Anthropic just dropped something that's going to change how a lot of teams think about Slack. It's called Claude Tag, and the easiest way to describe it is this: it's an AI agent that doesn't just answer your questions when you ping it — it actually sticks around. It learns your channel. It remembers what happened yesterday. And it works alongside your team like, well, a teammate.
This is Anthropic's biggest swing yet at workplace collaboration software, and honestly, the timing makes sense. Companies have been bolting AI onto Slack for a while now. But most of those integrations work the same way: you tag the bot, it answers, the context disappears. Claude Tag is built to break that cycle.
What Makes Claude Tag Different From a Regular Slack Bot
Here's the thing about Anthropic's older Slack integrations — you could DM Claude or tag it for a quick answer, and that was basically it. One-off interactions. No real memory of what came before.
Claude Tag flips that. It introduces persistent memory and context that actually carries across conversations and across team members. And this part's interesting: everyone in a given Slack channel shares the same Claude identity. So if you tag Claude on Monday and your coworker picks up that thread on Wednesday, Claude already knows what's going on. Nobody has to catch it up.
Anthropic put it simply — anyone in the channel can see what Claude's been working on and just pick up right where the last person left off.
How Claude Tag Handles Tasks
When you hand Claude Tag a task, it doesn't just fire back a response. It breaks the task into stages, works through them using whatever tools it's been given access to, and posts the results right back into the Slack thread. So you're watching it actually move through the work, step by step, instead of getting a single canned reply.
Ambient Mode: The Part That Works Without Being Asked
This might be the most ambitious piece of the whole release. Claude Tag includes what Anthropic calls "ambient mode," and it basically means the AI doesn't wait to be summoned. It proactively surfaces updates. It flags relevant information from elsewhere in the organization. It follows up on threads that got dropped and forgotten — without anyone having to remember to ask it to.
Anthropic explained that as Claude follows along in its channel, it keeps learning more about the work happening there. And if it's been granted permission, it can automatically pull facts from other channels across the organization too.
Enterprise Controls: Who Gets to See What
Obviously, a tool that learns everything happening in a channel raises a fair question: what stops it from leaking context where it shouldn't?
Anthropic built in guardrails for exactly that. System administrators get to define which tools, data sources, and channels each individual Claude identity can actually access. That means a Claude instance set up for the legal team can't accidentally seed memories or context into the engineering channel. The departments stay separated, even though it's technically the same underlying product.
Right now, Claude Tag is available in research preview, and it's limited to Claude Enterprise and Claude Team customers. Anthropic has said it plans to expand the feature to additional platforms beyond Slack down the line.
Where Claude Tag Fits in the Bigger AI Workplace Race
This launch doesn't happen in a vacuum. Anthropic is now going head-to-head with some serious competition in the workplace AI space:
- Microsoft Copilot, which pulls organizational context from Microsoft Graph
- Glean, an enterprise search company building its own intelligence layers
- Snowflake, a data platform also working on intelligence infrastructure for AI agents
According to reporting from Reuters, this move reinforces just how central the enterprise sector has become for AI companies right now — it's turned into one of the most important battlegrounds in the whole industry.