Anthropic’s New Claude Tag Acts as an Agentic AI Coworker Within Slack

Anthropic has unveiled a new AI agent called Claude Tag, which can join Slack as a team member to learn a company's workflows and context over time. Teams can assign tasks to the agent, ask it questions, or request analysis based on ongoing discussions. Anything in an assigned channel is fair game because Claude Tag is always on.

From each Slack channel and connected tool, Claude Tag builds a memory of company projects, naming conventions, and past decisions. Over time, it can do things like answer questions about older conversations, find sales numbers, or pull in incident reports without needing to manually search through chats. Organizations can even create separate "Claudes" for different purposes and tag them as they're needed.

"Think of it as creating separate Claude identities for different uses: everything, including its memories, will stay scoped to the channels defined by the administrators," Anthropic's announcement reads. "For example, a model set up for sales work won't pass on memories to one set up for engineering; nor will it give engineers access to any sales data or tools."

The system also includes an ambient mode that lets Claude proactively step in when it detects likely gaps or stalled threads. In this mode, Claude can do things like remind teams about unresolved issues or highlight old discussions relevant to the current topic.

Admins can choose which channels Claude can read and which tools it can connect to. While it can be added to private channels, Claude Tag can't report information from those channels outside them.

The agent is currently available in beta for Claude Team and Enterprise customers.