Anthropic’s latest evolution transforms AI from a solitary tool into a proactive, multiplayer teammate that writes code, crunches data, and keeps your projects moving forward.
- Multiplayer Collaboration: Claude acts as a shared team member within Slack channels, allowing anyone to view its progress, delegate complex tasks, and seamlessly pick up where a teammate left off.
- Proactive Autonomy: With “ambient” capabilities enabled, Claude independently flags relevant information, follows up on quiet threads, and schedules autonomous tasks over hours or days.
- Enterprise-Grade Control: Administrators maintain granular oversight, tightly scoping Claude’s memory and tool access to specific channels while tracking token limits and request logs to ensure strict data security.
For years, interacting with artificial intelligence has been largely a solitary pursuit: a user typing a prompt into an isolated chat window and waiting for a response. Today, the paradigm of human-AI collaboration takes a massive leap forward. Introducing Claude Tag, an evolution of Claude Code that reimagines how organizations get things done. By integrating directly into collaborative environments—starting with Slack—Claude transforms from a reactive digital assistant into a proactive, embedded team member.
At Anthropic, tagging @Claude has fundamentally shifted daily operations. Today, a staggering 65% of the product team’s code is created by the internal version of Claude Tag. But this evolution is no longer confined to engineering. Teams are routinely tagging Claude to chase down vital product metrics, resolve complex support tickets, and run deep data analysis to find the root cause of tricky bugs. Now, this transformative capability is available in beta for Claude Enterprise and Team customers.
Delegating the Heavy Lifting
If you are familiar with Claude Code or Cowork, the foundation of Claude Tag will feel remarkably intuitive, yet exponentially more powerful. When you tag @Claude with a request in plain English, it doesn’t just generate a text response. It breaks the assignment down into actionable stages, leverages the specific tools and data sources it has been granted access to, and systematically works through the problem. Once the task is completed—whether that means writing a pull request, merging code, or analyzing a dataset—Claude responds directly in the Slack thread with its results.
This unlocks a radically new, asynchronous workflow. You can tag Claude with a demanding task and immediately pivot to other high-level priorities. Claude can even schedule tasks for itself, autonomously pursuing a multi-step project over hours or days. At Anthropic, it is now common practice to aggressively parallelize work, delegating distinct assignments to multiple Claudes simultaneously. Should you need a private sounding board, you can also send Claude a direct message; it will respond privately using the personal tools and connectors you have authorized.
The Multiplayer Advantage
What truly sets Claude Tag apart is its collaborative nature. It brings four distinct advantages to the modern team dynamic:
- It is Multiplayer: Within a designated Slack channel, there is a single, unified Claude interacting with the entire team. Because the workflow is public, any teammate can see what the AI is building and pick up the conversation exactly where the last person left off. It bridges the gap between individual AI usage and true team collaboration.
- It Learns Context Over Time: Nobody likes repeating themselves. As Claude follows the conversation in a channel, it steadily builds context about your team’s ongoing work. By pulling tacit knowledge from permitted Slack channels and connected data sources—while strictly refraining from reporting out of private channels—Claude ensures you rarely have to explain projects from scratch.
- It Takes Initiative: Through its unique “ambient” behavior, Claude sheds its passive nature. If turned on, Claude proactively nudges the team. It will flag highly relevant information from across its connected tools and channels, and even take the initiative to follow up on project threads that have mysteriously gone quiet.
- It Works Asynchronously: The true power of delegation lies in unblocking human potential. While you step into a meeting or focus on deep work, Claude is running data, drafting code, and preparing your next steps.
Built for the Enterprise: Security and Scoping
Transforming an AI into a teammate requires absolute trust, which is why Claude Tag was engineered from the ground up for complex organizations. Access to sensitive company data and task-specific tools is rigorously controlled by system administrators.
To bring Claude Tag online, administrators explicitly define what information and tools the model can access, and precisely in which channels that access is granted. This essentially creates bespoke “identities” for Claude across the company. For instance, a Claude configured for a sales team operates within a strictly scoped memory and toolset; it will never pass its context to an engineering channel, nor will it inadvertently grant engineers access to sensitive sales pipelines.
IT leaders maintain complete visibility. Administrators can configure token spending limits at both the organizational and individual channel levels. A comprehensive audit log tracks everything @Claude executes, including exactly who requested the task, ensuring compliance and peace of mind.
