Anthropic launches Claude Tag in Slack, bringing AI into group chats
Anthropic put Claude directly into Slack channels, letting coworkers tag an AI that can read threads, split tasks and surface updates across a team.

Anthropic pushed Claude into Slack group chats on June 23, giving anyone in a channel a way to summon the AI by tagging @Claude. The new Claude Tag is available in beta for Claude Enterprise and Claude Team customers. It can read conversations, break work into pieces, remember relevant channel details and flag updates when ambient behavior is turned on.
Claude can be granted access to selected channels, connected tools, data and codebases, which means the model can sit inside the same conversations where employees assign work, make decisions and decide who gets looped in. Cat Wu, Anthropic’s head of product, said the advantage is not only the model’s capabilities but that it can be tagged like another worker and can post alerts back into Slack, including when connected to Gmail to watch for important messages.
Slack has framed its platform as an “agentic operating system” since October 2025, and partners including Anthropic are building native agents directly into Slack. Slack’s marketplace listing lets users start a direct message with Claude or tag it in a thread without leaving the app, and Anthropic calls the new version a shared, multiplayer presence in a channel rather than a one-on-one chatbot.

Anthropic says 65% of its product team’s code is created by its internal version of Claude Tag, and it uses the tool to chase product metrics, work through support tickets and find bug root causes. Slack’s customer story credits Anthropic with $4.5 million in savings as it scaled, underscoring that Anthropic is both a seller and a showcase customer in the enterprise AI market.
Anthropic’s May 25 engineering post on containing its agentic products said higher-access deployments can make developers more productive, but also widen the blast radius as access expands. Conversations are not used to train Anthropic models by default, Anthropic retains inputs and outputs for up to two years, some trust-and-safety scores for up to seven years if a prompt is flagged, and users can delete conversations by disconnecting the Claude account from Slack. Data is processed in U.S. data centers by default, and Slack warns it does not have a business associate agreement with third-party app providers.
This article was produced by Prism’s automated news system from verified source data, official records, and press releases, then run through automated quality and moderation checks before publishing. The system is built and supervised by the people who set the standards it runs under. Read our full AI policy.
Know something we missed? Have a correction or additional information?
Submit a Tip

