Anthropic launches Claude Tag for Slack, turning chat into a persistent AI teammate

Anthropic is launching Claude Tag in research preview: an “always-on Claude” that lives inside Slack and acts as a persistent AI teammate. The key behavior is simple but consequential. Users can tag @Claude directly in Slack conversations to pull insights in the middle of work, and to assign tasks without leaving their team’s chat environment.

Anthropic says Claude Tag is available starting today to Claude Enterprise and Claude Team customers. That matters because it is not an experimental demo that no one can deploy. It is a product entry point for organizations that already pay for Anthropic in enterprise settings, meaning the decision is likely to land quickly with IT, security, and the folks responsible for internal productivity tooling.

If you have been watching the AI market over the last year, this launch fits a clear pattern. The winners are not just building smarter models. They are building places where people already work. Slack is that place for many teams: it is where questions happen, decisions get recorded, and tasks quietly turn into action. By embedding Claude inside Slack as a persistent presence, Anthropic is aiming to reduce the friction between “I need an answer” and “I got an answer,” and to reduce the friction between “we should do this” and “someone is actually driving it.”

The “always-on” angle is the real hook, even if the feature sounds almost too neat. A persistent AI teammate changes how workflows feel. Instead of treating AI like a tool you summon occasionally, it starts to feel like a role in the room. Tagging @Claude is the interface. But the bigger shift is cultural: the default assumption becomes that insight and tasking can be requested inline, during an ongoing conversation. That can accelerate execution, but it also raises a classic enterprise question: who is accountable for what gets generated, assigned, or acted on.

The source notes that Claude Tag lets users “tag @Claude to get insights in conversations and assign tasks.” That is two distinct capabilities, and each has different governance implications. Insights during conversations touch the quality and reliability of outputs, plus the risk of outdated or incorrect information getting repeated. Task assignment touches operational discipline: if an AI suggests a task, the organization needs to ensure the task actually has the right ownership, the right context, and the right definition of done.

Claude Tag being a research preview adds another layer. Research preview typically signals that the feature is being tested with real users, with the expectation of iteration. For executive teams, that is not a reason to ignore it. It is a reason to treat the rollout like a controlled experiment. Pilot it with a narrow set of teams. Establish guardrails on what kinds of conversations are appropriate. Decide how to log outputs for audit and how to handle mistakes when they happen. If you let “always-on” run wild, you will find out what your governance process is made of the hard way.

There is also a product strategy angle. Anthropic is positioning this as an evolution of its broader Slack-era push, building a direct competitor to the idea that AI belongs in your existing enterprise communication stack. That puts pressure on other AI vendors and workplace platform providers, because the bar shifts. It is no longer enough to provide a chat box or a separate assistant. The expectation becomes that the assistant can operate where decisions are made and tasks are tracked.

For boards and senior operators, the question is less “is Slack the right place for AI” and more “what happens when AI becomes part of the operating system of work.” If Claude Tag is adopted widely, internal processes may change fast: meeting prep, customer support drafting, incident response coordination, and project triage can all look different when AI can be summoned and used without context switching. That can translate into productivity gains. It can also increase the volume of semi-automated work that requires review, and it can raise privacy and security concerns tied to what data teams discuss in Slack.

The smartest way to read this launch is as a signal. Anthropic is treating always-on AI as a near-term enterprise workflow layer, not a futuristic idea. Starting today, Claude Enterprise and Claude Team customers can put @Claude inside Slack. The second-order stake for executives is that AI usage will move closer to day-to-day decisions and operational tasking, which means your policies, tooling, and measurement need to keep up, or you risk speed without control.