Anthropic just shipped Claude Tag (branded as @Claude), turning its flagship model into a full Slack team member. You @mention it in channels, assign tasks, and it works in the background while your team keeps moving. It’s available today in beta for Claude Enterprise and Team customers, and it runs on Opus 4.8.
This is not the old Claude in Slack app getting a refresh. It’s a fundamentally different product.
What actually changed
The previous Claude in Slack integration was basically a chatbot with a text window. You asked questions, it answered. Useful, but not much different from opening claude.ai in a browser tab.
Claude Tag is designed as a persistent team member. You grant it access to specific Slack channels, connect it to your tools and data sources through Anthropic’s Agent Identity model, and then anyone in the channel can tag @Claude with a task. It breaks down the request, works through it step by step using whatever tools it has access to, and posts the result back in a thread.
Anthropic’s announcement highlights four structural differences from the old approach:
Multiplayer. One Claude per channel interacts with everyone. Anyone in the channel can see what it’s working on and pick up where someone else left off. It’s collaborative, not a private chat session.
Learns over time. @Claude builds context by following along with the channels it’s in. You don’t have to re-explain your project every time. It can also pull context from other authorized channels and data sources. (It does not report from private channels unless explicitly granted access.)
Takes initiative. With “ambient” behavior enabled, Claude proactively surfaces updates, flags threads that went quiet, and tags you back when a decision is needed. It doesn’t wait to be asked.
Works asynchronously. You set a task, Claude runs it while you do other work. It can schedule work for itself and pursue multi-day projects autonomously. Anthropic says it now spends “much more of our time delegating tasks to many Claudes in parallel.”
You can also DM @Claude for private interactions using the personal tools and connectors you’ve set up.
The internal number that matters
Anthropic dropped a stat that puts this release in perspective: 65% of its product team’s code is now created by the internal version of Claude Tag. That is not “AI assists coding.” That is “the majority of production code is generated and handled by the AI team member in Slack.”
More telling: the same pattern has spread beyond engineering. Anthropic says teams are tagging Claude to chase down product metrics, work through support tickets, and find root causes of bugs. This is the internal dogfooding result Anthropic clearly wants to sell to its enterprise customers.
How admin controls work
Anthropic is positioning this for serious organizations, and the admin layer reflects it. System administrators specify exactly which tools and data the model has access to, in which channels. Think of it as creating separate Claude identities for different use cases. A Claude set up for sales work won’t pass context to one set up for engineering. Memories stay scoped to the channels the admin defines.
Admins can also set token spend limits at the organizational and per-channel level, and view an audit log of everything @Claude did and who requested each task.
The access model is documented in Anthropic’s Agent Identity framework, which is worth reading if you’re evaluating this for a regulated environment.
Migration and availability
Claude Tag replaces the existing Claude in Slack app. Administrators have 30 days to opt in and migrate. Anthropic is offering an introductory launch credit to eligible Enterprise and Team organizations so teams can try it out without immediate budget impact.
The beta is limited to Claude Enterprise and Team customers starting today. Anthropic says it plans to expand availability but did not give a timeline for broader rollout.
What this means
This is Anthropic’s most direct move yet into enterprise collaboration tools. Last month, I covered the two Claude deadlines that hit teams in June, and Anthropic previously showed what Claude can do when it controls your Mac directly. Now Claude Tag extends that agentic capability into the chat layer where most team communication already happens.
It also signals where Anthropic sees the market going: not better chat UIs, but AI that lives in the tools teams already use. The @Claude model is a more natural fit for enterprise adoption than asking teams to adopt a separate AI platform. You don’t have to learn a new interface. You just tag it in.
The big open question is how well it works at scale with complex toolchains and real enterprise access controls. The beta will answer that.
Bottom line
Anthropic turned Claude into a Slack-native team member that codes, researches, monitors channels, and runs async tasks. Enterprise and Team customers can try it today on Opus 4.8. If you are comparing AI subscriptions, I ranked Claude pricing against every major competitor. If the dogfooding numbers are real, this is going to put pressure on every other AI company that still thinks a chat window is the right interface for team AI.
