One thing quietly important occurred inside Slack on June 23, 2026. Anthropic launched Claude Tag — an AI agent embedded immediately into Salesforce’s Slack app — and the transfer is much less a couple of new chatbot function than it’s about redefining what office collaboration appears to be like like when an AI is all the time within the room.
Key takeaways
- Anthropic launched Claude Tag in Salesforce’s Slack on June 23, 2026, as a persistent AI teammate accessible through @Claude in any thread.
- The agent reads conversations, breaks down duties into phases, offers proactive updates, and retains context over time — together with studying from different channels it’s granted entry to.
- Claude Tag is at the moment in beta for Claude Enterprise and Claude Staff prospects, with growth to different platforms deliberate within the coming weeks.
- Directors can tightly management which information, instruments, and channels every Claude id can entry.
- Anthropic’s valuation has risen to $965 billion, surpassing OpenAI, and the corporate has filed confidentially for an IPO.
Anthropic Launches Claude Tag as a Persistent Slack Teammate
In contrast to earlier AI integrations that answered when spoken to, Claude Tag is designed to be all the time current. Customers summon it by typing @Claude in a Slack thread, however that’s solely half the story. The agent additionally operates in an ambient mode — proactively leaping into conversations, flagging related updates from throughout a company, and following up on duties or threads which have gone quiet, with out being explicitly requested.
That’s a significant shift. Earlier Anthropic integrations in Slack let customers DM @Claude or tag it in channels for on-demand assist. Claude Tag goes additional by introducing persistent context and reminiscence: because the agent follows together with a channel over time, it learns extra concerning the work occurring there. If granted permission, it will possibly additionally pull in info from different channels throughout the group.
Anthropic describes the expertise as working with “an actual colleague — one that may produce work in public view, with far better context and understanding than earlier than.”
Options and Capabilities
When assigned a selected job, Claude Tag breaks it down into phases and works by way of them utilizing no matter instruments it has been given entry to, posting updates again into the Slack thread. It may monitor exercise, ship alerts about posts which will have an effect on a person’s day, drop feedback into conversations, and even repair code points — all inside the acquainted Slack interface.
Crucially, everybody in a given Slack channel shares a single Claude id. Meaning any crew member can see what Claude has been engaged on and choose up the dialog from the place a colleague left off. It’s a genuinely completely different mannequin from the non-public, one-on-one AI assistant interactions most customers are accustomed to.
Cat Wu, Anthropic’s head of product for Claude Code, captured why the format change issues: “A number of the capabilities did exist, however really the shape issue of with the ability to tag it the identical manner that you’d a coworker is actually highly effective.”
Beta Entry and Who Will get It First
Claude Tag is at the moment accessible in analysis preview — in beta for Claude Enterprise and Claude Staff prospects by way of Slack. Anthropic has stated it plans to carry the potential to different platforms within the coming weeks, although no particular timeline or record of platforms has been disclosed.
Enterprise Integration and Information Controls
Enterprise adoption of AI instruments hinges closely on belief, and Anthropic has constructed a governance layer into Claude Tag from the beginning. System directors specify which instruments, info sources, and channels every Claude id can entry. A Claude set as much as help a authorized crew, for instance, can not bleed its recollections or context into the engineering channel.
This scoping mechanism is greater than a privateness function — it’s a design philosophy. Every Claude id stays contained to the channels directors outline, which suggests enterprises can deploy the agent throughout departments with out dropping management over info boundaries.
Making AI Collaboration Seen
Rob Seaman, Slack’s common supervisor, framed the launch with an idea that will get at why this issues for enterprise groups: “making AI multiplayer. As an alternative of a non-public back-and-forth, Claude Tag exhibits up within the open.”
That openness is strategically vital. When AI work occurs in a shared channel relatively than in non-public conversations, it turns into auditable, collaborative, and simpler to combine into present workflows. A crew can collectively redirect, right, or construct on what Claude has executed — which is a basically completely different relationship with AI than the solo-user assistant mannequin.
It additionally positions Anthropic in a crowded race for enterprise context. Microsoft channels organizational information by way of Copilot and Work IQ. Snowflake and Databricks are constructing back-end platforms that brokers can draw on. Glean is setting up an intelligence layer that sits between fashions and enterprise information. Claude Tag’s reply to all of that is to stay immediately contained in the dialog — studying from the messages themselves.
Anthropic’s Market Place and the Stakes Behind This Launch
This product launch doesn’t exist in isolation. Anthropic’s valuation has climbed to $965 billion, surpassing OpenAI, pushed by robust enterprise demand for its AI instruments. Earlier in June, the corporate filed confidentially for an preliminary public providing — a big milestone that raises the stakes for each product choice the corporate makes.
In that context, Claude Tag isn’t just a function replace. It’s an announcement about the place Anthropic sees the frontier of enterprise AI adoption: not in standalone productiveness instruments, however embedded contained in the platforms the place work already occurs. Getting Claude into Slack threads — the place tens of millions {of professional} conversations happen each day — is a distribution play as a lot as a product one.
The aggressive strain this creates is actual. OpenAI has its personal enterprise integrations, and Microsoft’s deep integration with Groups offers Copilot a structural benefit in sure enterprise environments. Anthropic’s guess is that the standard of Claude’s contextual understanding, mixed with the openness and adaptability of the Slack platform, can carve out a sturdy place out there earlier than the IPO window opens.
Whether or not the ambient, always-learning AI teammate mannequin resonates with enterprise patrons — or raises considerations about what it means to have an AI studying each message in a channel — could be the defining query for Claude Tag’s subsequent section.
FAQ
What’s Claude Tag and the way does it work in Slack?
Claude Tag is an AI agent launched by Anthropic inside Salesforce’s Slack app. Customers can summon it by typing @Claude in any thread, the place it reads conversations, breaks down duties into phases, and offers proactive updates. It additionally operates in an ambient mode, flagging related info and following up on forgotten threads with out being requested. Importantly, it retains context over time and may study from different channels it’s granted permission to entry.
Who can entry Claude Tag at the moment?
Claude Tag is at the moment accessible in beta — described as a analysis preview — for Claude Enterprise and Claude Staff prospects by way of Slack.
Can directors management Claude Tag’s entry to Slack information?
Sure. Anthropic stated directors can tightly management the info, instruments, and channels that every Claude id can entry. Every Claude id stays scoped to the channels directors outline, stopping info from one division bleeding into one other.
What are Anthropic’s plans for Claude Tag past Slack?
Anthropic has stated it plans to develop Claude Tag to different platforms within the coming weeks, although no particular platforms or dates have been introduced.
Article produced with the help of synthetic intelligence and reviewed by the editorial crew.