Anthropic tests Claude Tag, an always-on AI helper in Slack

In short: Anthropic is rolling out a research preview of Claude Tag, a version of its Claude AI that stays in Slack channels and can remember what a team is working on.

What happened

Anthropic introduced Claude Tag as a research preview, meaning it is an early test release. It is available through Slack for Claude Enterprise and Claude Team customers.

Slack is a workplace chat app, like a set of group text messages for a company. With Claude Tag, people can type @Claude in a chat to ask for help, get summaries, or assign tasks.

The big change is that Claude Tag is designed to keep ongoing context and “memory” within a specific Slack channel. Think of it like a shared notebook that sits in the channel, so the AI can follow decisions and past discussions without someone re-explaining everything each time.

Anthropic says everyone in a channel shares the same Claude identity. That means teammates can see what Claude has been doing and continue the conversation where someone else left off.

Company administrators control what Claude can access. Anthropic says each Claude identity can be limited to certain channels, so a Claude used for legal work, for example, cannot carry information into an engineering channel.

Claude Tag can also work in an “ambient mode,” where it may jump in on its own to post updates, flag relevant items, or remind the team about forgotten threads, if it has permission.

Why it matters

Many workers already live in tools like Slack, and companies often lose track of knowledge buried in chat messages. An AI that can safely follow along could make it easier to find past decisions and keep projects moving, but it also raises practical questions about access and privacy, since it depends on what the AI is allowed to read.

Source: TechCrunch AI