On June 23, 2026, Anthropic introduced Claude Tag, a new way for teams to bring Claude into the channels where work already happens, starting with Slack. Instead of opening a separate chat window, anyone in a Slack channel can now tag @Claude, hand it a task in plain language, and keep working while it plans, executes, and reports back in the thread. For engineering leaders and founders weighing how much of their roadmap should run through in-house teams, contractors, or outsourced partners, Claude Tag is worth understanding on its own terms — not as another chatbot, but as a shift in how AI shows up inside a team’s daily workflow. This guide breaks down what Claude Tag actually is, how it differs from the Claude in Slack app it replaces, how it stacks up against Slack’s native AI and Microsoft Copilot in Teams, and where it realistically fits for teams evaluating their AI-assisted development stack.
What Is Claude Tag?
Claude Tag is Anthropic’s answer to a specific limitation of chat-based AI tools: they only work for one person, in one conversation, at a time. Claude Tag instead gives a Slack channel a single, shared Claude that everyone in that channel can see, interact with, and hand work to. An admin grants Claude access to selected channels and connects it to whichever tools, data sources, and even codebases the team decides on. From that point, anyone in the channel can tag @Claude with a request, and it breaks the task into stages, works through them using the tools it has access to, and posts the result back into the thread — visible to everyone, not just the person who asked.
Claude Tag replaces Anthropic’s existing Claude in Slack app. Administrators get a 30-day window to opt into the migration, and Anthropic is issuing an introductory launch credit to eligible Enterprise and Team organizations so the switch doesn’t come with a surprise usage bill. It’s available today in beta for Claude Enterprise and Team customers, runs on Claude Opus 4.8, and Anthropic has been explicit that Slack is the starting point, not the destination — the stated goal is to expand @Claude into the other places teams already work.
Anthropic frames Claude Tag as an evolution of Claude Code rather than a completely separate product: the same agentic capability, made more proactive and built to work with a full team instead of a single developer in a terminal. The internal adoption numbers are the most concrete evidence of that framing — Anthropic says 65% of its own product team’s code is now created by its internal version of Claude Tag, and the same delegation pattern has spread well past engineering into chasing down product metrics, working support tickets, and finding the root cause of bugs.
Why Slack First
Slack is a reasonable place to start for a tool built around shared, visible collaboration. It’s already the surface where most cross-functional coordination happens at a typical software company — standups, incident channels, support triage, ad hoc requests between engineering and product. Anthropic’s own explanation for the choice is straightforward: Slack is where collaborative work between teams and AI already happens, and where most of Anthropic’s day-to-day work takes place. For teams evaluating Claude Tag, the practical implication is that it’s not a general-purpose interface yet — it lives inside the channels you already use, not a standalone dashboard.
How Claude Tag Works
Four characteristics separate Claude Tag from using Claude in a private chat, and they matter more than the “AI in Slack” framing suggests on the surface.
Multiplayer by Design
Within a given channel, there’s exactly one Claude that interacts with everyone in it. That means anyone can see what it’s currently working on, and anyone can pick up a task where the last person left off, without re-explaining the context from scratch. This is a genuinely different interaction model from a single chat thread: it behaves less like a tool one person opens and closes, and more like a teammate the whole channel shares.
It Learns Over Time
As Claude follows a channel, it accumulates context about the work happening there, so people don’t have to re-explain background information every time they tag it. With permission, it can also draw on other Slack channels and connected data sources to build broader tacit knowledge — though Anthropic has been specific that it does not report information from private channels, which matters for any team weighing how much access to grant.
It Takes Initiative
If a channel enables “ambient” behavior, Claude will proactively surface information it thinks the channel needs — flagging relevant updates from other connected channels or tools, and following up on threads or tasks that have gone quiet without being resolved. This is the part of Claude Tag that moves it furthest from a traditional chatbot: it isn’t purely reactive.
It Works Asynchronously
You can hand Claude a task and move on to something else while it works, including tasks that run over hours or days — it can schedule follow-on work for itself as part of pursuing a longer project. Anthropic notes this has changed how its own teams operate: staff report spending much more of their time delegating tasks to multiple Claudes working in parallel, rather than waiting on one thing at a time.
Direct Messages Still Work
Claude Tag isn’t only a channel feature. You can message it directly for private, one-on-one use, where it responds using your own personal tools and connectors rather than the channel’s shared configuration.
Getting Started With Claude Tag
Anthropic designed the rollout with administrators as the gatekeepers, since @Claude’s access to sensitive data and task-specific tools needs to be tightly scoped from day one. Setup runs in four steps: pair Claude Tag with your Slack workspace, give Claude access to your tools, set a limit on the organization’s monthly spend, and test Claude in a private channel to confirm it behaves as expected before opening it up more broadly.
The scoping model is worth understanding before rollout, because it’s the mechanism that makes this safe to hand to a whole team rather than a single trusted user. Administrators effectively create separate Claude identities for different uses. A model configured for sales work doesn’t carry memories over to one set up for engineering, and it won’t hand sales data or sales-specific tools to the engineering team’s Claude. Everything — including what it remembers — stays scoped to the channels an admin has explicitly defined.
If your organization already uses the older Claude in Slack app, migrating is opt-in rather than automatic: administrators have 30 days to move over, and Anthropic is offering a launch credit to eligible Enterprise and Team organizations to offset the cost of trying it with the whole company rather than a single pilot channel.
Claude Tag vs the Alternatives
Claude Tag doesn’t have one obvious head-to-head competitor — it’s closer to a new category than a drop-in replacement for an existing tool. But three comparisons come up constantly when teams evaluate whether it’s worth adopting.
Claude Tag vs Slack’s Native AI
Slack’s own AI features are built around summarization and search: catching up on a channel you missed, pulling a quick answer from thread history, recapping a huddle. They’re genuinely useful, but they’re reactive and read-only by design — they help you understand what already happened, not delegate new work. Claude Tag is built to do the opposite: it takes a task, uses tools and connectors to act on it, and reports back with a result. Teams that only need better search and recap inside Slack don’t need Claude Tag for that. Teams that want to hand off actual work — not just get a summary of it — are looking at a different capability entirely.
Claude Tag vs Microsoft Copilot in Teams
Microsoft’s Copilot in Teams plays a broadly similar role inside the Microsoft 365 ecosystem: meeting summaries, drafting help in chat, and surface-level assistance woven into Teams and the wider Office suite. The practical difference teams report is depth of tool access and follow-through. Copilot in Teams is strongest at assisting with communication and content inside Microsoft’s own apps. Claude Tag is explicitly positioned as an evolution of Claude Code — designed to be handed genuinely agentic, multi-step tasks that touch codebases, internal tools, and data sources an admin has connected, not just drafting and summarizing inside the chat app itself. For organizations already standardized on Microsoft 365, Copilot’s native integration is a real advantage; for teams that want an agent capable of doing engineering and operational work rather than assisting with communication, Claude Tag’s tool and codebase access goes further.
Claude Tag vs Claude Code or Cowork Alone
It’s worth being clear that Claude Tag isn’t a replacement for Claude Code or Cowork — Anthropic describes it as the next step in the same lineage, built to work better with a full team rather than one person at a terminal or in a desktop app. A developer running Claude Code solo still gets a highly capable individual coding agent. Claude Tag adds the multiplayer layer on top: visibility across the whole channel, shared memory of what’s already been tried, and the ability for anyone on the team — not just the person who opened the session — to pick up where things left off.
Admin Controls, Permissions and Cost Management
For any organization considering handing an AI agent standing access to Slack channels, the admin control layer is the part that actually determines whether this is safe to roll out beyond a single pilot team. Claude Tag’s permission model is scoped per channel, not global: administrators decide exactly which tools and information the model can reach, and in which channels, before anyone starts tagging it. Because scoping is enforced at the identity level rather than left to individual judgment, a channel connected to a codebase doesn’t automatically expose that codebase to a channel that hasn’t been granted it.
Cost control works the same way spend controls typically work for any usage-based AI tool, applied at two levels: administrators can cap total organizational monthly spend and set separate limits for individual channels, so one high-traffic channel can’t quietly consume the whole organization’s budget. On the audit side, administrators can view a log of everything @Claude has done and who requested each task — which matters both for cost accountability and for understanding how the tool is actually being used across the organization before deciding whether to expand it further.
Real-World Use Cases by Role
Because Claude Tag lives inside existing channels rather than a separate tool, its use tends to spread organically once a team tries it in one place. A few patterns show up consistently.
Engineering teams use it for the closest thing to its original use case: reviewing pull requests, chasing down the root cause of a flaky test or a production bug, and handling smaller implementation tasks that would otherwise sit in a backlog waiting for someone to have a free afternoon. Because the whole channel sees the work, a second engineer can jump in and redirect it without needing a handoff meeting.
Support and operations teams use the same delegation pattern for ticket triage — tagging @Claude in a support channel to investigate a customer-reported issue, pull relevant logs, or draft a first-pass response for a human to review before it goes out.
Product and data teams use it to chase down metrics that would otherwise require pinging an analyst or writing a one-off query — asking a question in the channel and letting Claude pull the number from whichever connected dashboard or database has been granted access.
Leadership and cross-functional leads get the most direct benefit from the “ambient” behavior: rather than checking five channels for status, a lead can rely on Claude to flag genuinely important updates and follow up on threads that have gone quiet, cutting down on the manual context-gathering that eats up a disproportionate share of a manager’s day.
How Claude Tag Fits With the Rest of Anthropic’s Agent Lineup
Claude Tag doesn’t exist in isolation — it’s the latest piece of a broader shift in how Anthropic packages agentic Claude for different work contexts. Claude Code remains the terminal-first tool for an individual developer running long, autonomous coding sessions. Claude Cowork brings that same agentic architecture into the desktop app for non-coding knowledge work — organizing files, building spreadsheets, drafting reports — with Dispatch layered on top so a task started on a phone can hand off to a session running on a desktop. Claude Tag adds the piece none of those cover on their own: a shared, standing presence inside a team’s actual collaboration surface, where the unit of interaction is a channel full of people rather than one person’s session.
For a team actually building out its AI tooling stack, the practical question isn’t which of these to pick — it’s which job each one is suited for. An individual engineer deep in a refactor is usually better served by Claude Code directly, where they control the full session. A non-technical operations lead assembling a report from scattered files is Cowork’s use case. A cross-functional channel that needs a standing, shared agent everyone can delegate to without spinning up their own session is exactly what Claude Tag is built for. Teams that try to force one tool to cover all three tend to end up fighting the tool’s design rather than their actual problem.
This matters for anyone evaluating whether to build this capability in-house, train existing staff on it, or bring in a partner who already has opinions about which tool fits which workflow. The tooling landscape here is moving fast enough that picking wrong isn’t catastrophic — most of these are still betas or research previews — but it does mean the evaluation itself is real work, not a five-minute decision.
Claude Tag: Pros and Cons
Pros
- Shared, multiplayer visibility — anyone in the channel can see and pick up delegated work, cutting down on repeated context-setting
- Builds and retains context over time instead of starting from zero in every conversation
- Ambient behavior proactively surfaces relevant updates and follows up on stalled threads without being asked
- Tight, per-channel permission scoping makes it realistic to roll out beyond a single trusted pilot user
- Runs on Claude Opus 4.8, giving it the same agentic capability underlying Claude Code
- Launch credits for eligible Enterprise/Team organizations lower the cost of rolling it out org-wide rather than testing in one channel
Cons
- Beta access is limited to Claude Enterprise and Team customers — not yet available on individual Pro plans
- Slack-only for now, with no fixed timeline for expansion to other surfaces
- Meaningful admin setup investment required before the whole team can safely use it — this isn’t a five-minute install
- Ambient/proactive behavior needs deliberate configuration, and teams unfamiliar with agentic AI may find the initial setup decisions unintuitive
- As a research-preview-adjacent, rapidly evolving product, workflows and permission defaults may shift as Anthropic expands it
Frequently Asked Questions
Is Claude Tag different from Claude in Slack?
Yes. Claude Tag replaces the existing Claude in Slack app rather than sitting alongside it. Administrators have a 30-day window to opt into the migration, and Anthropic is issuing launch credits to eligible Enterprise and Team organizations to help offset the switch. The core difference isn’t just branding — Claude Tag adds the multiplayer, memory, ambient, and asynchronous capabilities described above, which weren’t part of the original Claude in Slack integration.
Is my data safe if I give Claude Tag access to a channel?
Access is scoped per channel by administrators, and Claude Tag does not report information from private channels even when it has broader visibility elsewhere. Anthropic’s stated model treats each configured use case as a separate Claude identity with its own memory and tool access, so a channel granted access to one system doesn’t automatically leak that access to an unrelated channel. That said, any organization should apply the same due diligence to Claude Tag’s permissions that it would to any tool with standing access to internal systems — start with a private test channel and expand deliberately, exactly as Anthropic’s own setup instructions recommend.
Does Claude Tag work outside Slack?
Not yet. Anthropic has said Slack is the starting point and that the goal is to expand @Claude into other places teams work, but no other surfaces are live at this stage. Organizations that don’t use Slack as their primary collaboration tool will need to wait for that expansion, or continue using Claude through Claude Code, Cowork, or direct chat in the meantime.
What does Claude Tag cost?
Claude Tag itself is included in beta for Claude Enterprise and Team plans rather than being a separate paid add-on, though usage still draws against the organization’s Claude spend, which administrators can cap at both the organization and individual channel level. Eligible Enterprise and Team organizations also qualify for an introductory launch credit to reduce the cost of an org-wide rollout during the beta period.
Can Claude Tag replace a developer or an outsourced team?
No, and Anthropic doesn’t position it that way — even at Anthropic itself, Claude Tag is described as changing how much of the team’s work is delegated to Claude, not as removing the need for engineers to define problems, review output, and make architectural calls. What it does change is the calculus around how a team allocates its people: work that used to require pulling an engineer off their primary project for a quick fix, a metrics pull, or a bug investigation can now be delegated and picked up by anyone in the channel, which shifts where a growing team’s next hire — or its outsourcing partner — is best spent.
Conclusion
Claude Tag is a genuine step change from single-user AI chat toward something closer to a shared, persistent team member — one with memory, initiative, and the ability to work asynchronously across a channel rather than a single conversation. It’s not a universal replacement for Slack AI’s summarization, Microsoft Copilot’s Office-embedded assistance, or a developer running Claude Code solo — each of those still has a clear role. What it does change is how much delegated, multi-step work a team can hand off without adding headcount, and how visible that work is to everyone involved rather than sitting inside one person’s private chat history.
For founders and engineering leaders already thinking hard about the build-vs-outsource-vs-augment tradeoff, tools like Claude Tag are one more input into that decision — not a reason to defer it. Evaluating whether AI-augmented team structures change what you should build in-house versus hand to a partner? Talk to Lycore — 300+ projects across 30+ industries, helping engineering teams figure out where AI tooling genuinely changes the calculus and where it doesn’t.
