Anthropic Transforms Slack with Claude Tag, the Integrated AI

Anthropic Transforms Slack with Claude Tag, the Integrated AI

With Claude Tag, unveiled Tuesday evening, Anthropic offers a new way to integrate artificial intelligence directly into the heart of the collaborative platform Slack. It's a kind of virtual teammate that you tag, taking charge of tasks. Imagine delegating a complex task to the AI with a simple “@” in Slack in the coming days or weeks, while you focus on something else. This is the promise of Claude Tag, launched Tuesday, June 23, 2026, at the end of the day by Anthropic, initially in beta for its Enterprise and Team clients. Here, the American company, which must navigate the limitations of its Fable 5 and Mythos 5 models, demonstrates its intention to make Claude a true team member, capable of learning, anticipating, and working autonomously for hours or even days.

Claude Tag Joins Slack as a True Team Member

With Claude Tag, administrators invite Claude into the Slack channels of their choice, just like adding a new colleague to a discussion group. Once in place, any member of the channel can write to him by mentioning @Claude, a process identical to any mention on Slack or on social media or messaging platforms. Claude receives the request, breaks it down into subtasks, executes them in order, and then returns to post the results directly in the conversation thread, without anyone having to intervene in the meantime.

Of course, compared to other chatbot interfaces, Claude Tag stands out due to its collaborative nature. In a given channel, everyone shares the same Claude. Each person can see what he is working on, pick up the conversation where a colleague left off, or even advance a project as a team. Siloed exchanges may well belong to the past, with AI becoming a common resource.

A Direct Evolution of Claude Code

Anthropic sells the concept well, but the company also experiences it from the inside. Today, 65% of the code produced by its teams is generated using its internal version of Claude Tag. The tool is presented as a direct evolution of Claude Code, designed to go further by being more proactive and built to work with an entire team rather than a single user.

Claude Tag Learns, Anticipates, and Works While You Do Something Else

One of Claude Tag's qualities is its ability to remember the context of its channels. Over time, it accumulates relevant information without needing to re-explain everything for each task. Administrators can even grant it access to other channels and data sources to enrich its knowledge base. Private channels, however, remain out of reach.

Claude Tag does not just respond in a conversation when called upon. If you activate it in "ambient" mode, it monitors exchanges, gathers relevant information, and autonomously brings up unresolved topics. It can also receive long-term assignments and carry them out independently for hours or days. Meanwhile, the user can work on something else. Direct messages remain possible, and Claude responds privately, prioritizing the user's personal tools.

Availability and Governance

For the curious, Claude Tag is available today in beta, particularly for Enterprise and Team subscribers, powered by the Opus 4.8 model. It replaces the old Claude application in Slack, noting that administrators have 30 days to switch to the new version. To encourage this transition, Anthropic offers a launch credit to eligible organizations.

On the governance side, admins retain full control, with the ability to define which tools Claude can use, set a monthly budget of tokens to avoid surprises on the bill, and access a complete log of everything Claude has done, including the name of the person who initiated each action. The future of teamwork now has a @.