled only through real-world use within each organization.
