Breaking

Anthropic Unveils Slack-Resident AI 'Claude Tag', Writing 65% of Internal Code

June 23, 2026 at 13:14 EDT

AI Agents
Software Dev & Coding
Sales & Business

On June 23, Anthropic announced Claude Tag, a new feature that lets Claude join Slack as a persistent "team member," releasing it as a beta (research preview). Users delegate work by tagging @Claude in threads, and the company's internal version already writes 65% of its product team's code.

Anthropic · Claude Tag Beta

Tag @Claude Like a Teammate — and Let It Run the Workflow

Anthropic's new Slack-integrated agent lets teams delegate tasks to @Claude in real time. After running it internally for nearly a year, Claude now writes 65% of its own product team's code — and built most of Claude Tag itself.

65%

of the Claude Code team's product code is written by Claude

3×

faster incident response, internally reported

80%

cut in investigation time

The 70–80% reality

Practitioners report Claude handles most of the task — but completeness stalls, and humans still finish the rest.

70–80%

Claude automates

20–30%

human finishes & reviews

What happens when you tag @Claude on an incident

Page fires → retrieve graphs & deploy diffs

→

Root-cause analysis

→

Create fix PR

→

Monitor metrics & resolve

What works

Faster prototyping and async, proactive collaboration
Automate 70–80%, humans finish the rest
Covers ops, review & docs — not just code

The friction

Completeness stalls near 70–80%
Human review remains essential
Usage limits; context & permissions are key

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.