Anthropic launched Claude Tag today, a Slack-native version of its assistant that behaves like a shared virtual employee any team member can summon, watch, and redirect mid-task. The product is rolling out as a research preview for Claude Enterprise and Claude Team customers, and it already runs heavily inside Anthropic itself — approving and incorporating 65% of code changes submitted by the product team, according to Anthropic's Cat Wu. The bot inherits the agentic backbone of Claude Code and Cowork but shifts the surface from a solo IDE to a multiplayer channel.
The pitch is enterprise collaboration. A user tags Claude in Slack, the agent decomposes the request into steps, executes them in the background, and posts the result back to the channel where any teammate can intervene. Anthropic says the bot maintains a single organization-wide identity, so a half-finished workflow can be picked up by a colleague without re-explaining context. Over time it learns the company across channels, which is meant to remove the per-task onboarding overhead that has slowed AI adoption at large firms.
That positions Claude Tag against Salesforce's Slackbot and AI startups such as Viktor, both of which already sell Slack-resident agents. It also follows our earlier coverage of Anthropic embedding Claude directly into Slack as an always-on teammate — Claude Tag is the productized, multiplayer evolution of that work.
Key facts
- 01Anthropic launched Claude Tag as a research preview inside Slack for Claude Enterprise and Claude Team customers.
- 02Claude Tag is already approving and incorporating 65% of code changes submitted by Anthropic's own product team.
- 03Per Ramp's May AI Index, 34.4% of firms pay for Anthropic services vs. 32.3% for OpenAI — the first time Anthropic has led.
- 04Ramp's data set covers more than 50,000 U.S. companies, with Claude Code cited as the main driver of Anthropic's lead.
- 05Claude Tag will compete directly with Salesforce's Slackbot and AI startups like Viktor.
Wu frames the design shift bluntly: Claude Code, Cowork, and chat are built for one user at a time, while Claude Tag is built for a channel. The bot's output is visible to everyone in the room, and anyone can grab the wheel. That matters because the most common enterprise complaint about agents has been opacity — work happens in private threads, nothing is auditable, and handoffs break.
The release lands at a moment when Anthropic has, for the first time, edged past OpenAI in paid business adoption. Ramp's May AI Index, drawn from corporate spending data across more than 50,000 U.S. companies, shows 34.4% of firms paying for Anthropic services against 32.3% paying for OpenAI. Ramp credits Claude Code as the primary driver of that shift, which makes Claude Tag's productization of similar agentic muscle inside Slack a logical next move — and a direct attempt to widen the gap before OpenAI and Google counter.
Security and cost — the two reasons enterprise AI pilots stall — get explicit guardrails. Administrators can scope which tools, channels, memories, and data each instance can touch, so an HR-side Claude cannot leak personnel data into an engineering channel. Token-spend caps can be set per channel and per organization, addressing the budget overruns that have soured CFOs on usage-based agent pricing.
“If you're handling very sensitive data, such as personnel data, you can also DM Claude Tag … So you can be confident that sensitive information doesn't leak.”— Cat Wu, Anthropic head of product for Claude Code and Cowork
An "ambient" mode lets Claude Tag proactively surface updates, follow up on stalled threads, and remind users of forgotten tasks across the workspace. That feature collides directly with Slackbot's existing nudge surface and with the workflow-automation layer Salesforce has been building since acquiring Slack. The difference is that Claude Tag's nudges are generated by a frontier reasoning model with access to organization-wide context, not by static rules.
The 65% figure for code-change acceptance inside Anthropic is the most concrete productivity claim attached to the launch, though it carries the usual caveat: Anthropic's own engineers are the most sophisticated possible users of a Claude agent, and acceptance rates outside the lab are likely to be lower at first. Enterprises that have struggled to consolidate data across siloed teams will still have to do that consolidation work before Claude Tag's cross-channel learning pays off, and the research-preview gating means broader availability and pricing details are still pending.
Anthropic is heading into a likely IPO this year, and the enterprise channel is the revenue base public-market investors will scrutinize. Claude Tag is the clearest signal yet that the company intends to defend its lead over OpenAI in business adoption by going where work already happens rather than asking customers to come to a separate Claude surface. If Slack proves to be the right beachhead, Microsoft Teams is the obvious next port — and that is the front where the agent war actually gets decided.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
