Anthropic just released Claude Tag, a way to add Claude to your team's Slack as an actual member. You tag @Claude, hand off a task in plain language, and it goes and does it while you focus on something else. This guide walks you through the setup, end to end, and gives you a first task to run the day it goes live.
For most of the last two years, working with AI meant opening a separate chat window, re-explaining your context, copying the answer back out, and doing it all again the next time. It worked, but it lived outside the place your team actually gets things done. Claude Tag closes that gap. It puts Claude inside Slack as a member of the channel, so delegating to it feels less like using a tool and more like handing work to a capable teammate.
Here is the part worth paying attention to: at Anthropic, tagging @Claude is now one of the main ways they get work done. Sixty-five percent of their product team's code is created by their internal version of this, and the same pattern is spreading past engineering into chasing down metrics, working support tickets, and finding the root cause of tricky bugs. This is not a demo. It is how the company that builds Claude runs its own days.
The setup is genuinely simple, and once it is done, anyone in the channel can start tagging. Below is the full system: what you need, how to pair it, how to scope what Claude can see and do, and the first task to hand it so you feel the difference immediately.
An administrator starts in the Claude admin settings and connects Claude Tag to your Slack workspace. If your team already used the older Claude in Slack app, Claude Tag replaces it, and administrators have 30 days to opt in and migrate. Eligible Enterprise and Team organizations also get an introductory launch credit, so the whole company can try it without touching the budget on day one.
You only do this once. After the workspace is paired, the rest of the setup is about deciding what Claude can see.
This is the most important step, and it is where Claude Tag earns your trust. Administrators specify exactly which tools and information Claude can reach, and in which channels. Everything stays scoped to those channels, including what Claude remembers. A Claude set up for sales will not pass its memories or data to one set up for engineering, and the two never cross.
Think of it as creating separate Claude identities for separate jobs. Decide what each channel's Claude is for, then connect only the tools that job needs.
Before anyone starts tagging, set a cap on monthly spend. Administrators can set limits for the whole organization and for individual channels, so a busy channel cannot quietly run up a bill. You also get a full log of everything Claude has done, along with who requested each task, which means nothing happens that you cannot trace back.
Set the limit at a number that feels comfortable for a first month. You can always raise it once you see how your team uses it.
Do not roll this out to your whole team before you have watched it work once. Drop into the private test channel you set aside, tag @Claude, and give it a small, real task. Watch how it breaks the request into stages, works through them, and replies in the thread with what it created.
Here is a first task that shows you the shape of how it works without risking anything:
@Claude read the last two weeks of messages in this channel and give me a short summary of the open questions that never got a clear answer. For each one, note who asked it and suggest what the next step should be. Keep it to a simple list I can scan in under a minute.
When it replies, you will see the multiplayer part right away: the answer sits in the channel where anyone can read it and pick up where you left off. That is the whole difference. You are not having a private chat, you are working alongside a teammate everyone can see.
Once you trust it in the test channel, start delegating the work that actually eats your week. Tag @Claude the way you would message a capable coworker, in plain language, and let it work while you do something else. A few ways founders are using it that map directly to what Claude Tag was built to do:
Claude can also work asynchronously and even schedule tasks for itself, pursuing a project over hours or days. The shift, once it clicks, is that you stop doing the task and start handing it off. You spend your time deciding what needs to happen, not doing every piece of it yourself.
You paired it, scoped exactly what it can touch, capped the spend, tested it where it was safe, and handed it your first real task. That is a working AI teammate inside the place your team already lives, with controls tight enough to trust and a log that shows you everything.
The founders who win with this are not the ones who tag Claude once. They are the ones who build the habit of delegating, channel by channel, until a real share of the work runs without them. That is exactly the kind of system we build inside Her AI Systems™.
Claude Tag is a way to add Claude to your team's Slack as a member. You grant it access to selected channels and tools, then anyone in the channel can tag @Claude with a request and Claude breaks the task into stages and works through them. It is available in beta for Claude Enterprise and Team customers and runs on Opus 4.8.
Yes. Claude Tag is available in beta to Claude Enterprise and Team customers. If you are on one of those plans, an administrator can pair it with your Slack workspace from the admin settings. It replaces the older Claude in Slack app, and administrators have 30 days to opt in and migrate.
Yes. Administrators specify which tools and information Claude can access in which channels, so its memories and permissions stay scoped to those channels. A Claude set up for one team will not pass memories or data to another. Administrators can also set spend limits for the organization and for individual channels, and view a log of every task Claude has run and who requested it.
You tag @Claude with a request in plain language and it works through it using the tools it has access to, then replies in the Slack thread. Anthropic uses it internally to chase down product metrics and data, work through support tickets, and help find the root cause of tricky bugs. It can also work asynchronously and schedule tasks for itself over hours or days.
Free Download
Your future self will thank you for saving this.
Follow along for the next guide
New free guide every Tuesday. Find them first on Instagram, TikTok, or by subscribing to the email list.
Want more?
One new free guide. Every Tuesday. No spam, no fluff, just the systems and prompts I'm building for clients, simplified for you.
Subscribe