What is Claude Tag?
Anthropic just put an AI teammate inside your Slack channels — one shared @Claude that works on its own for hours, schedules its own follow-ups, and remembers the channel. Here is what it actually is, and what changes.
A shared teammate, not a private chatbot
Most AI at work is single-player: you open a private chat, ask, and the answer is yours alone. Claude Tag inverts that. Within a Slack channel there is one Claude that interacts with everyone — tag @Claude, hand it a task, and the whole channel can watch it work and jump in. Anthropic frames it as working with a colleague that produces work in public view rather than a tool you ping in a direct message.
The product lead for Claude Code and Cowork, Cat Wu, put the distinction plainly: Claude Code, Cowork, and chat are very single-player, whereas Claude Tag is built to be interactive and multiplayer. That one design choice is what most of the genuinely new behavior flows from.
Autonomous, scheduled, and aware of the channel
You set Claude a task and move on; it works in the background and posts results back to the thread. It is built for long-horizon work, not just instant replies.
- Autonomous over hours or days — hand it a project and it pursues it without you babysitting each step.
- Schedules its own tasks — it can set up future runs and recurring checks, then act on them.
- Channel and workspace memory — it remembers relevant context from the channels it is in, so it does not start from zero each time.
- Ambient mode (optional) — when enabled, it acts without being tagged: posting when a job finishes, nudging a stalled thread, or surfacing something it thinks you need to know.
- Connected tools, data, and codebases — admins decide what it can reach; code work shows up as the Claude GitHub App authoring a pull request, linked back to the Slack thread that started it.
Admins decide what it can see and spend
Because it lives where the work and the data already are, control is the heart of the product. Only a Slack Primary Owner or Owner can set up Claude Tag. Access is scoped in three tiers — organization-wide, per-workspace, and per-private-channel — and memory stays scoped to the channels it was defined in. A sales-configured Claude cannot read engineering data or memories, and it does not report from private channels.
Spending is bounded too. Admins set token-spend limits for the organization and for individual channels, with alerts at 75% and 95%; work that would exceed a limit is declined rather than silently cut short. An audit view lists every scheduled and one-time task plus the calls made under the agent identity, so there is a trail of what Claude did and who asked.
Beta today, and the August 3 cutover
Claude Tag is available in beta for Claude Enterprise and Team customers and runs on Claude Opus 4.8. It replaces the existing Claude in Slack app: administrators have a 30-day window to opt in, and the old app is switched over to the new experience on August 3, 2026. Anthropic is offering an introductory launch credit to eligible organizations, and channel work is billed to the organization while direct messages are billed to the individual.
Slack is the starting point, not the destination. Anthropic says it wants to expand so teams can tag @Claude in the other places they work, though it has not named which platforms come next.
A capability that still needs a harness
Anthropic reports that 65% of its own product team’s code is now created by its internal version of Claude Tag, and that the pattern has spread well beyond engineering into metrics, support tickets, and bug triage. That is a real signal of what a shared, autonomous agent can do when it is wired into the right tools and trusted with the right scope.
It is also the catch. The capability is only as good as the harness around it: which channels it lives in, which tools and data it can touch, where the human approval gates sit, and whether its memory and outputs respect where your data is allowed to live. Standing it up safely is a configuration and governance project, not a toggle. That wrapper is exactly the work AIMOCS does — putting models like this to work inside a business with bounded authority, monitoring, and the data rules the business actually operates under.
What is Claude Tag?
Claude Tag is Anthropic’s shared AI teammate for Slack, launched June 23, 2026. You tag @Claude in a channel and it takes on real work using connected tools and the channel’s context. It is multiplayer (one Claude per channel that everyone can steer), works autonomously over hours or days, can schedule its own tasks, and runs on Claude Opus 4.8.
How is Claude Tag different from the old Claude in Slack app?
Claude Tag replaces the Claude in Slack app. The old app was largely a private assistant; Claude Tag is a shared, autonomous teammate with channel memory, self-scheduled tasks, an ambient mode, and admin-controlled access to tools and data. Administrators have 30 days to opt in, and the old app is switched over on August 3, 2026.
What is ambient mode?
Ambient mode is an optional setting that lets Claude act without being tagged. When it is on, Claude proactively keeps the channel updated — posting when a job finishes, following up on a stalled thread, or surfacing information from connected tools it thinks the team needs.
Is Claude Tag multiplayer?
Yes. Within a given channel there is one Claude that everyone interacts with, so colleagues can see what it is doing and pick up a task where someone else left off. This shared identity is the main thing that sets it apart from single-player AI chats.
Who can use Claude Tag and what does it run on?
It is in beta for Claude Enterprise and Team customers, set up by a Slack Primary Owner or Owner, and it runs on Claude Opus 4.8. Channel work is billed to the organization and direct messages to the individual, with admin-set token-spend limits and alerts at 75% and 95%.
We don't advise on AI. We run it for you.
Proven on your data before you commit.