Anthropic launched Claude Tag on June 23, 2026, a new feature that embeds a persistent AI agent directly into Slack channels for Enterprise and Team customers. This is not the old Slack integration where you ask Claude a question and get an answer. Claude Tag is designed to work like a standing member of the team: it reads channel history, builds memory over time, executes multi-step tasks asynchronously, and takes initiative without being prompted.

The product replaces Anthropic’s existing Slack app, with automatic migration set for August 3, 2026 for any organization that does not opt in manually first.

What Claude Tag Actually Does

When an administrator activates Claude Tag in a Slack channel, the agent immediately begins indexing that channel’s conversation history, files, pinned messages, and documents. It builds what Anthropic describes as a knowledge graph: a live map of project context, terminology, open decisions, team roles, and recurring topics.

From that point, team members can tag @Claude to assign work. Unlike a basic prompt-response tool, Claude Tag will break down complex requests into steps, use connected tools to execute each one, and post back results to the thread when it is done. Team members can walk away; Claude keeps working.

That asynchronous execution model is a significant shift. Most AI tools in the workplace today require someone to be present and actively steering the conversation. Claude Tag is built to close tickets, draft documents, summarize meeting threads, and coordinate follow-ups without constant supervision.

The enterprise security architecture is worth noting. Admins can create separate Claude identities scoped to individual channels or use cases, each with its own toolset, memory boundary, and token-spend limit. A Claude instance configured for the sales team has no access to engineering data and cannot share memory across channels. That scoping means organizations can deploy AI at the team level without building a universal integration that touches everything at once.

Anthropic also reports that its own internal product teams now route approximately 65% of all code changes through an internal version of Claude Tag. That figure is the clearest signal of what the company actually believes this product can do.

Why This Matters Beyond Slack

The Claude Tag launch reflects a broader transition in how enterprise AI tools are being designed. For the past two years, most AI deployments in the workplace have been personal assistants: tools that help an individual user think faster, write better, or search more efficiently. The value accrues to the individual using the tool.

Claude Tag is a different architecture. The value accrues to the team. Because the agent accumulates shared institutional knowledge, it becomes more useful the longer it runs in a channel and the more people interact with it. A new hire can ask what decisions were made last month and get a real answer. A project lead can offload the task of writing up meeting outcomes and trust it will happen without delegation overhead.

This matters for how businesses think about AI adoption. Deploying AI as a team layer changes the ROI equation. Instead of measuring whether one individual is more productive, you are measuring whether the team moves faster, loses fewer things to handoff gaps, and can run more work in parallel.

What This Means for Business

If you are running a business that uses Slack, the practical question right now is: which channels would benefit from an agent that never forgets and never misses a thread?

The answer for most teams is some version of project management, client work, and internal knowledge. Claude Tag is reasonably well suited to all three. It does not replace the human judgment that drives strategy or client relationships, but it can carry the administrative load that currently slips through the cracks between meetings.

For businesses that have been watching AI tool development but waiting for something that fits into existing workflows without requiring a new platform, Claude Tag is probably the most natural enterprise AI insertion point we have seen. It does not ask people to change where they work. It brings the capability into the place they already spend their day.

The migration deadline of August 3 also creates a forcing function for Enterprise and Team customers who have not made a decision yet. This one is worth evaluating before the clock runs out.

For organizations looking to understand how to structure AI deployments at the team level, including which workflows to start with and how to measure the outcome, the Enterprise DNA advisory service is a practical starting point.