Meet @Claude, the teammate that doesn’t need onboarding.
For the past few years, using AI at work usually meant opening another browser tab, typing out a prompt, then pasting the results back into Slack.
Useful? Absolutely. Collaborative? Not really.
Anthropic wants to change that with Claude Tag, a new feature that lets anyone in a Slack channel simply tag @Claude like they would a teammate.
From there, Claude can summarise meetings, investigate bugs, analyse product data, review code, draft documentation or take on longer tasks before reporting back into the thread.
The biggest shift is that Claude becomes part of the conversation instead of another private chat. Rather than everyone working with their own separate AI assistant, teams share a single Claude inside selected Slack channels. Anyone can see what it’s working on, pick up an existing task or build on someone else’s prompt.
Claude also builds context over time from the conversations, tools and documents it’s been given permission to access, meaning there’s less need to explain the same project from scratch every time.
It can even work asynchronously, tackling jobs in the background before posting updates back into the thread. With its optional “ambient” mode enabled, Claude can follow up on unresolved conversations or surface useful information without being prompted.
Anthropic says the feature has already become central to its own workflow, with 65% of its product team’s code now created using an internal version of Claude Tag.
It’s a surprisingly simple idea, a single @ symbol, but it could mark one of the biggest shifts yet in how teams actually work with AI.
