Claude Tag is now available in beta via Claude Enterprise in AWS Marketplace
Teams can now @mention Claude directly in Slack channels with per-channel governance, async task delegation, and AWS spend commitment support.
View original announcement →Visual Summary
graph TD
A{{Claude Tag}}:::announced
B((Slack Channels)):::external
C(Claude Enterprise):::compute
D(AWS Marketplace):::compute
E([Per-Channel Scoping]):::feature
F([Persistent Context]):::feature
G([Spend Controls]):::feature
H((Team Members)):::external
I([Tool & Data Integration]):::feature
H ==>|"@mentions"| A
A ==>|"embedded in"| B
A -->|"powered by"| C
C -->|"procured via"| D
A -->|"enforces"| E
A -->|"builds"| F
A -.->|"limits"| G
A -->|"connects to"| I
classDef announced fill:#ff9900,stroke:#ec7211,color:#fff,font-weight:bold
classDef compute fill:#e3f2fd,stroke:#1565c0,color:#1565c0
classDef storage fill:#e8f5e9,stroke:#2e7d32,color:#2e7d32
classDef feature fill:#fff3e0,stroke:#e65100,color:#e65100
classDef external fill:#f5f5f5,stroke:#616161,color:#616161
What's New
Anthropic has launched Claude Tag in beta, a new capability that embeds Claude directly into Slack channels, enabling teams to collaborate with AI in their existing communication workflows. Available exclusively to AWS customers who access Claude Enterprise through AWS Marketplace, Claude Tag allows any channel member to invoke Claude via @mention, delegate tasks, and connect it to tools, data, and codebases. The feature is designed with enterprise governance in mind, operating under a scoped per-channel identity with spend controls and ambient mode disabled by default.
How It Works
- Channel-level provisioning: An admin provisions Claude's agent identity through the Claude admin console (approximately one hour of setup) and scopes permissions individually per Slack channel, limiting blast radius and enforcing least-privilege access.
- @mention invocation: Any member of a channel where Claude has been granted access can tag @Claude to delegate tasks, making collaboration multiplayer rather than limited to a single user session.
- Persistent context memory: Claude builds and retains contextual awareness from conversations within the channels it monitors, enabling more relevant and informed responses over time without requiring users to re-explain background.
- Tool and data integration: Administrators can connect Claude Tag to external tools, data sources, and codebases, allowing it to act on real organizational assets rather than operating in isolation.
- Future task planning: Claude can schedule and plan tasks to be completed at a later time, supporting asynchronous workflows where teams delegate work and return to results.
- Consumption-based pricing: Usage is metered by consumption rather than headcount, with org-wide budget visibility and the ability to set per-channel spend limits, drawing down from existing AWS EDP or PPA commitments.
- Security defaults: Claude Tag operates under its own distinct identity scoped per channel, with ambient listening mode off by default, ensuring it only engages when explicitly invoked.
Why It's Important
- Reduces context-switching friction: By bringing Claude into Slack—where teams already communicate—organizations eliminate the need to switch between a standalone AI tool and their collaboration platform, lowering adoption barriers significantly.
- Enables true team-scale AI collaboration: The multiplayer model means AI assistance is no longer siloed to individual users; entire teams can delegate, review, and build on Claude's outputs within a shared workspace.
- Accelerates async workflows: The ability to delegate tasks and have Claude complete them while team members focus elsewhere mirrors how teams use human collaborators, unlocking productivity gains beyond simple Q&A interactions.
- Strengthens enterprise governance posture: Per-channel identity scoping, spend controls, and ambient mode off by default give security and compliance teams granular control, making it easier to audit and govern AI usage at scale.
- Leverages existing AWS commercial agreements: AWS customers can apply Claude Tag usage against existing EDP or PPA commitments, simplifying procurement and avoiding new vendor relationships or budget cycles.
How It's Different
- Multiplayer vs. single-user AI: Unlike most AI chat tools that operate in individual sessions, Claude Tag is natively collaborative—multiple team members in the same channel can interact with and build on a shared Claude context simultaneously.
- Channel-scoped identity vs. org-wide agent: Rather than deploying a single AI agent with broad organizational access, Claude Tag's per-channel identity model limits scope and exposure, a meaningful architectural difference for security-conscious enterprises.
- Ambient mode off by default: Many AI integrations passively monitor conversations; Claude Tag explicitly requires an @mention to engage, reducing unintended data exposure and giving users clear control over when AI is active.
- Consumption pricing vs. seat-based licensing: Traditional enterprise AI tools charge per seat regardless of usage; Claude Tag's consumption model aligns cost directly with value delivered and avoids paying for inactive users.
- AWS Marketplace parity with first-party: Unlike some marketplace-distributed products that offer reduced functionality, Claude Tag via AWS Marketplace delivers identical capabilities, setup, and controls as direct Anthropic Claude Enterprise, with the added benefit of AWS commercial alignment.
- Integrated tool and codebase connectivity: Beyond chat, Claude Tag can be connected to live data sources and codebases within a channel context, enabling it to act as a functional team member rather than a passive question-answering tool.
When to Prefer It
- Engineering teams using Slack for incident response: When on-call engineers need rapid analysis of logs, code, or runbooks during an incident, Claude Tag can be invoked directly in the incident channel with access to the relevant codebase.
- Cross-functional project teams: When multiple stakeholders from different departments collaborate in a shared Slack channel and need consistent AI assistance without each person maintaining their own AI tool subscription.
- Organizations with strict AI governance requirements: When security or compliance teams need auditable, scoped AI access with clear identity boundaries and spend controls rather than broad, ungoverned AI tool proliferation.
- Teams with async or distributed workflows: When team members work across time zones and need to delegate research, summarization, or task execution to Claude and return to completed work, rather than requiring synchronous AI interaction.
- AWS-committed customers seeking to consolidate spend: When an organization has existing AWS EDP or PPA commitments and wants to apply that spend toward enterprise AI tooling without adding a separate vendor contract.
- Organizations already standardized on Claude Enterprise: When a company is already using Claude Enterprise via AWS Marketplace and wants to extend that investment into Slack without additional procurement or onboarding overhead.
Availability
- Status: Available in public beta as of June 23, 2026; not yet generally available.
- Access requirement: Exclusively available to AWS customers who have purchased Claude Enterprise through AWS Marketplace; not available via direct Anthropic subscription or other cloud marketplaces at this time.
- Initial platform support: Slack is the first supported channel; additional platforms have not been announced.
- Pricing model: Consumption-based, metered by usage rather than headcount; org-wide budget visibility with per-channel spend limits; eligible to draw down from AWS EDP or PPA commitments.
- Setup time: Admin provisioning of the Claude Tag agent identity in the Claude admin console takes approximately one hour.
- Regional availability: Specific AWS region support is not detailed in the announcement; availability is tied to Claude Enterprise on AWS Marketplace regional coverage.
- HIPAA eligibility: Claude Enterprise (the underlying product) is HIPAA-eligible with a BAA available, though applicability of this to Claude Tag specifically in beta should be confirmed with Anthropic.
