Anthropic's Claude Tag Brings an Always-On AI Teammate Into Slack

Anthropic has launched Claude Tag, an "always-on" AI agent that works inside Slack. Team members summon it by typing @Claude in a channel or thread, much as they would mention a colleague. Once tagged, it can read the conversation, take on multi-step tasks, and hold onto context and memory across a channel over time.

The tool was announced on June 23, 2026, and is available the same day in beta for Claude Enterprise and Claude Team customers. For now, Slack is the only place it lives, with Anthropic saying it plans to expand to other platforms "in the coming weeks." Under the hood, Claude Tag runs on Claude Opus 4.8.

It replaces Anthropic's older "Claude in Slack" app. Administrators must opt in within 30 days to migrate, and the legacy app is being switched over on August 3, 2026.

The central idea is that Claude Tag accumulates knowledge about how a team works. As Anthropic puts it, "As Claude follows along with its channel, it learns ever more about the work."

A standout feature is an "ambient" mode, which lets Claude proactively join conversations to flag updates and follow up on threads that might otherwise be forgotten. It is also multiplayer — there is one shared Claude identity per channel that everyone can see and direct, rather than separate private assistants. With administrator permission, it can pull context from other channels and connected tools.

Cat Wu, Anthropic's head of product for Claude Code, said the appeal lies in the format rather than entirely new abilities: "A lot of the capabilities did exist, but actually the form factor of being able to tag it the same way that you would a coworker is really powerful." She described Claude Tag as "an evolution of Claude Code... built to be interactive and multiplayer." Slack's general manager Rob Seaman framed the move as "making AI multiplayer."

Anthropic has built in controls aimed at keeping the agent's knowledge appropriately contained. Administrators can scope Claude per channel — so, for example, a "legal" Claude cannot seed its memories into an engineering channel — and they can set spending limits.

The company is also offering introductory launch credits to eligible organisations to encourage adoption. As a sign of how far it expects the tool to go, Anthropic says 65% of its own product team's code is now created by an internal version of Claude Tag.

It is worth noting that the "always-listening" angle raised in some coverage is editorial analysis around privacy and compliance, not a reported incident. As with any tool that monitors conversations and retains memory, businesses will want to think through governance before switching it on.

Claude Tag lands in an increasingly crowded field. Reporting placed it alongside competitive efforts from Microsoft, with its Work IQ, Copilot and Graph offerings, as well as Glean — cited at roughly $300 million in annual recurring revenue and a $7 billion valuation — plus Snowflake, Databricks, Salesforce's Slackbot and the startup Viktor.

The launch is part of Anthropic's broader enterprise push. Reuters framed it against a reported company valuation of roughly $965 billion ahead of a likely public listing. That valuation figure is attributed to Reuters, and businesses should treat it as a press-reported number rather than a confirmed fact.

The common thread across all these players is the same: bringing AI agents directly into the tools where work already happens, so they can act on a company's own knowledge rather than sitting in a separate window.

The promise of an AI teammate that "learns ever more about the work" rests on one thing: the quality and accessibility of the information it can reach. A Slack channel captures conversations, but the most valuable context about your business usually lives elsewhere — in your CRM, your accounting system, your project records and your operational tools.

That's the practical lesson here. An AI agent is only as useful as the data it can draw on, and most mid-market businesses still have that data scattered across disconnected systems. Tools like Claude Tag can read a channel, but they can't reason well about your customers, finances or operations if those records are siloed and inconsistent.

This is where a unified, AI-ready data foundation earns its keep. Bringing your various sources together into one clean, consistent database gives any AI tool you adopt — whether it lives in Slack or somewhere else — a complete and trustworthy picture to work from. The arrival of agents like Claude Tag is a good prompt to get that foundation in order, so that when you do put AI to work, it has something solid to learn from.