Boris Cherny, the guy behind Claude Code at Anthropic, just dropped something interesting on X. His team isn’t just using Claude Code internally anymore. They’ve built something called Claude Tag and it’s already writing 65% of the product team’s new code.
This number suggests that Anthropic itself has moved past the “helpful coding assistant” phase and into something more embedded.
From “ask me anything” to “I’m in the channel”
Claude Tag lives in Slack. You add it to a channel, mention it with @.Claude, and it starts acting like a teammate instead of a tool you open in another tab. It has memory across conversations, understands the context of each channel, and can take initiative based on high-level instructions rather than constant prompting.
Introducing Claude Tag, a new way for teams to work with Claude.
— Claude (@claudeai) June 23, 2026
In Slack, Claude joins as a team member with access to the channels and tools you choose. Tag Claude in and delegate tasks to it while you focus on other work. pic.twitter.com/R2C6A5Kcye
The interesting part is what’s under the hood. It’s still powered by the same Claude Code model, but now it runs in isolated sandboxes per thread. It can clone repos, write code, run tests, and clean up after itself. No lingering state. No shared workspace with other agents. Just focused, disposable instances that disappear when the thread ends.
People inside Anthropic are already using it for PR reviews, incident investigations, data analysis, and onboarding questions. Instead of digging through docs or pinging the right person, they just tag Claude and it pulls the answer from the channel history and connected systems.
This is a big deal because it’s coming from the company that built one of the most respected coding models in the first place. They’re not just shipping this to customers — they’re running it on their own product work at scale. That 65% figure isn’t marketing. It’s internal usage data.
What’s emerging isn’t another chatbot. It’s an always-present teammate that lives where the work already happens. Whether this becomes the default way teams interact with AI or just one more notification to ignore will depend on how well the security and permission model holds up in real organizations.
For now, it’s in beta for Claude Enterprise and Team customers on Slack, with more surfaces promised later. The original thread is here if you want the full details straight from Boris.
Comments