What is Claude Tag for Slack?

Claude Tag is Anthropic's new persistent AI teammate inside Slack, launched in beta on June 23-24, 2026 in partnership with Salesforce. It replaces the previous 'Claude in Slack' app with a richer integration: users summon Claude by typing @Claude in any channel, and Claude can collaborate on workflows, requests, and ideation in real-time across messages, threads, and connected tools. With 'ambient behavior' enabled by an admin, Claude Tag can proactively monitor channels, follow up on unresolved tasks or inactive threads, and provide updates without being explicitly summoned. It is available in beta for Claude Enterprise and Claude Team plan subscribers.

How is Claude Tag different from the old Claude in Slack app?

Four key differences. (1) Persistent vs request-response: the old app required you to message Claude in a DM or invoke it for a specific request; Claude Tag stays in channels as an ambient teammate. (2) Multi-channel context: Claude Tag can learn over time from channel conversations and connected data sources (with admin permission), while the old app had session-scoped context. (3) Agentic execution: Claude Tag can break down tasks, execute them across connected tools (codebases, databases, third-party apps via MCP), and respond in threads — the old app was conversational only. (4) Proactive ambient mode: Claude Tag can follow up on stalled threads and unresolved tasks autonomously when ambient behavior is enabled; the old app only responded when addressed.

Who can use Claude Tag?

As of June 24, 2026, Claude Tag is in beta for Claude Enterprise and Claude Team plan subscribers. Individual Claude Pro and Claude Free users cannot use Claude Tag (they still have access to other Claude integrations and the Claude.ai web/desktop app). Slack workspace administrators must enable Claude Tag and configure which channels Claude can join, which tools and data sources it can access, and whether ambient behavior is on. Enterprise admins also control which employees can grant Claude Tag access to their workflows. A general-availability rollout timeline has not been published — beta access is rolling out through July 2026 with broader availability expected in Q3 2026.

What can Claude Tag actually do in a channel?

Five practical capabilities. (1) Summarize threads and channel activity on demand (@Claude summarize the last 50 messages). (2) Delegate multi-step tasks (@Claude pull the open bugs assigned to the backend team and rank them by user impact). (3) Connect to and act on connected tools — codebases via Anthropic's GitHub integration, databases via MCP servers, third-party SaaS apps via available connectors. (4) Generate documents, code, and structured outputs directly in thread. (5) Ambient follow-up — when enabled, proactively re-pings stalled threads, follows up on unresolved tasks, and surfaces updates without being asked. The combination effectively turns Slack from a messaging tool into a workflow surface with an embedded AI teammate.

Quick Answer

What Is Claude Tag in Slack? (Anthropic, June 2026)

Published: June 25, 2026

What Is Claude Tag in Slack? (Anthropic, June 2026)

On June 23-24, 2026, Anthropic and Salesforce launched Claude Tag — a new persistent AI teammate integration inside Slack that replaces the previous “Claude in Slack” app. Users summon Claude by typing @Claude in any channel, and Claude can collaborate on workflows, execute tasks across connected tools, and (when ambient behavior is enabled) proactively follow up on unresolved threads. It’s available in beta for Claude Enterprise and Claude Team subscribers. Here’s what it is, how it works, and why it matters.

Last verified: June 25, 2026.

TL;DR

Launched: Beta rollout starting June 23-24, 2026 (Anthropic + Salesforce joint announcement)
What it is: Persistent @Claude AI teammate in Slack channels — not just a DM-based bot
Replaces: The previous “Claude in Slack” app
Who can use it: Claude Enterprise and Claude Team plan subscribers (no Pro/Free access)
Key new capability: Ambient behavior — proactive monitoring of channels, follow-up on unresolved tasks, autonomous re-pings of stalled threads
Admin controls: Workspace admins control which channels Claude joins, which tools/data sources it accesses, and whether ambient mode is on
GA timing: Beta through Q3 2026; general availability not yet announced

What’s new about Claude Tag

The previous “Claude in Slack” app was a conventional chatbot: you DM’d Claude or @-mentioned it for a specific task, Claude responded, conversation ended. Useful, but not transformative.

Claude Tag is structurally different in four ways:

1. Persistent in channels, not just request-response

Claude Tag joins channels as a member, not a bot. It sees channel context over time, learns the team’s workflow patterns (with permission), and stays in the channel between messages. When you @Claude, it has the surrounding context already.

2. Multi-channel and multi-tool context

Claude Tag can be granted access to multiple channels, multiple data sources, and connected tools (GitHub repos, internal documents via MCP, SaaS apps via available connectors). When you ask it something in one channel, it can pull context from connected sources rather than starting from scratch.

3. Agentic execution, not just conversation

Claude Tag can break down a multi-step request, execute the steps across connected tools, and respond with the result. Example: “@Claude, pull the open P0 bugs assigned to the backend team in GitHub, rank them by user impact based on Sentry error frequency, and post a prioritized list in #engineering-triage” — Claude Tag can actually do that end-to-end, not just describe how to do it.

4. Ambient behavior

The most novel capability. When workspace admins enable ambient behavior, Claude Tag can proactively:

Re-ping stalled threads where someone asked a question and never got an answer
Follow up on unresolved tasks that were assigned but not closed
Surface relevant context when a new topic comes up in a channel (“this was discussed in #planning last week — here’s the conclusion”)
Provide periodic updates on long-running tasks Claude is executing

This shifts Claude from “thing you ask” to “thing in the channel that helps when it can.” It’s the same conceptual move Microsoft made with Copilot in Teams, but Anthropic’s positioning is broader-scope and more agentic.

What Claude Tag requires

Three setup pieces:

Requirement	Detail
Claude plan	Claude Enterprise or Claude Team (not Pro, not Free)
Slack workspace admin permission	Admin must enable Claude Tag, configure channel access, connected tools, ambient behavior
Connected data sources	Optional but valuable — GitHub repos, MCP servers, SaaS connectors

Per-user activation is also controlled — admins can grant or restrict which employees can invite Claude Tag into their workflows.

How it compares to Slack AI, Microsoft Copilot in Teams, ChatGPT Enterprise

The competitive picture as of June 25, 2026:

Capability	Claude Tag (Anthropic)	Slack AI (Salesforce native)	Copilot in Teams (Microsoft)	ChatGPT Enterprise (OpenAI)
Channel presence	Persistent @Claude in any channel	Sidebar summaries, no channel presence	Persistent Copilot in chats/meetings	DM/sidebar bot via integrations
Ambient behavior	Yes (admin opt-in)	No	Limited (meeting follow-ups)	No
Multi-tool execution	Yes (MCP + connectors)	Limited	Yes (Graph + connectors)	Limited
Underlying model	Claude family (Fable 5, Sonnet 4.7, Opus 4.8)	Salesforce-routed (varies)	GPT-5.5 / GPT-5.6 family	GPT-5.5 / GPT-5.6 family
Plan required	Claude Enterprise / Team	Slack paid plan	M365 + Copilot license	ChatGPT Enterprise
Best for	Slack-first agentic workflows	Conversation summaries	Microsoft 365 shops	OpenAI-stack teams

For teams that are Slack-native and want agentic AI workflows across connected tools, Claude Tag is the most ambitious offering. For teams in Microsoft 365, Copilot in Teams is the obvious choice. For teams that want a less channel-embedded AI assistant, the existing ChatGPT Enterprise / Claude in Slack patterns still work.

Privacy and admin controls

Three categories of control admins should configure before enabling Claude Tag:

Channel scope

Which channels can Claude Tag join? Sensitive channels (security incidents, executive discussions, HR matters) typically should not have Claude Tag enabled. Most engineering, support, and operational channels are appropriate.

Tool and data source access

What can Claude Tag read and act on? GitHub access, MCP servers, third-party connectors all require explicit admin grants. Default to least-privilege — start narrow, expand as use cases prove out.

Ambient behavior

Whether Claude Tag can act without being explicitly summoned. This is the most powerful and most sensitive setting. Recommended approach: keep ambient off for the first 2-4 weeks of beta usage, observe what teams ask Claude for explicitly, then selectively enable ambient in channels where the pattern is clear.

Practical use cases

The most-requested workflows from early beta:

Engineering triage: rank open issues, generate context for on-call rotations, summarize PR review threads
Customer support: pull related historical tickets, draft response templates, escalate based on sentiment
Product planning: synthesize discussion across multiple planning channels, generate one-pagers from chat threads
Recruiting coordination: track candidate status across channels, follow up on stalled interview loops
Incident response: assemble incident timelines, draft post-mortems, identify on-call gaps

In every case, the value isn’t Claude doing something humans couldn’t — it’s Claude doing it without humans needing to assemble context manually from multiple channels and tools.

What to do today

If you’re a Claude Enterprise or Team admin

Request beta access via your Anthropic account team
Plan your channel scope, tool access, and ambient policy before turning it on
Pilot in one team for 2-4 weeks before rolling out broadly
Watch for the GA announcement (expected Q3 2026)

If you’re a Claude Pro or Free user

Claude Tag isn’t available to you yet
Continue using Claude.ai, Claude Code, and the Claude API
If your team has Claude Team or Enterprise, request workspace-admin to enable it

If you’re choosing between Claude Tag and competitors

Claude Tag wins for Slack-first agentic workflows with connected tools
Microsoft Copilot in Teams wins for M365-native teams
Existing Slack AI is good enough for pure summarization workflows

Bottom line

Claude Tag is the most ambitious Slack AI integration shipped to date. The shift from request-response bot to persistent ambient teammate is genuinely new, and the multi-tool agentic execution makes it useful for real workflows beyond conversation summarization.

For Claude Enterprise and Team subscribers in Slack-heavy organizations, this is worth piloting now. For everyone else, watch for the GA announcement and the broader rollout in Q3 2026.

The strategic point: Anthropic is moving Claude from “API you call” to “ambient teammate that lives where your work lives.” Slack is the first venue. Expect similar integrations across the workflow surfaces enterprises actually use — Teams, Notion, Linear, Jira, and the other places work happens.