Anthropic began a beta rollout of Claude Tag, a shared AI agent for Slack, on June 23, and while Slack owner Salesforce officially promoted it, concerns are spreading inside the company that the feature competes with its own products.
June 23, 2026 · Anthropic × Slack
Claude Tag turns Slack into a "multiplayer AI" workspace
Anthropic launched a shared AI team member that activates with @Claude , breaking tasks into steps and replying in-thread. As Salesforce integrates it, some staff reportedly fear the model provider could become a "Trojan horse."
65%
of Anthropic's own product-team code generated by Claude Tag internally
30-day
opt-in migration replacing the existing "Claude in Slack" app
Opus 4.8
model powering Claude Enterprise / Team plans
How a tagged task flows
@Claude
Tag it in natural language
→
Break into steps
Plans & executes the task
→
Reply in-thread
Plus ambient & async follow-up
Ambient mode: flags & follows up
Async self-scheduling: hourly / daily
Cross-channel memory
✓ What testers welcome
Remembers context, responds to plain-language tags
Fast queries via external tools like Supabase
Accurate cross-system search & document generation
⚠ What raises concern
Confusion & glitches from replacing the old app
Lock-in: context accumulates on Anthropic's side
Burden of memory limits & scope management
The strategic tension
The model provider sitting inside the software could outpace the platform built on top of it.
Direct competition
Overlaps with Slack's native Slackbot and Agentforce
Data influence
Expanding customer reach through deep data access
"Trojan horse"
How some Salesforce staff reportedly describe it
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…