Shared engineering tasks
Investigate bugs, prepare code changes, open pull requests, or summarize technical context in a visible team thread.
Independent tool overview
Claude Tag is Anthropic's beta Slack agent for Claude Team and Enterprise organizations, letting channels delegate tool-using work to a shared @Claude identity with scoped memory, scheduled tasks, proactive follow-up, and audit controls.
Visit the official Claude Tag site ↗
Overview
Claude Tag moves Claude from a private chat into Slack as a shared, asynchronous teammate. In a configured channel, anyone with access can mention @Claude, assign a task, watch its plan and progress in a thread, and continue from the same channel memory. Owners can connect repositories and workplace tools, let Claude schedule longer-running work, and optionally enable ambient behavior that proactively surfaces information or follows up on stalled threads.
That collaboration model is powerful because the agent can act with organization credentials rather than one employee's account. It is also the central risk. A broadly configured channel can turn a casual Slack message, malicious pasted content, guest access, or mistaken instruction into repository changes, external calls, data exposure, or spend. Claude Tag should be deployed as a privileged service identity with narrow scopes, isolated channels, hard budget limits, human approval for consequential actions, and continuous audit review.
Use cases
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Investigate bugs, prepare code changes, open pull requests, or summarize technical context in a visible team thread.
Research cases, connect approved systems, draft responses, and follow unresolved threads under human review.
Let authorized channel members delegate repeatable queries against approved data sources.
Schedule multi-stage research, analysis, or coding work that runs asynchronously and reports back.
Retain useful channel context so the agent does not need the same project explanation on every task.
Use ambient behavior to surface relevant updates or prompt owners when a thread appears unresolved.
Provide central permissions, spend limits, channel isolation, guest controls, and an audit trail.
Capabilities
Starts visible work when someone tags @Claude and returns progress and results in the Slack thread.
Everyone in a channel interacts with the same Claude context and can steer or continue the work.
Supports private conversations using the individual's enabled Claude tools and connectors.
Opens Claude in Slack's side panel without requiring a channel mention.
Remembers relevant information by workspace and channel instead of starting each task from zero.
Uses organization-configured credentials, services, and code repositories to complete tasks.
Acts under a distinct organization identity so posts, commits, pull requests, and network calls can be attributed.
Works in stages while people continue other work and reports back when finished.
Can create recurring or future tasks that continue over hours or days.
Optionally checks in proactively, surfaces relevant information, and follows up on stalled work.
Applies organization-wide, workspace, and private-channel credentials and repository scopes through inheritance.
Lets owners block Claude, restrict it to channel-only access, or allow full configured access when Slack guests are present.
Provides organization-wide and per-channel hard caps, inherited defaults, threshold alerts, and channel analytics.
Lets admins inspect, edit, and delete what Claude remembers for a channel or workspace.
Lists one-time and scheduled tasks plus Agent Identity network calls and the user who requested each action.
Process
Step 1
Choose one high-value, reversible workflow with clear inputs, allowed tools, output, owner, and success measure.
Step 2
Have a Primary Owner or Owner provision Claude Tag separately from any human's broad personal credentials.
Step 3
Grant only the repositories, read scopes, write scopes, environments, and data needed for the pilot.
Step 4
Use a private test channel, verify membership and guests, and avoid inheriting unrelated workspace-wide credentials.
Step 5
Require an accountable person before messages, ticket changes, merges, deployments, purchases, deletions, or sensitive data exports.
Step 6
Apply a conservative organization cap, default per-channel cap, and named process for reviewing increase requests.
Step 7
Try ambiguous requests, malicious pasted instructions, cross-channel questions, guest presence, stale tasks, and credential failures.
Step 8
Review Slack threads, commits, pull requests, downstream system logs, network calls, memory, and spend during the pilot.
Step 9
List active triggers regularly, assign an owner, set an expiration date, and disable recurring tasks that no longer have a purpose.
Step 10
Create separate scoped configurations for engineering, sales, legal, and other domains instead of one all-powerful Claude.
Cost
Claude Tag requires a Claude Team or Enterprise organization and adds consumption-based channel charges. Team seat prices are published, but Tag usage is not a flat per-seat add-on: channel tasks draw from the organization's usage balance, while direct messages draw from the individual's Claude account. Set hard limits before enabling it. U.S. Team prices below exclude tax; Enterprise seat and usage pricing requires the applicable live plan or sales quote.
$25/user/month monthly or $20/user/month annually
Base Team seat that makes the beta available to an eligible organization.
$125/user/month monthly or $100/user/month annually
Higher-capacity Team seat for power users; organizations can mix Standard and Premium seats.
Check live quote or checkout
Advanced governance and consumption model for larger or regulated organizations.
Consumption-based
Variable organization charge based on the work delegated in Slack.
Pricing checked . Check current pricing at the source ↗
Assessment
Compare
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Consumer
Choose Slackbot for Slack's native personal AI agent and tighter alignment with Slack's own product experience.
Explore Slackbot →Coding
Choose Slack Code when the main need is a team software-development channel built around multiple coding agents.
Explore Slack Code →Project Management
Choose standard Claude when private chat, Projects, Code, or Cowork is sufficient and shared Slack identity is unnecessary.
Explore Claude →Questions
Claude Tag is Anthropic's shared agent in Slack. Team members mention @Claude to delegate work using channel context, organization tools, persistent memory, scheduled tasks, and a distinct Agent Identity.
It is currently available in beta to Claude Team and Enterprise organizations. A Primary Owner or Owner must set it up and choose who and which channels can use it.
Yes. Anthropic says the earlier Claude in Slack experience switched to Claude Tag on August 3, 2026.
It requires Team or Enterprise access, then channel work is billed to the organization's usage balance based on consumption. Direct messages are billed to the individual's Claude account.
Yes. It retains relevant memory scoped to workspaces and channels so future tasks can use prior context. Admins can inspect, edit, and delete that memory.
Only channels and tools configured by an Owner or Primary Owner. Private channels can add credentials and repositories on top of inherited organization and workspace scopes.
Owners can block Claude entirely, allow channel-only access, or permit full configured access. Restrict is the default and is the safer starting point.
Yes, when given the necessary repository and tool access. Commits and pull requests can be attributed to the Claude GitHub App, but humans must review tests, security, and change scope before merge.
Yes. It can schedule work and follow up later. Teams should inventory active triggers, assign owners and expiration dates, and remove stale automation.
Owners can set an organization hard cap, per-channel caps, inherited defaults, and alerts at 75% and 95%. A task that would exceed the limit is declined.
Only under a carefully scoped control model. Use a dedicated identity, least privilege, isolated channels, guest restrictions, approval gates, sandboxed environments, hard spend limits, and continuous audit review.
Bottom line
Claude Tag is a meaningful step beyond a Slack chatbot: shared memory, a distinct identity, connected tools, long-running tasks, and proactive follow-up can make it useful operational infrastructure. Those same features make it a privileged agent rather than a convenience bot. The best deployments begin with read-only, reversible work in isolated channels and expand only after permissions, guests, memory, scheduled tasks, spend, and audit behavior have been tested.
Visit Claude Tag website ↗
Mistral OCR 4 - Mistral’s OCR model with layout-aware document understanding

🗣️ Mercury 2 - Inception's diffusion reasoning model built for realtime voice agents

Meta Enterprise Agent - AI agent for customer sales and support across Meta's apps

Presence - OpenAI's managed enterprise service deploying chat and voice agents

Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.