Anthropic's New Claude Tag Acts as a Virtual Coworker in Slack

Slack users can now summon an AI coworker with a single @ mention. Anthropic has launched Claude Tag, turning its flagship AI assistant into a virtual team member embedded directly inside Slack workspaces.

The feature — rolling out globally as a paid add-on — lets organizations add Claude as a user with access to specific channels and data. Employees can then @-mention Claude in conversations to request summaries, research, Q&A, brainstorming, or even code debugging. The move marks one of the deepest integrations of a major AI model into a mainstream enterprise communication tool.

Why now? Slack has been experimenting with AI features since late 2022, offering native summarization and search powered by OpenAI models. But enterprise customers — particularly those in regulated industries — have pressed for more control over data privacy and model safety. Anthropic, founded by former OpenAI researchers, has built its reputation on “constitutional AI” and rigorous safety guardrails. The Claude Tag integration aims to give companies the productivity gains of generative AI without sacrificing security or compliance.

The integration works through Slack’s existing app framework. Once an admin grants permission, Claude appears in the workspace directory alongside human coworkers. It can read and write messages in assigned channels, but — crucially — organizations decide which channels and data Claude can access. Anthropic says no customer data is used to train its models, and messages are encrypted in transit and at rest.

Early adopters include mid-size tech firms and professional services companies, according to Anthropic. Pricing is per-seat, added to existing Claude enterprise subscriptions. Slack’s parent company Salesforce has not commented on the integration, but the move puts Anthropic in direct competition with both OpenAI’s ChatGPT Enterprise plugin for Slack and Microsoft’s Copilot for Teams.

“What’s really interesting here is the access-control layer,” said Sarah Guo, founder of Conviction VC and a former AI investor at Greylock. “Enterprises don’t just want a chatbot; they want a tool that respects their permissions and data boundaries. Claude Tag does that by design.” Guo noted that the tag model — treating AI as a named team member — could become a template for how other AI assistants integrate into collaboration software.

Looking ahead, Anthropic plans to expand Claude Tag’s capabilities with “proactive suggestions” — where Claude volunteers insights or catches errors without being prompted. Slack itself is expected to deepen its own AI features, potentially blurring the line between native AI and third-party bots. For now, Claude Tag represents a pragmatic bet: workers already live in Slack, so let AI live there too.

For businesses, the calculus is clear: faster answers, less context-switching, and a single window into team knowledge. The open question is whether users will trust an AI that writes alongside them — and whether Anthropic’s safety-first approach will give it an edge in the increasingly crowded enterprise AI market.