Claude Tag Beta Release

Claude Tag Beta Release

Claude Tag enables collaborative AI delegation within Slack

Claude Tag is a new integration for Claude Enterprise and Team customers that allows teams to add Claude as a member of their Slack channels. By tagging @Claude, users can delegate complex tasks, connect the model to specific tools and codebases, and collaborate asynchronously with an AI that maintains context across a shared channel.

Core Capabilities of @Claude

Claude Tag evolves the functionality of Claude Code by moving from a single-user chat interface to a "multiplayer" environment. Key features include:

  • Multiplayer Collaboration: Unlike private chats, a single Claude instance interacts with everyone in a Slack channel. This allows team members to monitor the AI's progress and pick up conversations where others left off.
  • Persistent Organizational Memory: Claude builds context by remembering relevant information from the channels it is permitted to access. This reduces the need for users to repeatedly provide the same background information.
  • Proactive "Ambient" Behavior: When enabled, Claude can take initiative by flagging relevant information across connected tools and channels or following up on unresolved threads without being prompted.
  • Asynchronous Execution: Claude can plan and execute tasks over hours or days, allowing users to delegate work and return later to review the results in a Slack thread.

Administrative Control and Security

To manage data privacy and security, system administrators control Claude Tag's access at the channel level. This allows for the creation of scoped identities:

  • Scoped Access: Administrators specify which tools and data sources are available to Claude in specific channels. For example, an engineering-scoped Claude will not have access to sales data or tools, and its memories remain isolated from other channel identities.
  • Spend Management: Administrators can set token spend limits for the entire organization or for individual channels to control costs.
  • Auditability: A complete log of Claude's actions and the users who requested those tasks is available for review.

Deployment and Technical Specifications

Claude Tag is currently in beta and replaces the previous "Claude in Slack" app. It operates using the Opus 4.8 model. To implement Claude Tag, administrators must pair it with their Slack workspace, grant tool access, set spending limits, and verify functionality in a private channel.

Community Insights and Critical Perspectives

While Anthropic reports that 65% of its own product team's code is created by an internal version of Claude Tag, the announcement has sparked debate among technical users regarding security and utility:

Security and Compliance Concerns Some users have questioned the feasibility of enterprise security in this model, noting that AI permissions may not align perfectly with Slack channel memberships. One commenter noted:

"Claude needs to inherit permissions from somewhere, and those permissions will never align with the members of a slack channel."

Data Privacy Risks There are significant concerns regarding the risks of giving a third-party AI provider a "direct pipe" into sensitive internal discussions and business positioning.

Memory Reliability Users have reported issues with AI "learning" from incorrect or experimental data, leading to a "foundation of sand" where the model makes hidden, incorrect assumptions based on previous interactions.

Pricing and Billing Community members have pointed out that Claude Tag utilizes token-based pricing, which is separate from existing Claude subscriptions, potentially leading to unexpected costs if spending limits are not manually configured.

Sources